asterisk/main/rtp_engine.c

4031 lines
122 KiB
C
Raw Normal View History

/*
* Asterisk -- An open source telephony toolkit.
*
* Copyright (C) 1999 - 2008, Digium, Inc.
*
* Joshua Colp <jcolp@digium.com>
*
* See http://www.asterisk.org for more information about
* the Asterisk project. Please do not directly contact
* any of the maintainers of this project for assistance;
* the project provides a web site, mailing lists and IRC
* channels for your use.
*
* This program is free software, distributed under the terms of
* the GNU General Public License Version 2. See the LICENSE file
* at the top of the source tree.
*/
/*! \file
*
* \brief Pluggable RTP Architecture
*
* \author Joshua Colp <jcolp@digium.com>
*/
/*** MODULEINFO
<support_level>core</support_level>
Refactor RTCP events over to Stasis; associate with channels This patch does the following: * It merges Jaco Kroon's patch from ASTERISK-20754, which provides channel information in the RTCP events. Because Stasis provides a cache, Jaco's patch was modified to pass the channel uniqueid to the RTP layer as opposed to a pointer to the channel. This has the following benefits: (1) It keeps the RTP engine 'clean' of references back to channels (2) It prevents circular dependencies and other potential ref counting issues * The RTP engine now allows any RTP implementation to raise RTCP messages. Potentially, other implementations (such as res_rtp_multicast) could also raise RTCP information. The engine provides structs to represent RTCP headers and RTCP SR/RR reports. * Some general refactoring in res_rtp_asterisk was done to try and tame the RTCP code. It isn't perfect - that's *way* beyond the scope of this work - but it does feel marginally better. * A few random bugs were fixed in the RTCP statistics. (Example: performing an assignment of a = a is probably not correct) * We now raise RTCP events for each SR/RR sent/received. Previously we wouldn't raise an event when we sent a RR report. Note that this work will be of use to others who want to monitor call quality or build modules that report call quality statistics. Since the events are now moving across the Stasis message bus, this is far easier to accomplish. It is also a first step (though by no means the last step) towards getting Olle's pinefrog work incorporated. Again: note that the patch by Jaco Kroon was modified slightly for this work; however, he did all of the hard work in finding the right places to set the channel in the RTP engine across the channel drivers. Much thanks goes to Jaco for his hard work here. Review: https://reviewboard.asterisk.org/r/2603/ (closes issue ASTERISK-20574) Reported by: Jaco Kroon patches: asterisk-rtcp-channel.patch uploaded by jkroon (License 5671) (closes issue ASTERISK-21471) Reported by: Matt Jordan git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@393740 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2013-07-05 17:33:33 +00:00
***/
/*** DOCUMENTATION
<managerEvent language="en_US" name="RTCPSent">
<managerEventInstance class="EVENT_FLAG_REPORTING">
<synopsis>Raised when an RTCP packet is sent.</synopsis>
<syntax>
<channel_snapshot/>
Refactor RTCP events over to Stasis; associate with channels This patch does the following: * It merges Jaco Kroon's patch from ASTERISK-20754, which provides channel information in the RTCP events. Because Stasis provides a cache, Jaco's patch was modified to pass the channel uniqueid to the RTP layer as opposed to a pointer to the channel. This has the following benefits: (1) It keeps the RTP engine 'clean' of references back to channels (2) It prevents circular dependencies and other potential ref counting issues * The RTP engine now allows any RTP implementation to raise RTCP messages. Potentially, other implementations (such as res_rtp_multicast) could also raise RTCP information. The engine provides structs to represent RTCP headers and RTCP SR/RR reports. * Some general refactoring in res_rtp_asterisk was done to try and tame the RTCP code. It isn't perfect - that's *way* beyond the scope of this work - but it does feel marginally better. * A few random bugs were fixed in the RTCP statistics. (Example: performing an assignment of a = a is probably not correct) * We now raise RTCP events for each SR/RR sent/received. Previously we wouldn't raise an event when we sent a RR report. Note that this work will be of use to others who want to monitor call quality or build modules that report call quality statistics. Since the events are now moving across the Stasis message bus, this is far easier to accomplish. It is also a first step (though by no means the last step) towards getting Olle's pinefrog work incorporated. Again: note that the patch by Jaco Kroon was modified slightly for this work; however, he did all of the hard work in finding the right places to set the channel in the RTP engine across the channel drivers. Much thanks goes to Jaco for his hard work here. Review: https://reviewboard.asterisk.org/r/2603/ (closes issue ASTERISK-20574) Reported by: Jaco Kroon patches: asterisk-rtcp-channel.patch uploaded by jkroon (License 5671) (closes issue ASTERISK-21471) Reported by: Matt Jordan git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@393740 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2013-07-05 17:33:33 +00:00
<parameter name="SSRC">
<para>The SSRC identifier for our stream</para>
</parameter>
<parameter name="PT">
<para>The type of packet for this RTCP report.</para>
<enumlist>
<enum name="200(SR)"/>
<enum name="201(RR)"/>
</enumlist>
</parameter>
<parameter name="To">
<para>The address the report is sent to.</para>
</parameter>
<parameter name="ReportCount">
<para>The number of reports that were sent.</para>
<para>The report count determines the number of ReportX headers in
the message. The X for each set of report headers will range from 0 to
<literal>ReportCount - 1</literal>.</para>
</parameter>
<parameter name="SentNTP" required="false">
<para>The time the sender generated the report. Only valid when
PT is <literal>200(SR)</literal>.</para>
</parameter>
<parameter name="SentRTP" required="false">
<para>The sender's last RTP timestamp. Only valid when PT is
<literal>200(SR)</literal>.</para>
</parameter>
<parameter name="SentPackets" required="false">
<para>The number of packets the sender has sent. Only valid when PT
is <literal>200(SR)</literal>.</para>
</parameter>
<parameter name="SentOctets" required="false">
<para>The number of bytes the sender has sent. Only valid when PT is
<literal>200(SR)</literal>.</para>
</parameter>
<parameter name="ReportXSourceSSRC">
<para>The SSRC for the source of this report block.</para>
</parameter>
<parameter name="ReportXFractionLost">
<para>The fraction of RTP data packets from <literal>ReportXSourceSSRC</literal>
lost since the previous SR or RR report was sent.</para>
</parameter>
<parameter name="ReportXCumulativeLost">
<para>The total number of RTP data packets from <literal>ReportXSourceSSRC</literal>
lost since the beginning of reception.</para>
</parameter>
<parameter name="ReportXHighestSequence">
<para>The highest sequence number received in an RTP data packet from
<literal>ReportXSourceSSRC</literal>.</para>
</parameter>
<parameter name="ReportXSequenceNumberCycles">
<para>The number of sequence number cycles seen for the RTP data
received from <literal>ReportXSourceSSRC</literal>.</para>
</parameter>
<parameter name="ReportXIAJitter">
<para>An estimate of the statistical variance of the RTP data packet
interarrival time, measured in timestamp units.</para>
</parameter>
<parameter name="ReportXLSR">
<para>The last SR timestamp received from <literal>ReportXSourceSSRC</literal>.
If no SR has been received from <literal>ReportXSourceSSRC</literal>,
then 0.</para>
</parameter>
<parameter name="ReportXDLSR">
<para>The delay, expressed in units of 1/65536 seconds, between
receiving the last SR packet from <literal>ReportXSourceSSRC</literal>
and sending this report.</para>
</parameter>
</syntax>
<see-also>
<ref type="managerEvent">RTCPReceived</ref>
</see-also>
Refactor RTCP events over to Stasis; associate with channels This patch does the following: * It merges Jaco Kroon's patch from ASTERISK-20754, which provides channel information in the RTCP events. Because Stasis provides a cache, Jaco's patch was modified to pass the channel uniqueid to the RTP layer as opposed to a pointer to the channel. This has the following benefits: (1) It keeps the RTP engine 'clean' of references back to channels (2) It prevents circular dependencies and other potential ref counting issues * The RTP engine now allows any RTP implementation to raise RTCP messages. Potentially, other implementations (such as res_rtp_multicast) could also raise RTCP information. The engine provides structs to represent RTCP headers and RTCP SR/RR reports. * Some general refactoring in res_rtp_asterisk was done to try and tame the RTCP code. It isn't perfect - that's *way* beyond the scope of this work - but it does feel marginally better. * A few random bugs were fixed in the RTCP statistics. (Example: performing an assignment of a = a is probably not correct) * We now raise RTCP events for each SR/RR sent/received. Previously we wouldn't raise an event when we sent a RR report. Note that this work will be of use to others who want to monitor call quality or build modules that report call quality statistics. Since the events are now moving across the Stasis message bus, this is far easier to accomplish. It is also a first step (though by no means the last step) towards getting Olle's pinefrog work incorporated. Again: note that the patch by Jaco Kroon was modified slightly for this work; however, he did all of the hard work in finding the right places to set the channel in the RTP engine across the channel drivers. Much thanks goes to Jaco for his hard work here. Review: https://reviewboard.asterisk.org/r/2603/ (closes issue ASTERISK-20574) Reported by: Jaco Kroon patches: asterisk-rtcp-channel.patch uploaded by jkroon (License 5671) (closes issue ASTERISK-21471) Reported by: Matt Jordan git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@393740 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2013-07-05 17:33:33 +00:00
</managerEventInstance>
</managerEvent>
<managerEvent language="en_US" name="RTCPReceived">
<managerEventInstance class="EVENT_FLAG_REPORTING">
<synopsis>Raised when an RTCP packet is received.</synopsis>
<syntax>
<channel_snapshot/>
Refactor RTCP events over to Stasis; associate with channels This patch does the following: * It merges Jaco Kroon's patch from ASTERISK-20754, which provides channel information in the RTCP events. Because Stasis provides a cache, Jaco's patch was modified to pass the channel uniqueid to the RTP layer as opposed to a pointer to the channel. This has the following benefits: (1) It keeps the RTP engine 'clean' of references back to channels (2) It prevents circular dependencies and other potential ref counting issues * The RTP engine now allows any RTP implementation to raise RTCP messages. Potentially, other implementations (such as res_rtp_multicast) could also raise RTCP information. The engine provides structs to represent RTCP headers and RTCP SR/RR reports. * Some general refactoring in res_rtp_asterisk was done to try and tame the RTCP code. It isn't perfect - that's *way* beyond the scope of this work - but it does feel marginally better. * A few random bugs were fixed in the RTCP statistics. (Example: performing an assignment of a = a is probably not correct) * We now raise RTCP events for each SR/RR sent/received. Previously we wouldn't raise an event when we sent a RR report. Note that this work will be of use to others who want to monitor call quality or build modules that report call quality statistics. Since the events are now moving across the Stasis message bus, this is far easier to accomplish. It is also a first step (though by no means the last step) towards getting Olle's pinefrog work incorporated. Again: note that the patch by Jaco Kroon was modified slightly for this work; however, he did all of the hard work in finding the right places to set the channel in the RTP engine across the channel drivers. Much thanks goes to Jaco for his hard work here. Review: https://reviewboard.asterisk.org/r/2603/ (closes issue ASTERISK-20574) Reported by: Jaco Kroon patches: asterisk-rtcp-channel.patch uploaded by jkroon (License 5671) (closes issue ASTERISK-21471) Reported by: Matt Jordan git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@393740 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2013-07-05 17:33:33 +00:00
<parameter name="SSRC">
<para>The SSRC identifier for the remote system</para>
</parameter>
<xi:include xpointer="xpointer(/docs/managerEvent[@name='RTCPSent']/managerEventInstance/syntax/parameter[@name='PT'])" />
<parameter name="From">
<para>The address the report was received from.</para>
</parameter>
<parameter name="RTT">
<para>Calculated Round-Trip Time in seconds</para>
</parameter>
<parameter name="ReportCount">
<para>The number of reports that were received.</para>
<para>The report count determines the number of ReportX headers in
the message. The X for each set of report headers will range from 0 to
<literal>ReportCount - 1</literal>.</para>
</parameter>
<xi:include xpointer="xpointer(/docs/managerEvent[@name='RTCPSent']/managerEventInstance/syntax/parameter[@name='SentNTP'])" />
<xi:include xpointer="xpointer(/docs/managerEvent[@name='RTCPSent']/managerEventInstance/syntax/parameter[@name='SentRTP'])" />
<xi:include xpointer="xpointer(/docs/managerEvent[@name='RTCPSent']/managerEventInstance/syntax/parameter[@name='SentPackets'])" />
<xi:include xpointer="xpointer(/docs/managerEvent[@name='RTCPSent']/managerEventInstance/syntax/parameter[@name='SentOctets'])" />
<xi:include xpointer="xpointer(/docs/managerEvent[@name='RTCPSent']/managerEventInstance/syntax/parameter[contains(@name, 'ReportX')])" />
</syntax>
<see-also>
<ref type="managerEvent">RTCPSent</ref>
</see-also>
Refactor RTCP events over to Stasis; associate with channels This patch does the following: * It merges Jaco Kroon's patch from ASTERISK-20754, which provides channel information in the RTCP events. Because Stasis provides a cache, Jaco's patch was modified to pass the channel uniqueid to the RTP layer as opposed to a pointer to the channel. This has the following benefits: (1) It keeps the RTP engine 'clean' of references back to channels (2) It prevents circular dependencies and other potential ref counting issues * The RTP engine now allows any RTP implementation to raise RTCP messages. Potentially, other implementations (such as res_rtp_multicast) could also raise RTCP information. The engine provides structs to represent RTCP headers and RTCP SR/RR reports. * Some general refactoring in res_rtp_asterisk was done to try and tame the RTCP code. It isn't perfect - that's *way* beyond the scope of this work - but it does feel marginally better. * A few random bugs were fixed in the RTCP statistics. (Example: performing an assignment of a = a is probably not correct) * We now raise RTCP events for each SR/RR sent/received. Previously we wouldn't raise an event when we sent a RR report. Note that this work will be of use to others who want to monitor call quality or build modules that report call quality statistics. Since the events are now moving across the Stasis message bus, this is far easier to accomplish. It is also a first step (though by no means the last step) towards getting Olle's pinefrog work incorporated. Again: note that the patch by Jaco Kroon was modified slightly for this work; however, he did all of the hard work in finding the right places to set the channel in the RTP engine across the channel drivers. Much thanks goes to Jaco for his hard work here. Review: https://reviewboard.asterisk.org/r/2603/ (closes issue ASTERISK-20574) Reported by: Jaco Kroon patches: asterisk-rtcp-channel.patch uploaded by jkroon (License 5671) (closes issue ASTERISK-21471) Reported by: Matt Jordan git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@393740 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2013-07-05 17:33:33 +00:00
</managerEventInstance>
</managerEvent>
***/
#include "asterisk.h"
#include <math.h> /* for sqrt, MAX */
#include <sched.h> /* for sched_yield */
#include <sys/time.h> /* for timeval */
#include <time.h> /* for time_t */
#include "asterisk/_private.h" /* for ast_rtp_engine_init prototype */
#include "asterisk/astobj2.h" /* for ao2_cleanup, ao2_ref, etc */
#include "asterisk/channel.h" /* for ast_channel_name, etc */
#include "asterisk/codec.h" /* for ast_codec_media_type2str, etc */
#include "asterisk/format.h" /* for ast_format_cmp, etc */
#include "asterisk/format_cache.h" /* for ast_format_adpcm, etc */
#include "asterisk/format_cap.h" /* for ast_format_cap_alloc, etc */
#include "asterisk/json.h" /* for ast_json_ref, etc */
#include "asterisk/linkedlists.h" /* for ast_rtp_engine::<anonymous>, etc */
#include "asterisk/lock.h" /* for ast_rwlock_unlock, etc */
#include "asterisk/logger.h" /* for ast_log, ast_debug, etc */
#include "asterisk/manager.h"
#include "asterisk/module.h" /* for ast_module_unref, etc */
#include "asterisk/netsock2.h" /* for ast_sockaddr_copy, etc */
#include "asterisk/options.h" /* for ast_option_rtpptdynamic */
#include "asterisk/pbx.h" /* for pbx_builtin_setvar_helper */
#include "asterisk/res_srtp.h" /* for ast_srtp_res */
#include "asterisk/rtp_engine.h" /* for ast_rtp_codecs, etc */
#include "asterisk/stasis.h" /* for stasis_message_data, etc */
#include "asterisk/stasis_channels.h" /* for ast_channel_stage_snapshot, etc */
#include "asterisk/strings.h" /* for ast_str_append, etc */
#include "asterisk/time.h" /* for ast_tvdiff_ms, ast_tvnow */
#include "asterisk/translate.h" /* for ast_translate_available_formats */
#include "asterisk/utils.h" /* for ast_free, ast_strdup, etc */
#include "asterisk/vector.h" /* for AST_VECTOR_GET, etc */
struct ast_srtp_res *res_srtp = NULL;
struct ast_srtp_policy_res *res_srtp_policy = NULL;
/*! Structure that contains extmap negotiation information */
struct rtp_extmap {
/*! The RTP extension */
enum ast_rtp_extension extension;
/*! The current negotiated direction */
enum ast_rtp_extension_direction direction;
};
/*! Structure that represents an RTP session (instance) */
struct ast_rtp_instance {
/*! Engine that is handling this RTP instance */
struct ast_rtp_engine *engine;
/*! Data unique to the RTP engine */
void *data;
/*! RTP properties that have been set and their value */
int properties[AST_RTP_PROPERTY_MAX];
/*! Address that we are expecting RTP to come in to */
struct ast_sockaddr local_address;
/*! The original source address */
struct ast_sockaddr requested_target_address;
/*! Address that we are sending RTP to */
struct ast_sockaddr incoming_source_address;
/*! Instance that we are bridged to if doing remote or local bridging */
struct ast_rtp_instance *bridged;
/*! Payload and packetization information */
struct ast_rtp_codecs codecs;
/*! RTP timeout time (negative or zero means disabled, negative value means temporarily disabled) */
int timeout;
/*! RTP timeout when on hold (negative or zero means disabled, negative value means temporarily disabled). */
int holdtimeout;
/*! RTP keepalive interval */
int keepalive;
/*! Glue currently in use */
struct ast_rtp_glue *glue;
/*! SRTP info associated with the instance */
struct ast_srtp *srtp;
/*! SRTP info dedicated for RTCP associated with the instance */
struct ast_srtp *rtcp_srtp;
Refactor RTCP events over to Stasis; associate with channels This patch does the following: * It merges Jaco Kroon's patch from ASTERISK-20754, which provides channel information in the RTCP events. Because Stasis provides a cache, Jaco's patch was modified to pass the channel uniqueid to the RTP layer as opposed to a pointer to the channel. This has the following benefits: (1) It keeps the RTP engine 'clean' of references back to channels (2) It prevents circular dependencies and other potential ref counting issues * The RTP engine now allows any RTP implementation to raise RTCP messages. Potentially, other implementations (such as res_rtp_multicast) could also raise RTCP information. The engine provides structs to represent RTCP headers and RTCP SR/RR reports. * Some general refactoring in res_rtp_asterisk was done to try and tame the RTCP code. It isn't perfect - that's *way* beyond the scope of this work - but it does feel marginally better. * A few random bugs were fixed in the RTCP statistics. (Example: performing an assignment of a = a is probably not correct) * We now raise RTCP events for each SR/RR sent/received. Previously we wouldn't raise an event when we sent a RR report. Note that this work will be of use to others who want to monitor call quality or build modules that report call quality statistics. Since the events are now moving across the Stasis message bus, this is far easier to accomplish. It is also a first step (though by no means the last step) towards getting Olle's pinefrog work incorporated. Again: note that the patch by Jaco Kroon was modified slightly for this work; however, he did all of the hard work in finding the right places to set the channel in the RTP engine across the channel drivers. Much thanks goes to Jaco for his hard work here. Review: https://reviewboard.asterisk.org/r/2603/ (closes issue ASTERISK-20574) Reported by: Jaco Kroon patches: asterisk-rtcp-channel.patch uploaded by jkroon (License 5671) (closes issue ASTERISK-21471) Reported by: Matt Jordan git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@393740 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2013-07-05 17:33:33 +00:00
/*! Channel unique ID */
char channel_uniqueid[AST_MAX_UNIQUEID];
/*! Time of last packet sent */
time_t last_tx;
/*! Time of last packet received */
time_t last_rx;
/*! Enabled RTP extensions */
AST_VECTOR(, enum ast_rtp_extension_direction) extmap_enabled;
/*! Negotiated RTP extensions (using index based on extension) */
AST_VECTOR(, int) extmap_negotiated;
/*! Negotiated RTP extensions (using index based on unique id) */
AST_VECTOR(, struct rtp_extmap) extmap_unique_ids;
};
/*!
* \brief URIs for known RTP extensions
*/
static const char * const rtp_extension_uris[AST_RTP_EXTENSION_MAX] = {
[AST_RTP_EXTENSION_UNSUPPORTED] = "",
[AST_RTP_EXTENSION_ABS_SEND_TIME] = "http://www.webrtc.org/experiments/rtp-hdrext/abs-send-time",
[AST_RTP_EXTENSION_TRANSPORT_WIDE_CC] = "http://www.ietf.org/id/draft-holmer-rmcat-transport-wide-cc-extensions-01",
};
/*! List of RTP engines that are currently registered */
static AST_RWLIST_HEAD_STATIC(engines, ast_rtp_engine);
/*! List of RTP glues */
static AST_RWLIST_HEAD_STATIC(glues, ast_rtp_glue);
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
#define MAX_RTP_MIME_TYPES 128
/*! The following array defines the MIME Media type (and subtype) for each
of our codecs, or RTP-specific data type. */
static struct ast_rtp_mime_type {
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
/*! \brief A mapping object between the Asterisk codec and this RTP payload */
struct ast_rtp_payload_type payload_type;
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
/*! \brief The media type */
char type[16];
/*! \brief The format type */
char subtype[64];
/*! \brief Expected sample rate of the /c subtype */
unsigned int sample_rate;
} ast_rtp_mime_types[128]; /* This will Likely not need to grow any time soon. */
static ast_rwlock_t mime_types_lock;
static int mime_types_len = 0;
/*!
* \brief Mapping between Asterisk codecs and rtp payload types
*
* Static (i.e., well-known) RTP payload types for our "AST_FORMAT..."s:
* also, our own choices for dynamic payload types. This is our master
* table for transmission
*
* See http://www.iana.org/assignments/rtp-parameters for a list of
* assigned values
*/
static struct ast_rtp_payload_type *static_RTP_PT[AST_RTP_MAX_PT];
static ast_rwlock_t static_RTP_PT_lock;
Refactor RTCP events over to Stasis; associate with channels This patch does the following: * It merges Jaco Kroon's patch from ASTERISK-20754, which provides channel information in the RTCP events. Because Stasis provides a cache, Jaco's patch was modified to pass the channel uniqueid to the RTP layer as opposed to a pointer to the channel. This has the following benefits: (1) It keeps the RTP engine 'clean' of references back to channels (2) It prevents circular dependencies and other potential ref counting issues * The RTP engine now allows any RTP implementation to raise RTCP messages. Potentially, other implementations (such as res_rtp_multicast) could also raise RTCP information. The engine provides structs to represent RTCP headers and RTCP SR/RR reports. * Some general refactoring in res_rtp_asterisk was done to try and tame the RTCP code. It isn't perfect - that's *way* beyond the scope of this work - but it does feel marginally better. * A few random bugs were fixed in the RTCP statistics. (Example: performing an assignment of a = a is probably not correct) * We now raise RTCP events for each SR/RR sent/received. Previously we wouldn't raise an event when we sent a RR report. Note that this work will be of use to others who want to monitor call quality or build modules that report call quality statistics. Since the events are now moving across the Stasis message bus, this is far easier to accomplish. It is also a first step (though by no means the last step) towards getting Olle's pinefrog work incorporated. Again: note that the patch by Jaco Kroon was modified slightly for this work; however, he did all of the hard work in finding the right places to set the channel in the RTP engine across the channel drivers. Much thanks goes to Jaco for his hard work here. Review: https://reviewboard.asterisk.org/r/2603/ (closes issue ASTERISK-20574) Reported by: Jaco Kroon patches: asterisk-rtcp-channel.patch uploaded by jkroon (License 5671) (closes issue ASTERISK-21471) Reported by: Matt Jordan git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@393740 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2013-07-05 17:33:33 +00:00
/*! \brief \ref stasis topic for RTP related messages */
static struct stasis_topic *rtp_topic;
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
/*!
* \brief Set given json object into target with name
*
* \param target Target json.
* \param name key of given object.
* \param obj Json value will be set.
*/
#define SET_AST_JSON_OBJ(target, name, obj) ({ \
struct ast_json *j_tmp = obj; \
if (j_tmp) { \
ast_json_object_set(target, name, j_tmp); \
} \
})
/*!
* \internal
* \brief Destructor for \c ast_rtp_payload_type
*/
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
static void rtp_payload_type_dtor(void *obj)
{
struct ast_rtp_payload_type *payload = obj;
ao2_cleanup(payload->format);
}
static struct ast_rtp_payload_type *rtp_payload_type_alloc(struct ast_format *format,
int payload, int rtp_code, int primary_mapping)
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
{
struct ast_rtp_payload_type *type = ao2_alloc_options(
sizeof(*type), rtp_payload_type_dtor, AO2_ALLOC_OPT_LOCK_NOLOCK);
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
if (!type) {
return NULL;
}
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
type->format = ao2_bump(format);
type->asterisk_format = type->format != NULL;
type->payload = payload;
type->rtp_code = rtp_code;
type->primary_mapping = primary_mapping;
return type;
}
struct ast_rtp_payload_type *ast_rtp_engine_alloc_payload_type(void)
{
return rtp_payload_type_alloc(NULL, 0, 0, 0);
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
}
int ast_rtp_engine_register2(struct ast_rtp_engine *engine, struct ast_module *module)
{
struct ast_rtp_engine *current_engine;
/* Perform a sanity check on the engine structure to make sure it has the basics */
if (ast_strlen_zero(engine->name) || !engine->new || !engine->destroy || !engine->write || !engine->read) {
ast_log(LOG_WARNING, "RTP Engine '%s' failed sanity check so it was not registered.\n", !ast_strlen_zero(engine->name) ? engine->name : "Unknown");
return -1;
}
/* Link owner module to the RTP engine for reference counting purposes */
engine->mod = module;
AST_RWLIST_WRLOCK(&engines);
/* Ensure that no two modules with the same name are registered at the same time */
AST_RWLIST_TRAVERSE(&engines, current_engine, entry) {
if (!strcmp(current_engine->name, engine->name)) {
ast_log(LOG_WARNING, "An RTP engine with the name '%s' has already been registered.\n", engine->name);
AST_RWLIST_UNLOCK(&engines);
return -1;
}
}
/* The engine survived our critique. Off to the list it goes to be used */
AST_RWLIST_INSERT_TAIL(&engines, engine, entry);
AST_RWLIST_UNLOCK(&engines);
ast_verb(2, "Registered RTP engine '%s'\n", engine->name);
return 0;
}
int ast_rtp_engine_unregister(struct ast_rtp_engine *engine)
{
struct ast_rtp_engine *current_engine = NULL;
AST_RWLIST_WRLOCK(&engines);
if ((current_engine = AST_RWLIST_REMOVE(&engines, engine, entry))) {
ast_verb(2, "Unregistered RTP engine '%s'\n", engine->name);
}
AST_RWLIST_UNLOCK(&engines);
return current_engine ? 0 : -1;
}
int ast_rtp_glue_register2(struct ast_rtp_glue *glue, struct ast_module *module)
{
struct ast_rtp_glue *current_glue = NULL;
if (ast_strlen_zero(glue->type)) {
return -1;
}
glue->mod = module;
AST_RWLIST_WRLOCK(&glues);
AST_RWLIST_TRAVERSE(&glues, current_glue, entry) {
if (!strcasecmp(current_glue->type, glue->type)) {
ast_log(LOG_WARNING, "RTP glue with the name '%s' has already been registered.\n", glue->type);
AST_RWLIST_UNLOCK(&glues);
return -1;
}
}
AST_RWLIST_INSERT_TAIL(&glues, glue, entry);
AST_RWLIST_UNLOCK(&glues);
ast_verb(2, "Registered RTP glue '%s'\n", glue->type);
return 0;
}
int ast_rtp_glue_unregister(struct ast_rtp_glue *glue)
{
struct ast_rtp_glue *current_glue = NULL;
AST_RWLIST_WRLOCK(&glues);
if ((current_glue = AST_RWLIST_REMOVE(&glues, glue, entry))) {
ast_verb(2, "Unregistered RTP glue '%s'\n", glue->type);
}
AST_RWLIST_UNLOCK(&glues);
return current_glue ? 0 : -1;
}
static void instance_destructor(void *obj)
{
struct ast_rtp_instance *instance = obj;
/* Pass us off to the engine to destroy */
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
if (instance->data) {
/*
* Lock in case the RTP engine has other threads that
* need synchronization with the destruction.
*/
ao2_lock(instance);
instance->engine->destroy(instance);
ao2_unlock(instance);
}
if (instance->srtp) {
res_srtp->destroy(instance->srtp);
}
if (instance->rtcp_srtp) {
res_srtp->destroy(instance->rtcp_srtp);
}
ast_rtp_codecs_payloads_destroy(&instance->codecs);
AST_VECTOR_FREE(&instance->extmap_enabled);
AST_VECTOR_FREE(&instance->extmap_negotiated);
AST_VECTOR_FREE(&instance->extmap_unique_ids);
/* Drop our engine reference */
ast_module_unref(instance->engine->mod);
ast_debug(1, "Destroyed RTP instance '%p'\n", instance);
}
int ast_rtp_instance_destroy(struct ast_rtp_instance *instance)
{
ao2_cleanup(instance);
return 0;
}
struct ast_rtp_instance *ast_rtp_instance_new(const char *engine_name,
struct ast_sched_context *sched, const struct ast_sockaddr *sa,
void *data)
{
struct ast_sockaddr address = {{0,}};
struct ast_rtp_instance *instance = NULL;
struct ast_rtp_engine *engine = NULL;
struct ast_module *mod_ref;
AST_RWLIST_RDLOCK(&engines);
/* If an engine name was specified try to use it or otherwise use the first one registered */
if (!ast_strlen_zero(engine_name)) {
AST_RWLIST_TRAVERSE(&engines, engine, entry) {
if (!strcmp(engine->name, engine_name)) {
break;
}
}
} else {
engine = AST_RWLIST_FIRST(&engines);
}
/* If no engine was actually found bail out now */
if (!engine) {
ast_log(LOG_ERROR, "No RTP engine was found. Do you have one loaded?\n");
AST_RWLIST_UNLOCK(&engines);
return NULL;
}
/* Bump up the reference count before we return so the module can not be unloaded */
mod_ref = ast_module_running_ref(engine->mod);
AST_RWLIST_UNLOCK(&engines);
if (!mod_ref) {
/* BUGBUG: improve handling of this situation. */
return NULL;
}
/* Allocate a new RTP instance */
if (!(instance = ao2_alloc(sizeof(*instance), instance_destructor))) {
ast_module_unref(engine->mod);
return NULL;
}
instance->engine = engine;
ast_sockaddr_copy(&instance->local_address, sa);
ast_sockaddr_copy(&address, sa);
if (ast_rtp_codecs_payloads_initialize(&instance->codecs)) {
ao2_ref(instance, -1);
return NULL;
}
/* Initialize RTP extension support */
if (AST_VECTOR_INIT(&instance->extmap_enabled, 0) ||
AST_VECTOR_INIT(&instance->extmap_negotiated, 0) ||
AST_VECTOR_INIT(&instance->extmap_unique_ids, 0)) {
ao2_ref(instance, -1);
return NULL;
}
ast_debug(1, "Using engine '%s' for RTP instance '%p'\n", engine->name, instance);
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
/*
* And pass it off to the engine to setup
*
* Lock in case the RTP engine has other threads that
* need synchronization with the construction.
*/
ao2_lock(instance);
if (instance->engine->new(instance, sched, &address, data)) {
ast_debug(1, "Engine '%s' failed to setup RTP instance '%p'\n", engine->name, instance);
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_unlock(instance);
ao2_ref(instance, -1);
return NULL;
}
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_unlock(instance);
ast_debug(1, "RTP instance '%p' is setup and ready to go\n", instance);
return instance;
}
Refactor RTCP events over to Stasis; associate with channels This patch does the following: * It merges Jaco Kroon's patch from ASTERISK-20754, which provides channel information in the RTCP events. Because Stasis provides a cache, Jaco's patch was modified to pass the channel uniqueid to the RTP layer as opposed to a pointer to the channel. This has the following benefits: (1) It keeps the RTP engine 'clean' of references back to channels (2) It prevents circular dependencies and other potential ref counting issues * The RTP engine now allows any RTP implementation to raise RTCP messages. Potentially, other implementations (such as res_rtp_multicast) could also raise RTCP information. The engine provides structs to represent RTCP headers and RTCP SR/RR reports. * Some general refactoring in res_rtp_asterisk was done to try and tame the RTCP code. It isn't perfect - that's *way* beyond the scope of this work - but it does feel marginally better. * A few random bugs were fixed in the RTCP statistics. (Example: performing an assignment of a = a is probably not correct) * We now raise RTCP events for each SR/RR sent/received. Previously we wouldn't raise an event when we sent a RR report. Note that this work will be of use to others who want to monitor call quality or build modules that report call quality statistics. Since the events are now moving across the Stasis message bus, this is far easier to accomplish. It is also a first step (though by no means the last step) towards getting Olle's pinefrog work incorporated. Again: note that the patch by Jaco Kroon was modified slightly for this work; however, he did all of the hard work in finding the right places to set the channel in the RTP engine across the channel drivers. Much thanks goes to Jaco for his hard work here. Review: https://reviewboard.asterisk.org/r/2603/ (closes issue ASTERISK-20574) Reported by: Jaco Kroon patches: asterisk-rtcp-channel.patch uploaded by jkroon (License 5671) (closes issue ASTERISK-21471) Reported by: Matt Jordan git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@393740 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2013-07-05 17:33:33 +00:00
const char *ast_rtp_instance_get_channel_id(struct ast_rtp_instance *instance)
{
return instance->channel_uniqueid;
}
void ast_rtp_instance_set_channel_id(struct ast_rtp_instance *instance, const char *uniqueid)
{
ast_copy_string(instance->channel_uniqueid, uniqueid, sizeof(instance->channel_uniqueid));
}
void ast_rtp_instance_set_data(struct ast_rtp_instance *instance, void *data)
{
instance->data = data;
}
void *ast_rtp_instance_get_data(struct ast_rtp_instance *instance)
{
return instance->data;
}
int ast_rtp_instance_write(struct ast_rtp_instance *instance, struct ast_frame *frame)
{
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
int res;
ao2_lock(instance);
res = instance->engine->write(instance, frame);
ao2_unlock(instance);
return res;
}
struct ast_frame *ast_rtp_instance_read(struct ast_rtp_instance *instance, int rtcp)
{
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
struct ast_frame *frame;
ao2_lock(instance);
frame = instance->engine->read(instance, rtcp);
ao2_unlock(instance);
return frame;
}
int ast_rtp_instance_set_local_address(struct ast_rtp_instance *instance,
const struct ast_sockaddr *address)
{
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_lock(instance);
ast_sockaddr_copy(&instance->local_address, address);
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_unlock(instance);
return 0;
}
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
static void rtp_instance_set_incoming_source_address_nolock(struct ast_rtp_instance *instance,
const struct ast_sockaddr *address)
{
ast_sockaddr_copy(&instance->incoming_source_address, address);
if (instance->engine->remote_address_set) {
instance->engine->remote_address_set(instance, &instance->incoming_source_address);
}
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
}
int ast_rtp_instance_set_incoming_source_address(struct ast_rtp_instance *instance,
const struct ast_sockaddr *address)
{
ao2_lock(instance);
rtp_instance_set_incoming_source_address_nolock(instance, address);
ao2_unlock(instance);
return 0;
}
int ast_rtp_instance_set_requested_target_address(struct ast_rtp_instance *instance,
const struct ast_sockaddr *address)
{
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_lock(instance);
ast_sockaddr_copy(&instance->requested_target_address, address);
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
rtp_instance_set_incoming_source_address_nolock(instance, address);
ao2_unlock(instance);
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
return 0;
}
Merged revisions 293803 via svnmerge from https://origsvn.digium.com/svn/asterisk/branches/1.8 ........ r293803 | twilson | 2010-11-03 11:05:14 -0700 (Wed, 03 Nov 2010) | 25 lines Avoid valgrind warnings for ast_rtp_instance_get_xxx_address The documentation for ast_rtp_instance_get_(local/remote)_address stated that they returned 0 for success and -1 on failure. Instead, they returned 0 if the address structure passed in was already equivalent to the address instance local/remote address or 1 otherwise. 90% of the calls to these functions completely ignored the return address and passed in an uninitialized struct, which would make valgrind complain even though the operation was technically safe. This patch fixes the documentation and converts the get_xxx_address functions to void since all they really do is copy the address and cannot fail. Additionally two new functions (ast_rtp_instance_get_and_cmp_(local/remote)_address) are created for the 3 times where the return value was actually checked. The get_and_cmp_local_address function is currently unused, but exists for the sake of symmetry. The only functional change as a result of this change is that we will not do an ast_sockaddr_cmp() on (mostly uninitialized) addresses before doing the ast_sockaddr_copy() in the get_*_address functions. So, even though it is an API change, it shouldn't have a noticeable change in behavior. Review: https://reviewboard.asterisk.org/r/995/ ........ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@293809 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2010-11-03 18:43:18 +00:00
int ast_rtp_instance_get_and_cmp_local_address(struct ast_rtp_instance *instance,
struct ast_sockaddr *address)
{
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_lock(instance);
if (ast_sockaddr_cmp(address, &instance->local_address) != 0) {
ast_sockaddr_copy(address, &instance->local_address);
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_unlock(instance);
return 1;
}
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_unlock(instance);
return 0;
}
Merged revisions 293803 via svnmerge from https://origsvn.digium.com/svn/asterisk/branches/1.8 ........ r293803 | twilson | 2010-11-03 11:05:14 -0700 (Wed, 03 Nov 2010) | 25 lines Avoid valgrind warnings for ast_rtp_instance_get_xxx_address The documentation for ast_rtp_instance_get_(local/remote)_address stated that they returned 0 for success and -1 on failure. Instead, they returned 0 if the address structure passed in was already equivalent to the address instance local/remote address or 1 otherwise. 90% of the calls to these functions completely ignored the return address and passed in an uninitialized struct, which would make valgrind complain even though the operation was technically safe. This patch fixes the documentation and converts the get_xxx_address functions to void since all they really do is copy the address and cannot fail. Additionally two new functions (ast_rtp_instance_get_and_cmp_(local/remote)_address) are created for the 3 times where the return value was actually checked. The get_and_cmp_local_address function is currently unused, but exists for the sake of symmetry. The only functional change as a result of this change is that we will not do an ast_sockaddr_cmp() on (mostly uninitialized) addresses before doing the ast_sockaddr_copy() in the get_*_address functions. So, even though it is an API change, it shouldn't have a noticeable change in behavior. Review: https://reviewboard.asterisk.org/r/995/ ........ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@293809 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2010-11-03 18:43:18 +00:00
void ast_rtp_instance_get_local_address(struct ast_rtp_instance *instance,
struct ast_sockaddr *address)
{
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_lock(instance);
Merged revisions 293803 via svnmerge from https://origsvn.digium.com/svn/asterisk/branches/1.8 ........ r293803 | twilson | 2010-11-03 11:05:14 -0700 (Wed, 03 Nov 2010) | 25 lines Avoid valgrind warnings for ast_rtp_instance_get_xxx_address The documentation for ast_rtp_instance_get_(local/remote)_address stated that they returned 0 for success and -1 on failure. Instead, they returned 0 if the address structure passed in was already equivalent to the address instance local/remote address or 1 otherwise. 90% of the calls to these functions completely ignored the return address and passed in an uninitialized struct, which would make valgrind complain even though the operation was technically safe. This patch fixes the documentation and converts the get_xxx_address functions to void since all they really do is copy the address and cannot fail. Additionally two new functions (ast_rtp_instance_get_and_cmp_(local/remote)_address) are created for the 3 times where the return value was actually checked. The get_and_cmp_local_address function is currently unused, but exists for the sake of symmetry. The only functional change as a result of this change is that we will not do an ast_sockaddr_cmp() on (mostly uninitialized) addresses before doing the ast_sockaddr_copy() in the get_*_address functions. So, even though it is an API change, it shouldn't have a noticeable change in behavior. Review: https://reviewboard.asterisk.org/r/995/ ........ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@293809 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2010-11-03 18:43:18 +00:00
ast_sockaddr_copy(address, &instance->local_address);
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_unlock(instance);
Merged revisions 293803 via svnmerge from https://origsvn.digium.com/svn/asterisk/branches/1.8 ........ r293803 | twilson | 2010-11-03 11:05:14 -0700 (Wed, 03 Nov 2010) | 25 lines Avoid valgrind warnings for ast_rtp_instance_get_xxx_address The documentation for ast_rtp_instance_get_(local/remote)_address stated that they returned 0 for success and -1 on failure. Instead, they returned 0 if the address structure passed in was already equivalent to the address instance local/remote address or 1 otherwise. 90% of the calls to these functions completely ignored the return address and passed in an uninitialized struct, which would make valgrind complain even though the operation was technically safe. This patch fixes the documentation and converts the get_xxx_address functions to void since all they really do is copy the address and cannot fail. Additionally two new functions (ast_rtp_instance_get_and_cmp_(local/remote)_address) are created for the 3 times where the return value was actually checked. The get_and_cmp_local_address function is currently unused, but exists for the sake of symmetry. The only functional change as a result of this change is that we will not do an ast_sockaddr_cmp() on (mostly uninitialized) addresses before doing the ast_sockaddr_copy() in the get_*_address functions. So, even though it is an API change, it shouldn't have a noticeable change in behavior. Review: https://reviewboard.asterisk.org/r/995/ ........ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@293809 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2010-11-03 18:43:18 +00:00
}
int ast_rtp_instance_get_and_cmp_requested_target_address(struct ast_rtp_instance *instance,
struct ast_sockaddr *address)
{
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_lock(instance);
if (ast_sockaddr_cmp(address, &instance->requested_target_address) != 0) {
ast_sockaddr_copy(address, &instance->requested_target_address);
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_unlock(instance);
return 1;
}
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_unlock(instance);
return 0;
}
void ast_rtp_instance_get_incoming_source_address(struct ast_rtp_instance *instance,
struct ast_sockaddr *address)
{
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_lock(instance);
ast_sockaddr_copy(address, &instance->incoming_source_address);
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_unlock(instance);
}
void ast_rtp_instance_get_requested_target_address(struct ast_rtp_instance *instance,
struct ast_sockaddr *address)
Merged revisions 293803 via svnmerge from https://origsvn.digium.com/svn/asterisk/branches/1.8 ........ r293803 | twilson | 2010-11-03 11:05:14 -0700 (Wed, 03 Nov 2010) | 25 lines Avoid valgrind warnings for ast_rtp_instance_get_xxx_address The documentation for ast_rtp_instance_get_(local/remote)_address stated that they returned 0 for success and -1 on failure. Instead, they returned 0 if the address structure passed in was already equivalent to the address instance local/remote address or 1 otherwise. 90% of the calls to these functions completely ignored the return address and passed in an uninitialized struct, which would make valgrind complain even though the operation was technically safe. This patch fixes the documentation and converts the get_xxx_address functions to void since all they really do is copy the address and cannot fail. Additionally two new functions (ast_rtp_instance_get_and_cmp_(local/remote)_address) are created for the 3 times where the return value was actually checked. The get_and_cmp_local_address function is currently unused, but exists for the sake of symmetry. The only functional change as a result of this change is that we will not do an ast_sockaddr_cmp() on (mostly uninitialized) addresses before doing the ast_sockaddr_copy() in the get_*_address functions. So, even though it is an API change, it shouldn't have a noticeable change in behavior. Review: https://reviewboard.asterisk.org/r/995/ ........ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@293809 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2010-11-03 18:43:18 +00:00
{
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_lock(instance);
ast_sockaddr_copy(address, &instance->requested_target_address);
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_unlock(instance);
Merged revisions 293803 via svnmerge from https://origsvn.digium.com/svn/asterisk/branches/1.8 ........ r293803 | twilson | 2010-11-03 11:05:14 -0700 (Wed, 03 Nov 2010) | 25 lines Avoid valgrind warnings for ast_rtp_instance_get_xxx_address The documentation for ast_rtp_instance_get_(local/remote)_address stated that they returned 0 for success and -1 on failure. Instead, they returned 0 if the address structure passed in was already equivalent to the address instance local/remote address or 1 otherwise. 90% of the calls to these functions completely ignored the return address and passed in an uninitialized struct, which would make valgrind complain even though the operation was technically safe. This patch fixes the documentation and converts the get_xxx_address functions to void since all they really do is copy the address and cannot fail. Additionally two new functions (ast_rtp_instance_get_and_cmp_(local/remote)_address) are created for the 3 times where the return value was actually checked. The get_and_cmp_local_address function is currently unused, but exists for the sake of symmetry. The only functional change as a result of this change is that we will not do an ast_sockaddr_cmp() on (mostly uninitialized) addresses before doing the ast_sockaddr_copy() in the get_*_address functions. So, even though it is an API change, it shouldn't have a noticeable change in behavior. Review: https://reviewboard.asterisk.org/r/995/ ........ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@293809 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2010-11-03 18:43:18 +00:00
}
void ast_rtp_instance_set_extended_prop(struct ast_rtp_instance *instance, int property, void *value)
{
if (instance->engine->extended_prop_set) {
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_lock(instance);
instance->engine->extended_prop_set(instance, property, value);
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_unlock(instance);
}
}
void *ast_rtp_instance_get_extended_prop(struct ast_rtp_instance *instance, int property)
{
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
void *prop;
if (instance->engine->extended_prop_get) {
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_lock(instance);
prop = instance->engine->extended_prop_get(instance, property);
ao2_unlock(instance);
} else {
prop = NULL;
}
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
return prop;
}
void ast_rtp_instance_set_prop(struct ast_rtp_instance *instance, enum ast_rtp_property property, int value)
{
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_lock(instance);
instance->properties[property] = value;
if (instance->engine->prop_set) {
instance->engine->prop_set(instance, property, value);
}
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_unlock(instance);
}
int ast_rtp_instance_get_prop(struct ast_rtp_instance *instance, enum ast_rtp_property property)
{
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
int prop;
ao2_lock(instance);
prop = instance->properties[property];
ao2_unlock(instance);
return prop;
}
struct ast_rtp_codecs *ast_rtp_instance_get_codecs(struct ast_rtp_instance *instance)
{
return &instance->codecs;
}
int ast_rtp_instance_extmap_enable(struct ast_rtp_instance *instance, int id, enum ast_rtp_extension extension,
enum ast_rtp_extension_direction direction)
{
struct rtp_extmap extmap = {
.extension = extension,
.direction = direction,
};
ao2_lock(instance);
if (!instance->engine->extension_enable || !instance->engine->extension_enable(instance, extension)) {
ao2_unlock(instance);
return 0;
}
/* We store enabled extensions separately so we can easily do negotiation */
if (AST_VECTOR_REPLACE(&instance->extmap_enabled, extension, direction)) {
ao2_unlock(instance);
return -1;
}
if (id <= 0) {
/* We find a free unique identifier for this extension by just appending it to the
* vector of unique ids. The size of the vector will become its unique identifier.
* As well when we are asking for information on the extensions it will be returned,
* allowing it to be added to the SDP offer.
*/
if (AST_VECTOR_APPEND(&instance->extmap_unique_ids, extmap)) {
AST_VECTOR_REPLACE(&instance->extmap_enabled, extension, AST_RTP_EXTENSION_DIRECTION_NONE);
ao2_unlock(instance);
return -1;
}
id = AST_VECTOR_SIZE(&instance->extmap_unique_ids);
} else {
/* Otherwise we put it precisely where they want it */
if (AST_VECTOR_REPLACE(&instance->extmap_unique_ids, id - 1, extmap)) {
AST_VECTOR_REPLACE(&instance->extmap_enabled, extension, AST_RTP_EXTENSION_DIRECTION_NONE);
ao2_unlock(instance);
return -1;
}
}
/* Now that we have an id add the extension to here */
if (AST_VECTOR_REPLACE(&instance->extmap_negotiated, extension, id)) {
extmap.extension = AST_RTP_EXTENSION_UNSUPPORTED;
extmap.direction = AST_RTP_EXTENSION_DIRECTION_NONE;
AST_VECTOR_REPLACE(&instance->extmap_enabled, extension, AST_RTP_EXTENSION_DIRECTION_NONE);
AST_VECTOR_REPLACE(&instance->extmap_unique_ids, id - 1, extmap);
ao2_unlock(instance);
return -1;
}
ao2_unlock(instance);
return 0;
}
/*! \brief Helper function which negotiates two RTP extension directions to get our current direction */
static enum ast_rtp_extension_direction rtp_extmap_negotiate_direction(enum ast_rtp_extension_direction ours,
enum ast_rtp_extension_direction theirs)
{
if (theirs == AST_RTP_EXTENSION_DIRECTION_NONE || ours == AST_RTP_EXTENSION_DIRECTION_NONE) {
/* This should not occur but if it does tolerate either side not having this extension
* in use.
*/
return AST_RTP_EXTENSION_DIRECTION_NONE;
} else if (theirs == AST_RTP_EXTENSION_DIRECTION_INACTIVE) {
/* Inactive is always inactive on our side */
return AST_RTP_EXTENSION_DIRECTION_INACTIVE;
} else if (theirs == AST_RTP_EXTENSION_DIRECTION_SENDRECV) {
return ours;
} else if (theirs == AST_RTP_EXTENSION_DIRECTION_SENDONLY) {
/* If they are send only then we become recvonly if we are configured as sendrecv or recvonly */
if (ours == AST_RTP_EXTENSION_DIRECTION_SENDRECV || ours == AST_RTP_EXTENSION_DIRECTION_RECVONLY) {
return AST_RTP_EXTENSION_DIRECTION_RECVONLY;
}
} else if (theirs == AST_RTP_EXTENSION_DIRECTION_RECVONLY) {
/* If they are recv only then we become sendonly if we are configured as sendrecv or sendonly */
if (ours == AST_RTP_EXTENSION_DIRECTION_SENDRECV || ours == AST_RTP_EXTENSION_DIRECTION_SENDONLY) {
return AST_RTP_EXTENSION_DIRECTION_SENDONLY;
}
}
return AST_RTP_EXTENSION_DIRECTION_NONE;
}
int ast_rtp_instance_extmap_negotiate(struct ast_rtp_instance *instance, int id, enum ast_rtp_extension_direction direction,
const char *uri, const char *attributes)
{
/* 'attributes' is currently unused but exists in the API to ensure it does not need to be altered
* in the future in case we need to use it.
*/
int idx;
enum ast_rtp_extension extension = AST_RTP_EXTENSION_UNSUPPORTED;
/* Per the RFC the identifier has to be 1 or above */
if (id < 1) {
return -1;
}
/* Convert the provided URI to the internal representation */
for (idx = 0; idx < ARRAY_LEN(rtp_extension_uris); ++idx) {
if (!strcasecmp(rtp_extension_uris[idx], uri)) {
extension = idx;
break;
}
}
ao2_lock(instance);
/* We only accept the extension if it is enabled */
if (extension < AST_VECTOR_SIZE(&instance->extmap_enabled) &&
AST_VECTOR_GET(&instance->extmap_enabled, extension) != AST_RTP_EXTENSION_DIRECTION_NONE) {
struct rtp_extmap extmap = {
.extension = extension,
.direction = rtp_extmap_negotiate_direction(AST_VECTOR_GET(&instance->extmap_enabled, extension), direction),
};
/* If the direction negotiation failed then don't accept or use this extension */
if (extmap.direction != AST_RTP_EXTENSION_DIRECTION_NONE) {
if (extension != AST_RTP_EXTENSION_UNSUPPORTED) {
AST_VECTOR_REPLACE(&instance->extmap_negotiated, extension, id);
}
AST_VECTOR_REPLACE(&instance->extmap_unique_ids, id - 1, extmap);
}
}
ao2_unlock(instance);
return 0;
}
void ast_rtp_instance_extmap_clear(struct ast_rtp_instance *instance)
{
static const struct rtp_extmap extmap_none = {
.extension = AST_RTP_EXTENSION_UNSUPPORTED,
.direction = AST_RTP_EXTENSION_DIRECTION_NONE,
};
int idx;
ao2_lock(instance);
/* Clear both the known unique ids and the negotiated extensions as we are about to have
* new results set on us.
*/
for (idx = 0; idx < AST_VECTOR_SIZE(&instance->extmap_unique_ids); ++idx) {
AST_VECTOR_REPLACE(&instance->extmap_unique_ids, idx, extmap_none);
}
for (idx = 0; idx < AST_VECTOR_SIZE(&instance->extmap_negotiated); ++idx) {
AST_VECTOR_REPLACE(&instance->extmap_negotiated, idx, -1);
}
ao2_unlock(instance);
}
int ast_rtp_instance_extmap_get_id(struct ast_rtp_instance *instance, enum ast_rtp_extension extension)
{
int id = -1;
ao2_lock(instance);
if (extension < AST_VECTOR_SIZE(&instance->extmap_negotiated)) {
id = AST_VECTOR_GET(&instance->extmap_negotiated, extension);
}
ao2_unlock(instance);
return id;
}
size_t ast_rtp_instance_extmap_count(struct ast_rtp_instance *instance)
{
size_t count;
ao2_lock(instance);
count = AST_VECTOR_SIZE(&instance->extmap_unique_ids);
ao2_unlock(instance);
return count;
}
enum ast_rtp_extension ast_rtp_instance_extmap_get_extension(struct ast_rtp_instance *instance, int id)
{
enum ast_rtp_extension extension = AST_RTP_EXTENSION_UNSUPPORTED;
ao2_lock(instance);
/* The local unique identifier starts at '1' so the highest unique identifier
* can be the actual size of the vector. We compensate (as it is 0 index based)
* by dropping it down to 1 to get the correct information.
*/
if (0 < id && id <= AST_VECTOR_SIZE(&instance->extmap_unique_ids)) {
struct rtp_extmap *extmap = AST_VECTOR_GET_ADDR(&instance->extmap_unique_ids, id - 1);
extension = extmap->extension;
}
ao2_unlock(instance);
return extension;
}
enum ast_rtp_extension_direction ast_rtp_instance_extmap_get_direction(struct ast_rtp_instance *instance, int id)
{
enum ast_rtp_extension_direction direction = AST_RTP_EXTENSION_DIRECTION_NONE;
ao2_lock(instance);
if (0 < id && id <= AST_VECTOR_SIZE(&instance->extmap_unique_ids)) {
struct rtp_extmap *extmap = AST_VECTOR_GET_ADDR(&instance->extmap_unique_ids, id - 1);
direction = extmap->direction;
}
ao2_unlock(instance);
return direction;
}
const char *ast_rtp_instance_extmap_get_uri(struct ast_rtp_instance *instance, int id)
{
enum ast_rtp_extension extension = ast_rtp_instance_extmap_get_extension(instance, id);
if (extension == AST_RTP_EXTENSION_UNSUPPORTED ||
(unsigned int)extension >= ARRAY_LEN(rtp_extension_uris)) {
return NULL;
}
return rtp_extension_uris[extension];
}
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
int ast_rtp_codecs_payloads_initialize(struct ast_rtp_codecs *codecs)
{
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
int res;
codecs->framing = 0;
ast_rwlock_init(&codecs->codecs_lock);
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
res = AST_VECTOR_INIT(&codecs->payload_mapping_rx, AST_RTP_MAX_PT);
res |= AST_VECTOR_INIT(&codecs->payload_mapping_tx, AST_RTP_MAX_PT);
if (res) {
AST_VECTOR_FREE(&codecs->payload_mapping_rx);
AST_VECTOR_FREE(&codecs->payload_mapping_tx);
}
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
return res;
}
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
void ast_rtp_codecs_payloads_destroy(struct ast_rtp_codecs *codecs)
{
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
int idx;
struct ast_rtp_payload_type *type;
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
for (idx = 0; idx < AST_VECTOR_SIZE(&codecs->payload_mapping_rx); ++idx) {
type = AST_VECTOR_GET(&codecs->payload_mapping_rx, idx);
ao2_t_cleanup(type, "destroying ast_rtp_codec rx mapping");
}
AST_VECTOR_FREE(&codecs->payload_mapping_rx);
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
for (idx = 0; idx < AST_VECTOR_SIZE(&codecs->payload_mapping_tx); ++idx) {
type = AST_VECTOR_GET(&codecs->payload_mapping_tx, idx);
ao2_t_cleanup(type, "destroying ast_rtp_codec tx mapping");
}
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
AST_VECTOR_FREE(&codecs->payload_mapping_tx);
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
ast_rwlock_destroy(&codecs->codecs_lock);
}
void ast_rtp_codecs_payloads_clear(struct ast_rtp_codecs *codecs, struct ast_rtp_instance *instance)
{
ast_rtp_codecs_payloads_destroy(codecs);
ast_rtp_codecs_payloads_initialize(codecs);
if (instance && instance->engine && instance->engine->payload_set) {
int i;
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_lock(instance);
for (i = 0; i < AST_RTP_MAX_PT; i++) {
instance->engine->payload_set(instance, i, 0, NULL, 0);
}
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_unlock(instance);
}
}
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
/*!
* \internal
* \brief Clear the rx primary mapping flag on all other matching mappings.
* \since 14.0.0
*
* \param codecs Codecs that need rx clearing.
* \param to_match Payload type object to compare against.
*
* \note It is assumed that codecs is write locked before calling.
*/
static void payload_mapping_rx_clear_primary(struct ast_rtp_codecs *codecs, struct ast_rtp_payload_type *to_match)
{
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
int idx;
struct ast_rtp_payload_type *current;
struct ast_rtp_payload_type *new_type;
struct timeval now;
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
if (!to_match->primary_mapping) {
return;
}
now = ast_tvnow();
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
for (idx = 0; idx < AST_VECTOR_SIZE(&codecs->payload_mapping_rx); ++idx) {
current = AST_VECTOR_GET(&codecs->payload_mapping_rx, idx);
if (!current || current == to_match || !current->primary_mapping) {
continue;
}
if (current->asterisk_format && to_match->asterisk_format) {
if (ast_format_cmp(current->format, to_match->format) == AST_FORMAT_CMP_NOT_EQUAL) {
continue;
}
} else if (!current->asterisk_format && !to_match->asterisk_format) {
if (current->rtp_code != to_match->rtp_code) {
continue;
}
} else {
continue;
}
/* Replace current with non-primary marked version */
new_type = ast_rtp_engine_alloc_payload_type();
if (!new_type) {
continue;
}
*new_type = *current;
new_type->primary_mapping = 0;
new_type->when_retired = now;
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
ao2_bump(new_type->format);
AST_VECTOR_REPLACE(&codecs->payload_mapping_rx, idx, new_type);
ao2_ref(current, -1);
}
}
/*!
* \internal
* \brief Put the new_type into the rx payload type mapping.
* \since 14.0.0
*
* \param codecs Codecs structure to put new_type into
* \param payload type position to replace.
* \param new_type RTP payload mapping object to store.
*
* \note It is assumed that codecs is write locked before calling.
*/
static void rtp_codecs_payload_replace_rx(struct ast_rtp_codecs *codecs, int payload, struct ast_rtp_payload_type *new_type)
{
ao2_ref(new_type, +1);
if (payload < AST_VECTOR_SIZE(&codecs->payload_mapping_rx)) {
ao2_t_cleanup(AST_VECTOR_GET(&codecs->payload_mapping_rx, payload),
"cleaning up rx mapping vector element about to be replaced");
}
if (AST_VECTOR_REPLACE(&codecs->payload_mapping_rx, payload, new_type)) {
ao2_ref(new_type, -1);
return;
}
payload_mapping_rx_clear_primary(codecs, new_type);
}
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
/*!
* \internal
* \brief Copy the rx payload type mapping to the destination.
* \since 14.0.0
*
* \param src The source codecs structure
* \param dest The destination codecs structure that the values from src will be copied to
* \param instance Optionally the instance that the dst codecs structure belongs to
*
* \note It is assumed that src is at least read locked before calling.
* \note It is assumed that dest is write locked before calling.
*/
static void rtp_codecs_payloads_copy_rx(struct ast_rtp_codecs *src, struct ast_rtp_codecs *dest, struct ast_rtp_instance *instance)
{
int idx;
struct ast_rtp_payload_type *type;
for (idx = 0; idx < AST_VECTOR_SIZE(&src->payload_mapping_rx); ++idx) {
type = AST_VECTOR_GET(&src->payload_mapping_rx, idx);
if (!type) {
continue;
}
ast_debug(2, "Copying rx payload mapping %d (%p) from %p to %p\n",
idx, type, src, dest);
rtp_codecs_payload_replace_rx(dest, idx, type);
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
if (instance && instance->engine && instance->engine->payload_set) {
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_lock(instance);
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
instance->engine->payload_set(instance, idx, type->asterisk_format, type->format, type->rtp_code);
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_unlock(instance);
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
}
}
}
/*!
* \internal
* \brief Determine if a type of payload is already present in mappings.
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
* \since 14.0.0
*
* \param codecs Codecs to be checked for mappings.
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
* \param to_match Payload type object to compare against.
*
* \note It is assumed that codecs is write locked before calling.
*
* \retval 0 not found
* \retval 1 found
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
*/
static int payload_mapping_tx_is_present(const struct ast_rtp_codecs *codecs, const struct ast_rtp_payload_type *to_match)
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
{
int idx;
struct ast_rtp_payload_type *current;
for (idx = 0; idx < AST_VECTOR_SIZE(&codecs->payload_mapping_tx); ++idx) {
current = AST_VECTOR_GET(&codecs->payload_mapping_tx, idx);
if (!current) {
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
continue;
}
if (current == to_match) {
/* The exact object is already in the mapping. */
return 1;
}
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
if (current->asterisk_format && to_match->asterisk_format) {
if (ast_format_get_codec_id(current->format) != ast_format_get_codec_id(to_match->format)) {
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
continue;
} else if (current->payload == to_match->payload) {
return 0;
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
}
} else if (!current->asterisk_format && !to_match->asterisk_format) {
if (current->rtp_code != to_match->rtp_code) {
continue;
}
} else {
continue;
}
return 1;
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
}
return 0;
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
}
/*!
* \internal
* \brief Copy the tx payload type mapping to the destination.
* \since 14.0.0
*
* \param src The source codecs structure
* \param dest The destination codecs structure that the values from src will be copied to
* \param instance Optionally the instance that the dst codecs structure belongs to
*
* \note It is assumed that src is at least read locked before calling.
* \note It is assumed that dest is write locked before calling.
*/
static void rtp_codecs_payloads_copy_tx(struct ast_rtp_codecs *src, struct ast_rtp_codecs *dest, struct ast_rtp_instance *instance)
{
int idx;
struct ast_rtp_payload_type *type;
for (idx = 0; idx < AST_VECTOR_SIZE(&src->payload_mapping_tx); ++idx) {
type = AST_VECTOR_GET(&src->payload_mapping_tx, idx);
if (!type) {
continue;
}
ast_debug(2, "Copying tx payload mapping %d (%p) from %p to %p\n",
idx, type, src, dest);
ao2_ref(type, +1);
if (idx < AST_VECTOR_SIZE(&dest->payload_mapping_tx)) {
ao2_t_cleanup(AST_VECTOR_GET(&dest->payload_mapping_tx, idx),
"cleaning up tx mapping vector element about to be replaced");
}
if (AST_VECTOR_REPLACE(&dest->payload_mapping_tx, idx, type)) {
ao2_ref(type, -1);
continue;
}
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
if (instance && instance->engine && instance->engine->payload_set) {
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_lock(instance);
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
instance->engine->payload_set(instance, idx, type->asterisk_format, type->format, type->rtp_code);
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_unlock(instance);
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
}
}
}
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
void ast_rtp_codecs_payloads_copy(struct ast_rtp_codecs *src, struct ast_rtp_codecs *dest, struct ast_rtp_instance *instance)
{
int idx;
struct ast_rtp_payload_type *type;
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
ast_rwlock_wrlock(&dest->codecs_lock);
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
/* Deadlock avoidance because of held write lock. */
while (ast_rwlock_tryrdlock(&src->codecs_lock)) {
ast_rwlock_unlock(&dest->codecs_lock);
sched_yield();
ast_rwlock_wrlock(&dest->codecs_lock);
}
/*
* This represents a completely new mapping of what the remote party is
* expecting for payloads, so we clear out the entire tx payload mapping
* vector and replace it.
*/
for (idx = 0; idx < AST_VECTOR_SIZE(&dest->payload_mapping_tx); ++idx) {
type = AST_VECTOR_GET(&dest->payload_mapping_tx, idx);
ao2_t_cleanup(type, "destroying ast_rtp_codec tx mapping");
AST_VECTOR_REPLACE(&dest->payload_mapping_tx, idx, NULL);
}
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
rtp_codecs_payloads_copy_rx(src, dest, instance);
rtp_codecs_payloads_copy_tx(src, dest, instance);
dest->framing = src->framing;
ast_rwlock_unlock(&src->codecs_lock);
ast_rwlock_unlock(&dest->codecs_lock);
}
void ast_rtp_codecs_payloads_xover(struct ast_rtp_codecs *src, struct ast_rtp_codecs *dest, struct ast_rtp_instance *instance)
{
int idx;
struct ast_rtp_payload_type *type;
ast_rwlock_wrlock(&dest->codecs_lock);
if (src != dest) {
/* Deadlock avoidance because of held write lock. */
while (ast_rwlock_tryrdlock(&src->codecs_lock)) {
ast_rwlock_unlock(&dest->codecs_lock);
sched_yield();
ast_rwlock_wrlock(&dest->codecs_lock);
}
}
/* Crossover copy payload type tx mapping to rx mapping. */
for (idx = 0; idx < AST_VECTOR_SIZE(&src->payload_mapping_tx); ++idx) {
type = AST_VECTOR_GET(&src->payload_mapping_tx, idx);
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
if (!type) {
continue;
}
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
/* All tx mapping elements should have the primary flag set. */
ast_assert(type->primary_mapping);
ast_debug(2, "Crossover copying tx to rx payload mapping %d (%p) from %p to %p\n",
idx, type, src, dest);
rtp_codecs_payload_replace_rx(dest, idx, type);
if (instance && instance->engine && instance->engine->payload_set) {
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_lock(instance);
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
instance->engine->payload_set(instance, idx, type->asterisk_format, type->format, type->rtp_code);
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_unlock(instance);
}
}
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
dest->framing = src->framing;
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
if (src != dest) {
ast_rwlock_unlock(&src->codecs_lock);
}
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
ast_rwlock_unlock(&dest->codecs_lock);
}
void ast_rtp_codecs_payloads_set_m_type(struct ast_rtp_codecs *codecs, struct ast_rtp_instance *instance, int payload)
{
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
struct ast_rtp_payload_type *new_type;
if (payload < 0 || payload >= AST_RTP_MAX_PT || payload > AST_RTP_PT_LAST_STATIC) {
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
return;
}
ast_rwlock_rdlock(&static_RTP_PT_lock);
new_type = ao2_bump(static_RTP_PT[payload]);
ast_rwlock_unlock(&static_RTP_PT_lock);
if (!new_type) {
ast_debug(1, "Don't have a default tx payload type %d format for m type on %p\n",
payload, codecs);
return;
}
ast_debug(1, "Setting tx payload type %d based on m type on %p\n",
payload, codecs);
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
ast_rwlock_wrlock(&codecs->codecs_lock);
if (!payload_mapping_tx_is_present(codecs, new_type)) {
if (payload < AST_VECTOR_SIZE(&codecs->payload_mapping_tx)) {
ao2_t_cleanup(AST_VECTOR_GET(&codecs->payload_mapping_tx, payload),
"cleaning up replaced tx payload type");
}
if (AST_VECTOR_REPLACE(&codecs->payload_mapping_tx, payload, new_type)) {
ao2_ref(new_type, -1);
} else if (instance && instance->engine && instance->engine->payload_set) {
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_lock(instance);
instance->engine->payload_set(instance, payload, new_type->asterisk_format, new_type->format, new_type->rtp_code);
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_unlock(instance);
}
} else {
ao2_ref(new_type, -1);
}
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
ast_rwlock_unlock(&codecs->codecs_lock);
}
int ast_rtp_codecs_payloads_set_rtpmap_type_rate(struct ast_rtp_codecs *codecs, struct ast_rtp_instance *instance, int pt,
char *mimetype, char *mimesubtype,
enum ast_rtp_options options,
unsigned int sample_rate)
{
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
unsigned int idx;
int found = 0;
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
if (pt < 0 || pt >= AST_RTP_MAX_PT) {
return -1; /* bogus payload type */
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
}
ast_rwlock_rdlock(&mime_types_lock);
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
ast_rwlock_wrlock(&codecs->codecs_lock);
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
for (idx = 0; idx < mime_types_len; ++idx) {
const struct ast_rtp_mime_type *t = &ast_rtp_mime_types[idx];
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
struct ast_rtp_payload_type *new_type;
if (strcasecmp(mimesubtype, t->subtype)) {
continue;
}
if (strcasecmp(mimetype, t->type)) {
continue;
}
/* if both sample rates have been supplied, and they don't match,
* then this not a match; if one has not been supplied, then the
* rates are not compared */
if (sample_rate && t->sample_rate &&
(sample_rate != t->sample_rate)) {
continue;
}
found = 1;
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
new_type = ast_rtp_engine_alloc_payload_type();
if (!new_type) {
continue;
}
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
new_type->asterisk_format = t->payload_type.asterisk_format;
new_type->rtp_code = t->payload_type.rtp_code;
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
new_type->payload = pt;
new_type->primary_mapping = 1;
if (t->payload_type.asterisk_format
&& ast_format_cmp(t->payload_type.format, ast_format_g726) == AST_FORMAT_CMP_EQUAL
&& (options & AST_RTP_OPT_G726_NONSTANDARD)) {
new_type->format = ast_format_g726_aal2;
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
} else {
new_type->format = t->payload_type.format;
}
if (new_type->format) {
/* SDP parsing automatically increases the reference count */
new_type->format = ast_format_parse_sdp_fmtp(new_type->format, "");
}
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
if (!payload_mapping_tx_is_present(codecs, new_type)) {
if (pt < AST_VECTOR_SIZE(&codecs->payload_mapping_tx)) {
ao2_t_cleanup(AST_VECTOR_GET(&codecs->payload_mapping_tx, pt),
"cleaning up replaced tx payload type");
}
if (AST_VECTOR_REPLACE(&codecs->payload_mapping_tx, pt, new_type)) {
ao2_ref(new_type, -1);
} else if (instance && instance->engine && instance->engine->payload_set) {
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_lock(instance);
instance->engine->payload_set(instance, pt, new_type->asterisk_format, new_type->format, new_type->rtp_code);
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_unlock(instance);
}
} else {
ao2_ref(new_type, -1);
}
break;
}
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
ast_rwlock_unlock(&codecs->codecs_lock);
ast_rwlock_unlock(&mime_types_lock);
return (found ? 0 : -2);
}
int ast_rtp_codecs_payloads_set_rtpmap_type(struct ast_rtp_codecs *codecs, struct ast_rtp_instance *instance, int payload, char *mimetype, char *mimesubtype, enum ast_rtp_options options)
{
return ast_rtp_codecs_payloads_set_rtpmap_type_rate(codecs, instance, payload, mimetype, mimesubtype, options, 0);
}
void ast_rtp_codecs_payloads_unset(struct ast_rtp_codecs *codecs, struct ast_rtp_instance *instance, int payload)
{
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
struct ast_rtp_payload_type *type;
if (payload < 0 || payload >= AST_RTP_MAX_PT) {
return;
}
ast_debug(2, "Unsetting payload %d on %p\n", payload, codecs);
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
ast_rwlock_wrlock(&codecs->codecs_lock);
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
if (payload < AST_VECTOR_SIZE(&codecs->payload_mapping_tx)) {
type = AST_VECTOR_GET(&codecs->payload_mapping_tx, payload);
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
ao2_cleanup(type);
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
AST_VECTOR_REPLACE(&codecs->payload_mapping_tx, payload, NULL);
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
}
if (instance && instance->engine && instance->engine->payload_set) {
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_lock(instance);
instance->engine->payload_set(instance, payload, 0, NULL, 0);
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_unlock(instance);
}
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
ast_rwlock_unlock(&codecs->codecs_lock);
}
enum ast_media_type ast_rtp_codecs_get_stream_type(struct ast_rtp_codecs *codecs)
{
enum ast_media_type stream_type = AST_MEDIA_TYPE_UNKNOWN;
int payload;
struct ast_rtp_payload_type *type;
ast_rwlock_rdlock(&codecs->codecs_lock);
for (payload = 0; payload < AST_VECTOR_SIZE(&codecs->payload_mapping_rx); ++payload) {
type = AST_VECTOR_GET(&codecs->payload_mapping_rx, payload);
if (type && type->asterisk_format) {
stream_type = ast_format_get_type(type->format);
break;
}
}
ast_rwlock_unlock(&codecs->codecs_lock);
return stream_type;
}
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
struct ast_rtp_payload_type *ast_rtp_codecs_get_payload(struct ast_rtp_codecs *codecs, int payload)
{
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
struct ast_rtp_payload_type *type = NULL;
if (payload < 0 || payload >= AST_RTP_MAX_PT) {
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
return NULL;
}
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
ast_rwlock_rdlock(&codecs->codecs_lock);
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
if (payload < AST_VECTOR_SIZE(&codecs->payload_mapping_rx)) {
type = AST_VECTOR_GET(&codecs->payload_mapping_rx, payload);
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
ao2_bump(type);
}
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
ast_rwlock_unlock(&codecs->codecs_lock);
if (!type && payload <= AST_RTP_PT_LAST_STATIC) {
ast_rwlock_rdlock(&static_RTP_PT_lock);
type = ao2_bump(static_RTP_PT[payload]);
ast_rwlock_unlock(&static_RTP_PT_lock);
}
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
return type;
}
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
int ast_rtp_codecs_payload_replace_format(struct ast_rtp_codecs *codecs, int payload, struct ast_format *format)
{
struct ast_rtp_payload_type *type;
if (payload < 0 || payload >= AST_RTP_MAX_PT || !format) {
return -1;
}
type = ast_rtp_engine_alloc_payload_type();
if (!type) {
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
return -1;
}
ao2_ref(format, +1);
type->format = format;
type->asterisk_format = 1;
type->payload = payload;
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
type->primary_mapping = 1;
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
ast_rwlock_wrlock(&codecs->codecs_lock);
if (!payload_mapping_tx_is_present(codecs, type)) {
if (payload < AST_VECTOR_SIZE(&codecs->payload_mapping_tx)) {
ao2_cleanup(AST_VECTOR_GET(&codecs->payload_mapping_tx, payload));
}
if (AST_VECTOR_REPLACE(&codecs->payload_mapping_tx, payload, type)) {
ao2_ref(type, -1);
}
} else {
ao2_ref(type, -1);
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
}
ast_rwlock_unlock(&codecs->codecs_lock);
return 0;
}
struct ast_format *ast_rtp_codecs_get_payload_format(struct ast_rtp_codecs *codecs, int payload)
{
struct ast_rtp_payload_type *type;
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
struct ast_format *format = NULL;
if (payload < 0 || payload >= AST_RTP_MAX_PT) {
return NULL;
}
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
ast_rwlock_rdlock(&codecs->codecs_lock);
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
if (payload < AST_VECTOR_SIZE(&codecs->payload_mapping_tx)) {
type = AST_VECTOR_GET(&codecs->payload_mapping_tx, payload);
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
if (type && type->asterisk_format) {
format = ao2_bump(type->format);
}
}
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
ast_rwlock_unlock(&codecs->codecs_lock);
return format;
}
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
void ast_rtp_codecs_set_framing(struct ast_rtp_codecs *codecs, unsigned int framing)
{
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
if (!framing) {
return;
}
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
ast_rwlock_wrlock(&codecs->codecs_lock);
codecs->framing = framing;
ast_rwlock_unlock(&codecs->codecs_lock);
}
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
unsigned int ast_rtp_codecs_get_framing(struct ast_rtp_codecs *codecs)
{
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
unsigned int framing;
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
ast_rwlock_rdlock(&codecs->codecs_lock);
framing = codecs->framing;
ast_rwlock_unlock(&codecs->codecs_lock);
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
return framing;
}
void ast_rtp_codecs_payload_formats(struct ast_rtp_codecs *codecs, struct ast_format_cap *astformats, int *nonastformats)
{
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
int idx;
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
ast_format_cap_remove_by_type(astformats, AST_MEDIA_TYPE_UNKNOWN);
*nonastformats = 0;
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
ast_rwlock_rdlock(&codecs->codecs_lock);
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
for (idx = 0; idx < AST_VECTOR_SIZE(&codecs->payload_mapping_tx); ++idx) {
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
struct ast_rtp_payload_type *type;
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
type = AST_VECTOR_GET(&codecs->payload_mapping_tx, idx);
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
if (!type) {
continue;
}
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
if (type->asterisk_format) {
ast_format_cap_append(astformats, type->format, 0);
} else {
*nonastformats |= type->rtp_code;
}
}
if (codecs->framing) {
ast_format_cap_set_framing(astformats, codecs->framing);
}
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
ast_rwlock_unlock(&codecs->codecs_lock);
}
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
/*!
* \internal
* \brief Find the static payload type mapping for the format.
* \since 14.0.0
*
* \param asterisk_format Non-zero if the given Asterisk format is present
* \param format Asterisk format to look for
* \param code The non-Asterisk format code to look for
*
* \note It is assumed that static_RTP_PT_lock is at least read locked before calling.
*
* \return Numerical payload type
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
* \retval -1 if not found.
*/
static int find_static_payload_type(int asterisk_format, const struct ast_format *format, int code)
{
int idx;
int payload = -1;
if (!asterisk_format) {
for (idx = 0; idx < AST_RTP_MAX_PT; ++idx) {
if (static_RTP_PT[idx]
&& !static_RTP_PT[idx]->asterisk_format
&& static_RTP_PT[idx]->rtp_code == code) {
payload = idx;
break;
}
}
} else if (format) {
for (idx = 0; idx < AST_RTP_MAX_PT; ++idx) {
if (static_RTP_PT[idx]
&& static_RTP_PT[idx]->asterisk_format
&& ast_format_cmp(format, static_RTP_PT[idx]->format)
!= AST_FORMAT_CMP_NOT_EQUAL) {
payload = idx;
break;
}
}
}
return payload;
}
/*!
* \internal
* \brief Find the first unused payload type in a given range
*
* \param codecs The codec structure to look in
* \param start Starting index
* \param end Ending index
* \param ignore Skip these payloads types
*
* \note The static_RTP_PT_lock object must be locked before calling
*
* \return Numerical payload type
* \retval -1 if not found.
*/
static int find_unused_payload_in_range(const struct ast_rtp_codecs *codecs,
int start, int end, struct ast_rtp_payload_type *ignore[])
{
int x;
for (x = start; x < end; ++x) {
struct ast_rtp_payload_type *type;
if (ignore[x]) {
continue;
} else if (!codecs || x >= AST_VECTOR_SIZE(&codecs->payload_mapping_rx)) {
return x;
}
type = AST_VECTOR_GET(&codecs->payload_mapping_rx, x);
if (!type) {
return x;
}
}
return -1;
}
/*!
* \internal
* \brief Find an unused payload type
*
* \param codecs Codecs structure to look in
*
* \note Both static_RTP_PT_lock and codecs (if given) must be at least
* read locked before calling.
*
* \return Numerical payload type
* \retval -1 if not found.
*/
static int find_unused_payload(const struct ast_rtp_codecs *codecs)
{
int res;
/* find next available dynamic payload slot */
res = find_unused_payload_in_range(
codecs, AST_RTP_PT_FIRST_DYNAMIC, AST_RTP_MAX_PT, static_RTP_PT);
if (res != -1) {
return res;
}
if (ast_option_rtpusedynamic) {
/*
* We're using default values for some dynamic types. So if an unused
* slot was not found try again, but this time ignore the default
* values declared for dynamic types (except for 101 and 121) .
*/
static struct ast_rtp_payload_type *ignore[AST_RTP_MAX_PT] = {0};
ignore[101] = static_RTP_PT[101];
ignore[121] = static_RTP_PT[121];
res = find_unused_payload_in_range(
codecs, AST_RTP_PT_FIRST_DYNAMIC, AST_RTP_MAX_PT, ignore);
if (res != -1) {
return res;
}
}
/* http://www.iana.org/assignments/rtp-parameters
* RFC 3551, Section 3: "[...] applications which need to define more
* than 32 dynamic payload types MAY bind codes below 96, in which case
* it is RECOMMENDED that unassigned payload type numbers be used
* first". Updated by RFC 5761, Section 4: "[...] values in the range
* 64-95 MUST NOT be used [to avoid conflicts with RTCP]". Summaries:
* https://tools.ietf.org/html/draft-roach-mmusic-unified-plan#section-3.2.1.2
* https://tools.ietf.org/html/draft-wu-avtcore-dynamic-pt-usage#section-3
*/
res = find_unused_payload_in_range(
codecs, MAX(ast_option_rtpptdynamic, AST_RTP_PT_LAST_STATIC + 1),
AST_RTP_PT_LAST_REASSIGN, static_RTP_PT);
if (res != -1) {
return res;
}
/* Yet, reusing mappings below AST_RTP_PT_LAST_STATIC (35) is not supported
* in Asterisk because when Compact Headers are activated, no rtpmap is
* send for those below 35. If you want to use 35 and below
* A) do not use Compact Headers,
* B) remove that code in chan_sip/res_pjsip, or
* C) add a flag that this RTP Payload Type got reassigned dynamically
* and requires a rtpmap even with Compact Headers enabled.
*/
res = find_unused_payload_in_range(
codecs, MAX(ast_option_rtpptdynamic, 20),
AST_RTP_PT_LAST_STATIC + 1, static_RTP_PT);
if (res != -1) {
return res;
}
return find_unused_payload_in_range(
codecs, MAX(ast_option_rtpptdynamic, 0),
20, static_RTP_PT);
}
/*!
* \internal
* \brief Find the oldest non-primary dynamic rx payload type.
* \since 14.0.0
*
* \param codecs Codecs structure to look in
*
* \note It is assumed that codecs is at least read locked before calling.
*
* \return Numerical payload type
* \retval -1 if not found.
*/
static int rtp_codecs_find_non_primary_dynamic_rx(struct ast_rtp_codecs *codecs)
{
struct ast_rtp_payload_type *type;
struct timeval oldest;
int idx;
int payload = -1;
idx = AST_RTP_PT_FIRST_DYNAMIC;
for (; idx < AST_VECTOR_SIZE(&codecs->payload_mapping_rx); ++idx) {
type = AST_VECTOR_GET(&codecs->payload_mapping_rx, idx);
if (type
&& !type->primary_mapping
&& (payload == -1
|| ast_tvdiff_ms(type->when_retired, oldest) < 0)) {
oldest = type->when_retired;
payload = idx;
}
}
return payload;
}
/*!
* \internal
* \brief Assign a payload type for the rx mapping.
* \since 14.0.0
*
* \param codecs Codecs structure to look in
* \param asterisk_format Non-zero if the given Asterisk format is present
* \param format Asterisk format to look for
* \param code The format to look for
* \param explicit Require the provided code to be explicitly used
*
* \note It is assumed that static_RTP_PT_lock is at least read locked before calling.
*
* \return Numerical payload type
* \retval -1 if could not assign.
*/
static int rtp_codecs_assign_payload_code_rx(struct ast_rtp_codecs *codecs, int asterisk_format, struct ast_format *format, int code, int explicit)
{
int payload = code;
struct ast_rtp_payload_type *new_type;
if (!explicit) {
payload = find_static_payload_type(asterisk_format, format, code);
if (payload < 0 && (!asterisk_format || !ast_option_rtpusedynamic)) {
return payload;
}
}
new_type = rtp_payload_type_alloc(format, payload, code, 1);
if (!new_type) {
return -1;
}
ast_rwlock_wrlock(&codecs->codecs_lock);
if (payload > -1 && (payload < AST_RTP_PT_FIRST_DYNAMIC
|| AST_VECTOR_SIZE(&codecs->payload_mapping_rx) <= payload
|| !AST_VECTOR_GET(&codecs->payload_mapping_rx, payload))) {
/*
* The payload type is a static assignment
* or our default dynamic position is available.
*/
rtp_codecs_payload_replace_rx(codecs, payload, new_type);
} else if (!explicit && (-1 < (payload = find_unused_payload(codecs))
|| -1 < (payload = rtp_codecs_find_non_primary_dynamic_rx(codecs)))) {
/*
* We found the first available empty dynamic position
* or we found a mapping that should no longer be
* actively used.
*/
new_type->payload = payload;
rtp_codecs_payload_replace_rx(codecs, payload, new_type);
} else if (explicit) {
/*
* They explicitly requested this payload number be used but it couldn't be
*/
payload = -1;
} else {
/*
* There are no empty or non-primary dynamic positions
* left. Sadness.
*
* I don't think this is really possible.
*/
ast_log(LOG_WARNING, "No dynamic RTP payload type values available "
"for %s - %d!\n", format ? ast_format_get_name(format) : "", code);
}
ast_rwlock_unlock(&codecs->codecs_lock);
ao2_ref(new_type, -1);
return payload;
}
int ast_rtp_codecs_payload_code(struct ast_rtp_codecs *codecs, int asterisk_format, struct ast_format *format, int code)
{
struct ast_rtp_payload_type *type;
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
int idx;
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
int payload = -1;
ast_rwlock_rdlock(&static_RTP_PT_lock);
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
if (!asterisk_format) {
ast_rwlock_rdlock(&codecs->codecs_lock);
for (idx = 0; idx < AST_VECTOR_SIZE(&codecs->payload_mapping_rx); ++idx) {
type = AST_VECTOR_GET(&codecs->payload_mapping_rx, idx);
if (!type) {
continue;
}
if (!type->asterisk_format
&& type->primary_mapping
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
&& type->rtp_code == code) {
payload = idx;
break;
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
}
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
}
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
ast_rwlock_unlock(&codecs->codecs_lock);
} else if (format) {
ast_rwlock_rdlock(&codecs->codecs_lock);
for (idx = 0; idx < AST_VECTOR_SIZE(&codecs->payload_mapping_rx); ++idx) {
type = AST_VECTOR_GET(&codecs->payload_mapping_rx, idx);
if (!type) {
continue;
}
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
if (type->asterisk_format
&& type->primary_mapping
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
&& ast_format_cmp(format, type->format) == AST_FORMAT_CMP_EQUAL) {
payload = idx;
break;
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
}
}
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
ast_rwlock_unlock(&codecs->codecs_lock);
}
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
if (payload < 0) {
payload = rtp_codecs_assign_payload_code_rx(codecs, asterisk_format, format,
code, 0);
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
}
ast_rwlock_unlock(&static_RTP_PT_lock);
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
return payload;
}
int ast_rtp_codecs_payload_set_rx(struct ast_rtp_codecs *codecs, int code, struct ast_format *format)
{
return rtp_codecs_assign_payload_code_rx(codecs, 1, format, code, 1);
}
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
int ast_rtp_codecs_payload_code_tx(struct ast_rtp_codecs *codecs, int asterisk_format, const struct ast_format *format, int code)
{
struct ast_rtp_payload_type *type;
int idx;
int payload = -1;
if (!asterisk_format) {
ast_rwlock_rdlock(&codecs->codecs_lock);
for (idx = 0; idx < AST_VECTOR_SIZE(&codecs->payload_mapping_tx); ++idx) {
type = AST_VECTOR_GET(&codecs->payload_mapping_tx, idx);
if (!type) {
continue;
}
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
if (!type->asterisk_format
&& type->rtp_code == code) {
payload = idx;
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
break;
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
}
}
ast_rwlock_unlock(&codecs->codecs_lock);
} else if (format) {
ast_rwlock_rdlock(&codecs->codecs_lock);
for (idx = 0; idx < AST_VECTOR_SIZE(&codecs->payload_mapping_tx); ++idx) {
type = AST_VECTOR_GET(&codecs->payload_mapping_tx, idx);
if (!type) {
continue;
}
if (type->asterisk_format
&& ast_format_cmp(format, type->format) == AST_FORMAT_CMP_EQUAL) {
payload = idx;
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
break;
}
}
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
ast_rwlock_unlock(&codecs->codecs_lock);
}
if (payload < 0) {
ast_rwlock_rdlock(&static_RTP_PT_lock);
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
payload = find_static_payload_type(asterisk_format, format, code);
ast_rwlock_unlock(&static_RTP_PT_lock);
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
}
return payload;
}
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
int ast_rtp_codecs_find_payload_code(struct ast_rtp_codecs *codecs, int payload)
{
struct ast_rtp_payload_type *type;
int res = -1;
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
ast_rwlock_rdlock(&codecs->codecs_lock);
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
if (payload < AST_VECTOR_SIZE(&codecs->payload_mapping_tx)) {
type = AST_VECTOR_GET(&codecs->payload_mapping_tx, payload);
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
if (type) {
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
res = payload;
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
}
}
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
ast_rwlock_unlock(&codecs->codecs_lock);
return res;
}
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
const char *ast_rtp_lookup_mime_subtype2(const int asterisk_format,
const struct ast_format *format, int code, enum ast_rtp_options options)
{
int i;
const char *res = "";
ast_rwlock_rdlock(&mime_types_lock);
for (i = 0; i < mime_types_len; i++) {
if (ast_rtp_mime_types[i].payload_type.asterisk_format && asterisk_format && format &&
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
(ast_format_cmp(format, ast_rtp_mime_types[i].payload_type.format) != AST_FORMAT_CMP_NOT_EQUAL)) {
if ((ast_format_cmp(format, ast_format_g726_aal2) == AST_FORMAT_CMP_EQUAL) &&
(options & AST_RTP_OPT_G726_NONSTANDARD)) {
res = "G726-32";
break;
} else {
res = ast_rtp_mime_types[i].subtype;
break;
}
} else if (!ast_rtp_mime_types[i].payload_type.asterisk_format && !asterisk_format &&
ast_rtp_mime_types[i].payload_type.rtp_code == code) {
res = ast_rtp_mime_types[i].subtype;
break;
}
}
ast_rwlock_unlock(&mime_types_lock);
return res;
}
unsigned int ast_rtp_lookup_sample_rate2(int asterisk_format,
const struct ast_format *format, int code)
{
unsigned int i;
unsigned int res = 0;
ast_rwlock_rdlock(&mime_types_lock);
for (i = 0; i < mime_types_len; ++i) {
if (ast_rtp_mime_types[i].payload_type.asterisk_format && asterisk_format && format &&
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
(ast_format_cmp(format, ast_rtp_mime_types[i].payload_type.format) != AST_FORMAT_CMP_NOT_EQUAL)) {
res = ast_rtp_mime_types[i].sample_rate;
break;
} else if (!ast_rtp_mime_types[i].payload_type.asterisk_format && !asterisk_format &&
ast_rtp_mime_types[i].payload_type.rtp_code == code) {
res = ast_rtp_mime_types[i].sample_rate;
break;
}
}
ast_rwlock_unlock(&mime_types_lock);
return res;
}
char *ast_rtp_lookup_mime_multiple2(struct ast_str *buf, struct ast_format_cap *ast_format_capability, int rtp_capability, const int asterisk_format, enum ast_rtp_options options)
{
int found = 0;
const char *name;
if (!buf) {
return NULL;
}
if (asterisk_format) {
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
int x;
struct ast_format *tmp_fmt;
for (x = 0; x < ast_format_cap_count(ast_format_capability); x++) {
tmp_fmt = ast_format_cap_get_format(ast_format_capability, x);
name = ast_rtp_lookup_mime_subtype2(asterisk_format, tmp_fmt, 0, options);
ao2_ref(tmp_fmt, -1);
ast_str_append(&buf, 0, "%s|", name);
found = 1;
}
} else {
int x;
ast_str_append(&buf, 0, "0x%x (", (unsigned int) rtp_capability);
for (x = 1; x <= AST_RTP_MAX; x <<= 1) {
if (rtp_capability & x) {
name = ast_rtp_lookup_mime_subtype2(asterisk_format, NULL, x, options);
ast_str_append(&buf, 0, "%s|", name);
found = 1;
}
}
}
ast_str_append(&buf, 0, "%s", found ? ")" : "nothing)");
return ast_str_buffer(buf);
}
int ast_rtp_instance_dtmf_begin(struct ast_rtp_instance *instance, char digit)
{
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
int res;
if (instance->engine->dtmf_begin) {
ao2_lock(instance);
res = instance->engine->dtmf_begin(instance, digit);
ao2_unlock(instance);
} else {
res = -1;
}
return res;
}
int ast_rtp_instance_dtmf_end(struct ast_rtp_instance *instance, char digit)
{
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
int res;
if (instance->engine->dtmf_end) {
ao2_lock(instance);
res = instance->engine->dtmf_end(instance, digit);
ao2_unlock(instance);
} else {
res = -1;
}
return res;
}
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
int ast_rtp_instance_dtmf_end_with_duration(struct ast_rtp_instance *instance, char digit, unsigned int duration)
{
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
int res;
if (instance->engine->dtmf_end_with_duration) {
ao2_lock(instance);
res = instance->engine->dtmf_end_with_duration(instance, digit, duration);
ao2_unlock(instance);
} else {
res = -1;
}
return res;
}
int ast_rtp_instance_dtmf_mode_set(struct ast_rtp_instance *instance, enum ast_rtp_dtmf_mode dtmf_mode)
{
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
int res;
if (instance->engine->dtmf_mode_set) {
ao2_lock(instance);
res = instance->engine->dtmf_mode_set(instance, dtmf_mode);
ao2_unlock(instance);
} else {
res = -1;
}
return res;
}
enum ast_rtp_dtmf_mode ast_rtp_instance_dtmf_mode_get(struct ast_rtp_instance *instance)
{
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
int res;
if (instance->engine->dtmf_mode_get) {
ao2_lock(instance);
res = instance->engine->dtmf_mode_get(instance);
ao2_unlock(instance);
} else {
res = 0;
}
return res;
}
void ast_rtp_instance_update_source(struct ast_rtp_instance *instance)
{
if (instance->engine->update_source) {
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_lock(instance);
instance->engine->update_source(instance);
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_unlock(instance);
}
}
void ast_rtp_instance_change_source(struct ast_rtp_instance *instance)
{
if (instance->engine->change_source) {
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_lock(instance);
instance->engine->change_source(instance);
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_unlock(instance);
}
}
int ast_rtp_instance_set_qos(struct ast_rtp_instance *instance, int tos, int cos, const char *desc)
{
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
int res;
if (instance->engine->qos) {
ao2_lock(instance);
res = instance->engine->qos(instance, tos, cos, desc);
ao2_unlock(instance);
} else {
res = -1;
}
return res;
}
void ast_rtp_instance_stop(struct ast_rtp_instance *instance)
{
if (instance->engine->stop) {
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_lock(instance);
instance->engine->stop(instance);
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_unlock(instance);
}
}
int ast_rtp_instance_fd(struct ast_rtp_instance *instance, int rtcp)
{
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
int res;
if (instance->engine->fd) {
ao2_lock(instance);
res = instance->engine->fd(instance, rtcp);
ao2_unlock(instance);
} else {
res = -1;
}
return res;
}
struct ast_rtp_glue *ast_rtp_instance_get_glue(const char *type)
{
struct ast_rtp_glue *glue = NULL;
AST_RWLIST_RDLOCK(&glues);
AST_RWLIST_TRAVERSE(&glues, glue, entry) {
if (!strcasecmp(glue->type, type)) {
break;
}
}
AST_RWLIST_UNLOCK(&glues);
return glue;
}
/*!
* \brief Conditionally unref an rtp instance
*/
static void unref_instance_cond(struct ast_rtp_instance **instance)
{
if (*instance) {
ao2_ref(*instance, -1);
*instance = NULL;
}
}
struct ast_rtp_instance *ast_rtp_instance_get_bridged(struct ast_rtp_instance *instance)
{
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
struct ast_rtp_instance *bridged;
ao2_lock(instance);
bridged = instance->bridged;
ao2_unlock(instance);
return bridged;
}
void ast_rtp_instance_set_bridged(struct ast_rtp_instance *instance, struct ast_rtp_instance *bridged)
{
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_lock(instance);
instance->bridged = bridged;
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_unlock(instance);
}
void ast_rtp_instance_early_bridge_make_compatible(struct ast_channel *c_dst, struct ast_channel *c_src)
{
struct ast_rtp_instance *instance_dst = NULL, *instance_src = NULL,
*vinstance_dst = NULL, *vinstance_src = NULL,
*tinstance_dst = NULL, *tinstance_src = NULL;
struct ast_rtp_glue *glue_dst, *glue_src;
enum ast_rtp_glue_result audio_glue_dst_res = AST_RTP_GLUE_RESULT_FORBID, video_glue_dst_res = AST_RTP_GLUE_RESULT_FORBID;
enum ast_rtp_glue_result audio_glue_src_res = AST_RTP_GLUE_RESULT_FORBID, video_glue_src_res = AST_RTP_GLUE_RESULT_FORBID;
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
struct ast_format_cap *cap_dst = ast_format_cap_alloc(AST_FORMAT_CAP_FLAG_DEFAULT);
struct ast_format_cap *cap_src = ast_format_cap_alloc(AST_FORMAT_CAP_FLAG_DEFAULT);
/* Lock both channels so we can look for the glue that binds them together */
ast_channel_lock_both(c_dst, c_src);
if (!cap_src || !cap_dst) {
goto done;
}
/* Grab glue that binds each channel to something using the RTP engine */
if (!(glue_dst = ast_rtp_instance_get_glue(ast_channel_tech(c_dst)->type)) || !(glue_src = ast_rtp_instance_get_glue(ast_channel_tech(c_src)->type))) {
ast_debug(1, "Can't find native functions for channel '%s'\n", glue_dst ? ast_channel_name(c_src) : ast_channel_name(c_dst));
goto done;
}
audio_glue_dst_res = glue_dst->get_rtp_info(c_dst, &instance_dst);
video_glue_dst_res = glue_dst->get_vrtp_info ? glue_dst->get_vrtp_info(c_dst, &vinstance_dst) : AST_RTP_GLUE_RESULT_FORBID;
audio_glue_src_res = glue_src->get_rtp_info(c_src, &instance_src);
video_glue_src_res = glue_src->get_vrtp_info ? glue_src->get_vrtp_info(c_src, &vinstance_src) : AST_RTP_GLUE_RESULT_FORBID;
/* If we are carrying video, and both sides are not going to remotely bridge... fail the native bridge */
if (video_glue_dst_res != AST_RTP_GLUE_RESULT_FORBID && (audio_glue_dst_res != AST_RTP_GLUE_RESULT_REMOTE || video_glue_dst_res != AST_RTP_GLUE_RESULT_REMOTE)) {
audio_glue_dst_res = AST_RTP_GLUE_RESULT_FORBID;
}
if (video_glue_src_res != AST_RTP_GLUE_RESULT_FORBID && (audio_glue_src_res != AST_RTP_GLUE_RESULT_REMOTE || video_glue_src_res != AST_RTP_GLUE_RESULT_REMOTE)) {
audio_glue_src_res = AST_RTP_GLUE_RESULT_FORBID;
}
if (audio_glue_dst_res == AST_RTP_GLUE_RESULT_REMOTE && (video_glue_dst_res == AST_RTP_GLUE_RESULT_FORBID || video_glue_dst_res == AST_RTP_GLUE_RESULT_REMOTE) && glue_dst->get_codec) {
glue_dst->get_codec(c_dst, cap_dst);
}
if (audio_glue_src_res == AST_RTP_GLUE_RESULT_REMOTE && (video_glue_src_res == AST_RTP_GLUE_RESULT_FORBID || video_glue_src_res == AST_RTP_GLUE_RESULT_REMOTE) && glue_src->get_codec) {
glue_src->get_codec(c_src, cap_src);
}
/* If any sort of bridge is forbidden just completely bail out and go back to generic bridging */
if (audio_glue_dst_res != AST_RTP_GLUE_RESULT_REMOTE || audio_glue_src_res != AST_RTP_GLUE_RESULT_REMOTE) {
goto done;
}
/* Make sure we have matching codecs */
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
if (!ast_format_cap_iscompatible(cap_dst, cap_src)) {
goto done;
}
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
ast_rtp_codecs_payloads_xover(&instance_src->codecs, &instance_dst->codecs, instance_dst);
if (vinstance_dst && vinstance_src) {
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
ast_rtp_codecs_payloads_xover(&vinstance_src->codecs, &vinstance_dst->codecs, vinstance_dst);
}
if (tinstance_dst && tinstance_src) {
rtp_engine.c: Initial split of payload types into rx and tx mappings. There are numerous problems with the current implementation of the RTP payload type mapping in Asterisk. It uses only one mapping structure to associate payload types to codecs. The single mapping is overkill if all of the payload type values are well known values. Dynamic payload type mappings do not work as well with the single mapping because RFC3264 allows each side of the link to negotiate different dynamic mappings for what they want to receive. Not only could you have the same codec mapped for sending and receiving on different payload types you could wind up with the same payload type mapped to different codecs for each direction. 1) An independent payload type mapping is needed for sending and receiving. 2) The receive mapping needs to keep track of previous mappings because of the slack to when negotiation happens and current packets in flight using the old mapping arrive. 3) The transmit mapping only needs to keep track of the current negotiated values since we are sending the packets and know when the switchover takes place. * Needed to create ast_rtp_codecs_payload_code_tx() and make some callers use the new function because ast_rtp_codecs_payload_code() was used for mappings in both directions. * Needed to create ast_rtp_codecs_payloads_xover() for cases where we need to pass preferred codec mappings to the peer channel for early media bridging or when we need to prefer the offered mapping that RFC3264 says we SHOULD use. * ast_rtp_codecs_payloads_xover() and ast_rtp_codecs_payload_code_tx() are the only new public functions created. All the others were only used for the tx or rx mapping direction so the function doxygen now reflects which direction the function operates. * chan_mgcp.c: Removed call to ast_rtp_codecs_payloads_clear() as doing that makes no sense when processing an incoming SDP. We would be wiping out any mappings that we set for the possible outgoing SDP we sent earlier. ASTERISK-25166 Reported by: Kevin Harwell ASTERISK-17410 Reported by: Boris Fox Change-Id: Iaf6c227bca68cb7c414cf2fd4108a8ac98bd45ac
2015-07-24 00:24:04 +00:00
ast_rtp_codecs_payloads_xover(&tinstance_src->codecs, &tinstance_dst->codecs, tinstance_dst);
}
if (glue_dst->update_peer(c_dst, instance_src, vinstance_src, tinstance_src, cap_src, 0)) {
ast_log(LOG_WARNING, "Channel '%s' failed to setup early bridge to '%s'\n",
ast_channel_name(c_dst), ast_channel_name(c_src));
} else {
ast_debug(1, "Seeded SDP of '%s' with that of '%s'\n",
ast_channel_name(c_dst), ast_channel_name(c_src));
}
done:
ast_channel_unlock(c_dst);
ast_channel_unlock(c_src);
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
ao2_cleanup(cap_dst);
ao2_cleanup(cap_src);
unref_instance_cond(&instance_dst);
unref_instance_cond(&instance_src);
unref_instance_cond(&vinstance_dst);
unref_instance_cond(&vinstance_src);
unref_instance_cond(&tinstance_dst);
unref_instance_cond(&tinstance_src);
}
int ast_rtp_instance_early_bridge(struct ast_channel *c0, struct ast_channel *c1)
{
struct ast_rtp_instance *instance0 = NULL, *instance1 = NULL,
*vinstance0 = NULL, *vinstance1 = NULL,
*tinstance0 = NULL, *tinstance1 = NULL;
struct ast_rtp_glue *glue0, *glue1;
enum ast_rtp_glue_result audio_glue0_res = AST_RTP_GLUE_RESULT_FORBID, video_glue0_res = AST_RTP_GLUE_RESULT_FORBID;
enum ast_rtp_glue_result audio_glue1_res = AST_RTP_GLUE_RESULT_FORBID, video_glue1_res = AST_RTP_GLUE_RESULT_FORBID;
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
struct ast_format_cap *cap0 = ast_format_cap_alloc(AST_FORMAT_CAP_FLAG_DEFAULT);
struct ast_format_cap *cap1 = ast_format_cap_alloc(AST_FORMAT_CAP_FLAG_DEFAULT);
/* If there is no second channel just immediately bail out, we are of no use in that scenario */
if (!c1 || !cap1 || !cap0) {
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
ao2_cleanup(cap0);
ao2_cleanup(cap1);
return -1;
}
/* Lock both channels so we can look for the glue that binds them together */
ast_channel_lock_both(c0, c1);
/* Grab glue that binds each channel to something using the RTP engine */
if (!(glue0 = ast_rtp_instance_get_glue(ast_channel_tech(c0)->type)) || !(glue1 = ast_rtp_instance_get_glue(ast_channel_tech(c1)->type))) {
Replace direct access to channel name with accessor functions There are many benefits to making the ast_channel an opaque handle, from increasing maintainability to presenting ways to kill masquerades. This patch kicks things off by taking things a field at a time, renaming the field to '__do_not_use_${fieldname}' and then writing setters/getters and converting the existing code to using them. When all fields are done, we can move ast_channel to a C file from channel.h and lop off the '__do_not_use_'. This patch sets up main/channel_interal_api.c to be the only file that actually accesses the ast_channel's fields directly. The intent would be for any API functions in channel.c to use the accessor functions. No more monkeying around with channel internals. We should use our own APIs. The interesting changes in this patch are the addition of channel_internal_api.c, the moving of the AST_DATA stuff from channel.c to channel_internal_api.c (note: the AST_DATA stuff will have to be reworked to use accessor functions when ast_channel is really opaque), and some re-working of the way channel iterators/callbacks are handled so as to avoid creating fake ast_channels on the stack to pass in matching data by directly accessing fields (since "name" is a stringfield and the fake channel doesn't init the stringfields, you can't use the ast_channel_name_set() function). I went with ast_channel_name(chan) for a getter, and ast_channel_name_set(chan, name) for a setter. The majority of the grunt-work for this change was done by writing a semantic patch using Coccinelle ( http://coccinelle.lip6.fr/ ). Review: https://reviewboard.asterisk.org/r/1655/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@350223 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-01-09 22:15:50 +00:00
ast_log(LOG_WARNING, "Can't find native functions for channel '%s'\n", glue0 ? ast_channel_name(c1) : ast_channel_name(c0));
goto done;
}
audio_glue0_res = glue0->get_rtp_info(c0, &instance0);
video_glue0_res = glue0->get_vrtp_info ? glue0->get_vrtp_info(c0, &vinstance0) : AST_RTP_GLUE_RESULT_FORBID;
audio_glue1_res = glue1->get_rtp_info(c1, &instance1);
video_glue1_res = glue1->get_vrtp_info ? glue1->get_vrtp_info(c1, &vinstance1) : AST_RTP_GLUE_RESULT_FORBID;
/* If we are carrying video, and both sides are not going to remotely bridge... fail the native bridge */
if (video_glue0_res != AST_RTP_GLUE_RESULT_FORBID && (audio_glue0_res != AST_RTP_GLUE_RESULT_REMOTE || video_glue0_res != AST_RTP_GLUE_RESULT_REMOTE)) {
audio_glue0_res = AST_RTP_GLUE_RESULT_FORBID;
}
if (video_glue1_res != AST_RTP_GLUE_RESULT_FORBID && (audio_glue1_res != AST_RTP_GLUE_RESULT_REMOTE || video_glue1_res != AST_RTP_GLUE_RESULT_REMOTE)) {
audio_glue1_res = AST_RTP_GLUE_RESULT_FORBID;
}
if (audio_glue0_res == AST_RTP_GLUE_RESULT_REMOTE && (video_glue0_res == AST_RTP_GLUE_RESULT_FORBID || video_glue0_res == AST_RTP_GLUE_RESULT_REMOTE) && glue0->get_codec) {
glue0->get_codec(c0, cap0);
}
if (audio_glue1_res == AST_RTP_GLUE_RESULT_REMOTE && (video_glue1_res == AST_RTP_GLUE_RESULT_FORBID || video_glue1_res == AST_RTP_GLUE_RESULT_REMOTE) && glue1->get_codec) {
glue1->get_codec(c1, cap1);
}
/* If any sort of bridge is forbidden just completely bail out and go back to generic bridging */
if (audio_glue0_res != AST_RTP_GLUE_RESULT_REMOTE || audio_glue1_res != AST_RTP_GLUE_RESULT_REMOTE) {
goto done;
}
/* Make sure we have matching codecs */
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
if (!ast_format_cap_iscompatible(cap0, cap1)) {
goto done;
}
/* Bridge media early */
if (glue0->update_peer(c0, instance1, vinstance1, tinstance1, cap1, 0)) {
Replace direct access to channel name with accessor functions There are many benefits to making the ast_channel an opaque handle, from increasing maintainability to presenting ways to kill masquerades. This patch kicks things off by taking things a field at a time, renaming the field to '__do_not_use_${fieldname}' and then writing setters/getters and converting the existing code to using them. When all fields are done, we can move ast_channel to a C file from channel.h and lop off the '__do_not_use_'. This patch sets up main/channel_interal_api.c to be the only file that actually accesses the ast_channel's fields directly. The intent would be for any API functions in channel.c to use the accessor functions. No more monkeying around with channel internals. We should use our own APIs. The interesting changes in this patch are the addition of channel_internal_api.c, the moving of the AST_DATA stuff from channel.c to channel_internal_api.c (note: the AST_DATA stuff will have to be reworked to use accessor functions when ast_channel is really opaque), and some re-working of the way channel iterators/callbacks are handled so as to avoid creating fake ast_channels on the stack to pass in matching data by directly accessing fields (since "name" is a stringfield and the fake channel doesn't init the stringfields, you can't use the ast_channel_name_set() function). I went with ast_channel_name(chan) for a getter, and ast_channel_name_set(chan, name) for a setter. The majority of the grunt-work for this change was done by writing a semantic patch using Coccinelle ( http://coccinelle.lip6.fr/ ). Review: https://reviewboard.asterisk.org/r/1655/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@350223 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-01-09 22:15:50 +00:00
ast_log(LOG_WARNING, "Channel '%s' failed to setup early bridge to '%s'\n", ast_channel_name(c0), c1 ? ast_channel_name(c1) : "<unspecified>");
}
done:
ast_channel_unlock(c0);
ast_channel_unlock(c1);
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
ao2_cleanup(cap0);
ao2_cleanup(cap1);
unref_instance_cond(&instance0);
unref_instance_cond(&instance1);
unref_instance_cond(&vinstance0);
unref_instance_cond(&vinstance1);
unref_instance_cond(&tinstance0);
unref_instance_cond(&tinstance1);
ast_debug(1, "Setting early bridge SDP of '%s' with that of '%s'\n", ast_channel_name(c0), c1 ? ast_channel_name(c1) : "<unspecified>");
return 0;
}
int ast_rtp_red_init(struct ast_rtp_instance *instance, int buffer_time, int *payloads, int generations)
{
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
int res;
if (instance->engine->red_init) {
ao2_lock(instance);
res = instance->engine->red_init(instance, buffer_time, payloads, generations);
ao2_unlock(instance);
} else {
res = -1;
}
return res;
}
int ast_rtp_red_buffer(struct ast_rtp_instance *instance, struct ast_frame *frame)
{
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
int res;
if (instance->engine->red_buffer) {
ao2_lock(instance);
res = instance->engine->red_buffer(instance, frame);
ao2_unlock(instance);
} else {
res = -1;
}
return res;
}
int ast_rtp_instance_get_stats(struct ast_rtp_instance *instance, struct ast_rtp_instance_stats *stats, enum ast_rtp_instance_stat stat)
{
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
int res;
if (instance->engine->get_stat) {
ao2_lock(instance);
res = instance->engine->get_stat(instance, stats, stat);
ao2_unlock(instance);
} else {
res = -1;
}
return res;
}
char *ast_rtp_instance_get_quality(struct ast_rtp_instance *instance, enum ast_rtp_instance_stat_field field, char *buf, size_t size)
{
struct ast_rtp_instance_stats stats = { 0, };
enum ast_rtp_instance_stat stat;
/* Determine what statistics we will need to retrieve based on field passed in */
if (field == AST_RTP_INSTANCE_STAT_FIELD_QUALITY) {
stat = AST_RTP_INSTANCE_STAT_ALL;
} else if (field == AST_RTP_INSTANCE_STAT_FIELD_QUALITY_JITTER) {
stat = AST_RTP_INSTANCE_STAT_COMBINED_JITTER;
} else if (field == AST_RTP_INSTANCE_STAT_FIELD_QUALITY_LOSS) {
stat = AST_RTP_INSTANCE_STAT_COMBINED_LOSS;
} else if (field == AST_RTP_INSTANCE_STAT_FIELD_QUALITY_RTT) {
stat = AST_RTP_INSTANCE_STAT_COMBINED_RTT;
} else {
return NULL;
}
/* Attempt to actually retrieve the statistics we need to generate the quality string */
if (ast_rtp_instance_get_stats(instance, &stats, stat)) {
return NULL;
}
/* Now actually fill the buffer with the good information */
if (field == AST_RTP_INSTANCE_STAT_FIELD_QUALITY) {
snprintf(buf, size, "ssrc=%u;themssrc=%u;lp=%u;rxjitter=%f;rxcount=%u;txjitter=%f;txcount=%u;rlp=%u;rtt=%f",
stats.local_ssrc, stats.remote_ssrc, stats.rxploss, stats.rxjitter, stats.rxcount, stats.txjitter, stats.txcount, stats.txploss, stats.rtt);
} else if (field == AST_RTP_INSTANCE_STAT_FIELD_QUALITY_JITTER) {
snprintf(buf, size, "minrxjitter=%f;maxrxjitter=%f;avgrxjitter=%f;stdevrxjitter=%f;reported_minjitter=%f;reported_maxjitter=%f;reported_avgjitter=%f;reported_stdevjitter=%f;",
stats.local_minjitter, stats.local_maxjitter, stats.local_normdevjitter, sqrt(stats.local_stdevjitter), stats.remote_minjitter, stats.remote_maxjitter, stats.remote_normdevjitter, sqrt(stats.remote_stdevjitter));
} else if (field == AST_RTP_INSTANCE_STAT_FIELD_QUALITY_LOSS) {
snprintf(buf, size, "minrxlost=%f;maxrxlost=%f;avgrxlost=%f;stdevrxlost=%f;reported_minlost=%f;reported_maxlost=%f;reported_avglost=%f;reported_stdevlost=%f;",
stats.local_minrxploss, stats.local_maxrxploss, stats.local_normdevrxploss, sqrt(stats.local_stdevrxploss), stats.remote_minrxploss, stats.remote_maxrxploss, stats.remote_normdevrxploss, sqrt(stats.remote_stdevrxploss));
} else if (field == AST_RTP_INSTANCE_STAT_FIELD_QUALITY_RTT) {
snprintf(buf, size, "minrtt=%f;maxrtt=%f;avgrtt=%f;stdevrtt=%f;", stats.minrtt, stats.maxrtt, stats.normdevrtt, stats.stdevrtt);
}
return buf;
}
void ast_rtp_instance_set_stats_vars(struct ast_channel *chan, struct ast_rtp_instance *instance)
{
char quality_buf[AST_MAX_USER_FIELD];
char *quality;
struct ast_channel *bridge;
bridge = ast_channel_bridge_peer(chan);
if (bridge) {
ast_channel_lock_both(chan, bridge);
ast_channel_stage_snapshot(bridge);
} else {
ast_channel_lock(chan);
}
ast_channel_stage_snapshot(chan);
quality = ast_rtp_instance_get_quality(instance, AST_RTP_INSTANCE_STAT_FIELD_QUALITY,
quality_buf, sizeof(quality_buf));
if (quality) {
pbx_builtin_setvar_helper(chan, "RTPAUDIOQOS", quality);
if (bridge) {
pbx_builtin_setvar_helper(bridge, "RTPAUDIOQOSBRIDGED", quality);
}
}
quality = ast_rtp_instance_get_quality(instance,
AST_RTP_INSTANCE_STAT_FIELD_QUALITY_JITTER, quality_buf, sizeof(quality_buf));
if (quality) {
pbx_builtin_setvar_helper(chan, "RTPAUDIOQOSJITTER", quality);
if (bridge) {
pbx_builtin_setvar_helper(bridge, "RTPAUDIOQOSJITTERBRIDGED", quality);
}
}
quality = ast_rtp_instance_get_quality(instance,
AST_RTP_INSTANCE_STAT_FIELD_QUALITY_LOSS, quality_buf, sizeof(quality_buf));
if (quality) {
pbx_builtin_setvar_helper(chan, "RTPAUDIOQOSLOSS", quality);
if (bridge) {
pbx_builtin_setvar_helper(bridge, "RTPAUDIOQOSLOSSBRIDGED", quality);
}
}
quality = ast_rtp_instance_get_quality(instance,
AST_RTP_INSTANCE_STAT_FIELD_QUALITY_RTT, quality_buf, sizeof(quality_buf));
if (quality) {
pbx_builtin_setvar_helper(chan, "RTPAUDIOQOSRTT", quality);
if (bridge) {
pbx_builtin_setvar_helper(bridge, "RTPAUDIOQOSRTTBRIDGED", quality);
}
}
ast_channel_stage_snapshot_done(chan);
ast_channel_unlock(chan);
if (bridge) {
ast_channel_stage_snapshot_done(bridge);
ast_channel_unlock(bridge);
ast_channel_unref(bridge);
}
}
int ast_rtp_instance_set_read_format(struct ast_rtp_instance *instance, struct ast_format *format)
{
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
int res;
if (instance->engine->set_read_format) {
ao2_lock(instance);
res = instance->engine->set_read_format(instance, format);
ao2_unlock(instance);
} else {
res = -1;
}
return res;
}
int ast_rtp_instance_set_write_format(struct ast_rtp_instance *instance, struct ast_format *format)
{
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
int res;
if (instance->engine->set_read_format) {
ao2_lock(instance);
res = instance->engine->set_write_format(instance, format);
ao2_unlock(instance);
} else {
res = -1;
}
return res;
}
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
/* XXX Nothing calls this */
int ast_rtp_instance_make_compatible(struct ast_channel *chan, struct ast_rtp_instance *instance, struct ast_channel *peer)
{
struct ast_rtp_glue *glue;
struct ast_rtp_instance *peer_instance = NULL;
int res = -1;
if (!instance->engine->make_compatible) {
return -1;
}
ast_channel_lock(peer);
if (!(glue = ast_rtp_instance_get_glue(ast_channel_tech(peer)->type))) {
ast_channel_unlock(peer);
return -1;
}
glue->get_rtp_info(peer, &peer_instance);
if (!peer_instance) {
ast_log(LOG_ERROR, "Unable to get_rtp_info for peer type %s\n", glue->type);
ast_channel_unlock(peer);
return -1;
}
if (peer_instance->engine != instance->engine) {
ast_log(LOG_ERROR, "Peer engine mismatch for type %s\n", glue->type);
ast_channel_unlock(peer);
ao2_ref(peer_instance, -1);
return -1;
}
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
/*
* XXX Good thing nothing calls this function because we would need
* deadlock avoidance to get the two instance locks.
*/
res = instance->engine->make_compatible(chan, instance, peer, peer_instance);
ast_channel_unlock(peer);
ao2_ref(peer_instance, -1);
peer_instance = NULL;
return res;
}
void ast_rtp_instance_available_formats(struct ast_rtp_instance *instance, struct ast_format_cap *to_endpoint, struct ast_format_cap *to_asterisk, struct ast_format_cap *result)
{
if (instance->engine->available_formats) {
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_lock(instance);
instance->engine->available_formats(instance, to_endpoint, to_asterisk, result);
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_unlock(instance);
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
if (ast_format_cap_count(result)) {
return;
}
}
ast_translate_available_formats(to_endpoint, to_asterisk, result);
}
int ast_rtp_instance_activate(struct ast_rtp_instance *instance)
{
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
int res;
if (instance->engine->activate) {
ao2_lock(instance);
res = instance->engine->activate(instance);
ao2_unlock(instance);
} else {
res = 0;
}
return res;
}
void ast_rtp_instance_stun_request(struct ast_rtp_instance *instance,
struct ast_sockaddr *suggestion,
const char *username)
{
if (instance->engine->stun_request) {
instance->engine->stun_request(instance, suggestion, username);
}
}
void ast_rtp_instance_set_timeout(struct ast_rtp_instance *instance, int timeout)
{
instance->timeout = timeout;
}
void ast_rtp_instance_set_hold_timeout(struct ast_rtp_instance *instance, int timeout)
{
instance->holdtimeout = timeout;
}
void ast_rtp_instance_set_keepalive(struct ast_rtp_instance *instance, int interval)
{
instance->keepalive = interval;
}
int ast_rtp_instance_get_timeout(struct ast_rtp_instance *instance)
{
return instance->timeout;
}
int ast_rtp_instance_get_hold_timeout(struct ast_rtp_instance *instance)
{
return instance->holdtimeout;
}
int ast_rtp_instance_get_keepalive(struct ast_rtp_instance *instance)
{
return instance->keepalive;
}
struct ast_rtp_engine *ast_rtp_instance_get_engine(struct ast_rtp_instance *instance)
{
return instance->engine;
}
struct ast_rtp_glue *ast_rtp_instance_get_active_glue(struct ast_rtp_instance *instance)
{
return instance->glue;
}
int ast_rtp_engine_register_srtp(struct ast_srtp_res *srtp_res, struct ast_srtp_policy_res *policy_res)
{
if (res_srtp || res_srtp_policy) {
return -1;
}
if (!srtp_res || !policy_res) {
return -1;
}
res_srtp = srtp_res;
res_srtp_policy = policy_res;
return 0;
}
void ast_rtp_engine_unregister_srtp(void)
{
res_srtp = NULL;
res_srtp_policy = NULL;
}
int ast_rtp_engine_srtp_is_registered(void)
{
return res_srtp && res_srtp_policy;
}
int ast_rtp_instance_add_srtp_policy(struct ast_rtp_instance *instance, struct ast_srtp_policy *remote_policy, struct ast_srtp_policy *local_policy, int rtcp)
{
int res = 0;
struct ast_srtp **srtp;
if (!res_srtp) {
return -1;
}
srtp: Fix possible race condition, and add NULL checks Somehow it's possible for the srtp session object to be NULL even though the Asterisk srtp object itself is valid. When this happened it would cause a crash down in the srtp code when attempting to protect or unprotect data. After looking at the code there is at least one spot that makes this situation possible. If Asterisk fails to unprotect the data, and after several retries it still can't then the srtp->session gets freed, and set to NULL while still leaving the Asterisk srtp object around. However, according to the original issue reporter this does not appear to be their situation since they found no errors logged stating the above happened (which Asterisk does for that situation). An issue was found however, where a possible race condition could occur between the pjsip incoming negotiation, and the receiving of RTP packets. Both places could attempt to create/setup srtp for the same rtp instance at the same time. This potentially could be the cause of the problem as well. Given the above this patch adds locking around srtp setup for a given rtp, or rtcp instance. NULL checks for the session have also been added within the protect and unprotect functions as a precaution. These checks should at least stop Asterisk from crashing if it gets in this situation again. This patch also fixes one other issue noticed during investigation. When doing a replace the old object was freed before creating the replacement. If the new replacement object failed to create then the rtp/rtcp instance would now point to freed srtp data which could potentially cause a crash as well when the next attempt to reference it was made. This is now fixed so the old srtp object is kept upon replacement failure. Lastly, more logging has been added to help diagnose future issues. ASTERISK-28472 Change-Id: I240e11cbb1e9ea8083d59d50db069891228fe5cc
2019-08-07 22:54:34 +00:00
ao2_lock(instance);
srtp = rtcp ? &instance->rtcp_srtp : &instance->srtp;
if (!*srtp) {
res = res_srtp->create(srtp, instance, remote_policy);
} else if (remote_policy) {
res = res_srtp->replace(srtp, instance, remote_policy);
}
if (!res) {
res = res_srtp->add_stream(*srtp, local_policy);
}
srtp: Fix possible race condition, and add NULL checks Somehow it's possible for the srtp session object to be NULL even though the Asterisk srtp object itself is valid. When this happened it would cause a crash down in the srtp code when attempting to protect or unprotect data. After looking at the code there is at least one spot that makes this situation possible. If Asterisk fails to unprotect the data, and after several retries it still can't then the srtp->session gets freed, and set to NULL while still leaving the Asterisk srtp object around. However, according to the original issue reporter this does not appear to be their situation since they found no errors logged stating the above happened (which Asterisk does for that situation). An issue was found however, where a possible race condition could occur between the pjsip incoming negotiation, and the receiving of RTP packets. Both places could attempt to create/setup srtp for the same rtp instance at the same time. This potentially could be the cause of the problem as well. Given the above this patch adds locking around srtp setup for a given rtp, or rtcp instance. NULL checks for the session have also been added within the protect and unprotect functions as a precaution. These checks should at least stop Asterisk from crashing if it gets in this situation again. This patch also fixes one other issue noticed during investigation. When doing a replace the old object was freed before creating the replacement. If the new replacement object failed to create then the rtp/rtcp instance would now point to freed srtp data which could potentially cause a crash as well when the next attempt to reference it was made. This is now fixed so the old srtp object is kept upon replacement failure. Lastly, more logging has been added to help diagnose future issues. ASTERISK-28472 Change-Id: I240e11cbb1e9ea8083d59d50db069891228fe5cc
2019-08-07 22:54:34 +00:00
ao2_unlock(instance);
return res;
}
struct ast_srtp *ast_rtp_instance_get_srtp(struct ast_rtp_instance *instance, int rtcp)
{
if (rtcp && instance->rtcp_srtp) {
return instance->rtcp_srtp;
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
} else {
return instance->srtp;
}
}
int ast_rtp_instance_sendcng(struct ast_rtp_instance *instance, int level)
{
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
int res;
if (instance->engine->sendcng) {
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
ao2_lock(instance);
res = instance->engine->sendcng(instance, level);
ao2_unlock(instance);
} else {
res = -1;
}
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
return res;
}
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
static void rtp_ice_wrap_set_authentication(struct ast_rtp_instance *instance, const char *ufrag, const char *password)
{
ao2_lock(instance);
instance->engine->ice->set_authentication(instance, ufrag, password);
ao2_unlock(instance);
}
static void rtp_ice_wrap_add_remote_candidate(struct ast_rtp_instance *instance, const struct ast_rtp_engine_ice_candidate *candidate)
{
ao2_lock(instance);
instance->engine->ice->add_remote_candidate(instance, candidate);
ao2_unlock(instance);
}
static void rtp_ice_wrap_start(struct ast_rtp_instance *instance)
{
ao2_lock(instance);
instance->engine->ice->start(instance);
ao2_unlock(instance);
}
static void rtp_ice_wrap_stop(struct ast_rtp_instance *instance)
{
ao2_lock(instance);
instance->engine->ice->stop(instance);
ao2_unlock(instance);
}
static const char *rtp_ice_wrap_get_ufrag(struct ast_rtp_instance *instance)
{
const char *ufrag;
ao2_lock(instance);
ufrag = instance->engine->ice->get_ufrag(instance);
ao2_unlock(instance);
return ufrag;
}
static const char *rtp_ice_wrap_get_password(struct ast_rtp_instance *instance)
{
const char *password;
ao2_lock(instance);
password = instance->engine->ice->get_password(instance);
ao2_unlock(instance);
return password;
}
static struct ao2_container *rtp_ice_wrap_get_local_candidates(struct ast_rtp_instance *instance)
{
struct ao2_container *local_candidates;
ao2_lock(instance);
local_candidates = instance->engine->ice->get_local_candidates(instance);
ao2_unlock(instance);
return local_candidates;
}
static void rtp_ice_wrap_ice_lite(struct ast_rtp_instance *instance)
{
ao2_lock(instance);
instance->engine->ice->ice_lite(instance);
ao2_unlock(instance);
}
static void rtp_ice_wrap_set_role(struct ast_rtp_instance *instance,
enum ast_rtp_ice_role role)
{
ao2_lock(instance);
instance->engine->ice->set_role(instance, role);
ao2_unlock(instance);
}
static void rtp_ice_wrap_turn_request(struct ast_rtp_instance *instance,
enum ast_rtp_ice_component_type component, enum ast_transport transport,
const char *server, unsigned int port, const char *username, const char *password)
{
ao2_lock(instance);
instance->engine->ice->turn_request(instance, component, transport, server, port,
username, password);
ao2_unlock(instance);
}
static void rtp_ice_wrap_change_components(struct ast_rtp_instance *instance,
int num_components)
{
ao2_lock(instance);
instance->engine->ice->change_components(instance, num_components);
ao2_unlock(instance);
}
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
static struct ast_rtp_engine_ice rtp_ice_wrappers = {
.set_authentication = rtp_ice_wrap_set_authentication,
.add_remote_candidate = rtp_ice_wrap_add_remote_candidate,
.start = rtp_ice_wrap_start,
.stop = rtp_ice_wrap_stop,
.get_ufrag = rtp_ice_wrap_get_ufrag,
.get_password = rtp_ice_wrap_get_password,
.get_local_candidates = rtp_ice_wrap_get_local_candidates,
.ice_lite = rtp_ice_wrap_ice_lite,
.set_role = rtp_ice_wrap_set_role,
.turn_request = rtp_ice_wrap_turn_request,
.change_components = rtp_ice_wrap_change_components,
};
struct ast_rtp_engine_ice *ast_rtp_instance_get_ice(struct ast_rtp_instance *instance)
{
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
if (instance->engine->ice) {
return &rtp_ice_wrappers;
}
/* ICE not available */
return NULL;
}
#ifdef TEST_FRAMEWORK
struct ast_rtp_engine_test *ast_rtp_instance_get_test(struct ast_rtp_instance *instance)
{
return instance->engine->test;
}
#endif
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
static int rtp_dtls_wrap_set_configuration(struct ast_rtp_instance *instance,
const struct ast_rtp_dtls_cfg *dtls_cfg)
{
int set_configuration;
ao2_lock(instance);
set_configuration = instance->engine->dtls->set_configuration(instance, dtls_cfg);
ao2_unlock(instance);
return set_configuration;
}
static int rtp_dtls_wrap_active(struct ast_rtp_instance *instance)
{
int active;
ao2_lock(instance);
active = instance->engine->dtls->active(instance);
ao2_unlock(instance);
return active;
}
static void rtp_dtls_wrap_stop(struct ast_rtp_instance *instance)
{
ao2_lock(instance);
instance->engine->dtls->stop(instance);
ao2_unlock(instance);
}
static void rtp_dtls_wrap_reset(struct ast_rtp_instance *instance)
{
ao2_lock(instance);
instance->engine->dtls->reset(instance);
ao2_unlock(instance);
}
static enum ast_rtp_dtls_connection rtp_dtls_wrap_get_connection(struct ast_rtp_instance *instance)
{
enum ast_rtp_dtls_connection get_connection;
ao2_lock(instance);
get_connection = instance->engine->dtls->get_connection(instance);
ao2_unlock(instance);
return get_connection;
}
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
static enum ast_rtp_dtls_setup rtp_dtls_wrap_get_setup(struct ast_rtp_instance *instance)
{
enum ast_rtp_dtls_setup get_setup;
ao2_lock(instance);
get_setup = instance->engine->dtls->get_setup(instance);
ao2_unlock(instance);
return get_setup;
}
static void rtp_dtls_wrap_set_setup(struct ast_rtp_instance *instance,
enum ast_rtp_dtls_setup setup)
{
ao2_lock(instance);
instance->engine->dtls->set_setup(instance, setup);
ao2_unlock(instance);
}
static void rtp_dtls_wrap_set_fingerprint(struct ast_rtp_instance *instance,
enum ast_rtp_dtls_hash hash, const char *fingerprint)
{
ao2_lock(instance);
instance->engine->dtls->set_fingerprint(instance, hash, fingerprint);
ao2_unlock(instance);
}
static enum ast_rtp_dtls_hash rtp_dtls_wrap_get_fingerprint_hash(struct ast_rtp_instance *instance)
{
enum ast_rtp_dtls_hash get_fingerprint_hash;
ao2_lock(instance);
get_fingerprint_hash = instance->engine->dtls->get_fingerprint_hash(instance);
ao2_unlock(instance);
return get_fingerprint_hash;
}
static const char *rtp_dtls_wrap_get_fingerprint(struct ast_rtp_instance *instance)
{
const char *get_fingerprint;
ao2_lock(instance);
get_fingerprint = instance->engine->dtls->get_fingerprint(instance);
ao2_unlock(instance);
return get_fingerprint;
}
static struct ast_rtp_engine_dtls rtp_dtls_wrappers = {
.set_configuration = rtp_dtls_wrap_set_configuration,
.active = rtp_dtls_wrap_active,
.stop = rtp_dtls_wrap_stop,
.reset = rtp_dtls_wrap_reset,
.get_connection = rtp_dtls_wrap_get_connection,
.get_setup = rtp_dtls_wrap_get_setup,
.set_setup = rtp_dtls_wrap_set_setup,
.set_fingerprint = rtp_dtls_wrap_set_fingerprint,
.get_fingerprint_hash = rtp_dtls_wrap_get_fingerprint_hash,
.get_fingerprint = rtp_dtls_wrap_get_fingerprint,
};
struct ast_rtp_engine_dtls *ast_rtp_instance_get_dtls(struct ast_rtp_instance *instance)
{
rtp_engine/res_rtp_asterisk: Fix RTP struct reentrancy crashes. The struct ast_rtp_instance has historically been indirectly protected from reentrancy issues by the channel lock because early channel drivers held the lock for really long times. Holding the channel lock for such a long time has caused many deadlock problems in the past. Along comes chan_pjsip/res_pjsip which doesn't necessarily hold the channel lock because sometimes there may not be an associated channel created yet or the channel pointer isn't available. In the case of ASTERISK-26835 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up changing the rtp->rtcp->local_addr_str string and interfering with each other. The classic reentrancy problem resulted in a crash. In the case of ASTERISK-26853 a pjsip serializer thread was processing a message's SDP body while another thread was reading a RTP packet from the socket. Both threads wound up processing ICE candidates in PJPROJECT and interfering with each other. The classic reentrancy problem resulted in a crash. * rtp_engine.c: Make the ast_rtp_instance_xxx() calls lock the RTP instance struct. * rtp_engine.c: Make ICE and DTLS wrapper functions to lock the RTP instance struct for the API call. * res_rtp_asterisk.c: Lock the RTP instance to prevent a reentrancy problem with rtp->rtcp->local_addr_str in the scheduler thread running ast_rtcp_write(). * res_rtp_asterisk.c: Avoid deadlock when local RTP bridging in bridge_p2p_rtp_write() because there are two RTP instance structs involved. * res_rtp_asterisk.c: Avoid deadlock when trying to stop scheduler callbacks. We cannot hold the instance lock when trying to stop a scheduler callback. * res_rtp_asterisk.c: Remove the lock in struct dtls_details and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads to prevent a race condition between starting and stopping a timeout timer. The race condition is no longer present between dtls_perform_handshake() and __rtp_recvfrom() because the instance lock prevents these functions from overlapping each other with regards to the timeout timer. * res_rtp_asterisk.c: Remove the lock in struct ast_rtp and use the struct ast_rtp_instance ao2 object lock instead. The lock was used to synchronize two threads using a condition signal to know when TURN negotiations complete. * res_rtp_asterisk.c: Avoid deadlock when trying to stop the TURN ioqueue_worker_thread(). We cannot hold the instance lock when trying to create or shut down the worker thread without a risk of deadlock. This patch exposed a race condition between a PJSIP serializer thread setting up an ICE session in ice_create() and another thread reading RTP packets. * res_rtp_asterisk.c:ice_create(): Set the new rtp->ice pointer after we have re-locked the RTP instance to prevent the other thread from trying to process ICE packets on an incomplete ICE session setup. A similar race condition is between a PJSIP serializer thread resetting up an ICE session in ice_create() and the timer_worker_thread() processing the completion of the previous ICE session. * res_rtp_asterisk.c:ast_rtp_on_ice_complete(): Protect against an uninitialized/null remote_address after calling update_address_with_ice_candidate(). * res_rtp_asterisk.c: Eliminate the chance of ice_reset_session() destroying and setting the rtp->ice pointer to NULL while other threads are using it by adding an ao2 wrapper around the PJPROJECT ice pointer. Now when we have to unlock the RTP instance object to call a PJPROJECT ICE function we will hold a ref to the wrapper. Also added some rtp->ice NULL checks after we relock the RTP instance and have to do something with the ICE structure. ASTERISK-26835 #close ASTERISK-26853 #close Change-Id: I780b39ec935dcefcce880d50c1a7261744f1d1b4
2017-03-22 21:05:49 +00:00
if (instance->engine->dtls) {
return &rtp_dtls_wrappers;
}
/* DTLS not available */
return NULL;
}
int ast_rtp_dtls_cfg_parse(struct ast_rtp_dtls_cfg *dtls_cfg, const char *name, const char *value)
{
if (!strcasecmp(name, "dtlsenable")) {
dtls_cfg->enabled = ast_true(value) ? 1 : 0;
} else if (!strcasecmp(name, "dtlsverify")) {
if (!strcasecmp(value, "yes")) {
dtls_cfg->verify = AST_RTP_DTLS_VERIFY_FINGERPRINT | AST_RTP_DTLS_VERIFY_CERTIFICATE;
} else if (!strcasecmp(value, "fingerprint")) {
dtls_cfg->verify = AST_RTP_DTLS_VERIFY_FINGERPRINT;
} else if (!strcasecmp(value, "certificate")) {
dtls_cfg->verify = AST_RTP_DTLS_VERIFY_CERTIFICATE;
} else if (!strcasecmp(value, "no")) {
dtls_cfg->verify = AST_RTP_DTLS_VERIFY_NONE;
} else {
return -1;
}
} else if (!strcasecmp(name, "dtlsrekey")) {
if (sscanf(value, "%30u", &dtls_cfg->rekey) != 1) {
return -1;
}
} else if (!strcasecmp(name, "dtlsautogeneratecert")) {
dtls_cfg->ephemeral_cert = ast_true(value) ? 1 : 0;
} else if (!strcasecmp(name, "dtlscertfile")) {
if (!ast_strlen_zero(value) && !ast_file_is_readable(value)) {
ast_log(LOG_ERROR, "%s file %s does not exist or is not readable\n", name, value);
return -1;
}
ast_free(dtls_cfg->certfile);
dtls_cfg->certfile = ast_strdup(value);
} else if (!strcasecmp(name, "dtlsprivatekey")) {
if (!ast_strlen_zero(value) && !ast_file_is_readable(value)) {
ast_log(LOG_ERROR, "%s file %s does not exist or is not readable\n", name, value);
return -1;
}
ast_free(dtls_cfg->pvtfile);
dtls_cfg->pvtfile = ast_strdup(value);
} else if (!strcasecmp(name, "dtlscipher")) {
ast_free(dtls_cfg->cipher);
dtls_cfg->cipher = ast_strdup(value);
} else if (!strcasecmp(name, "dtlscafile")) {
if (!ast_strlen_zero(value) && !ast_file_is_readable(value)) {
ast_log(LOG_ERROR, "%s file %s does not exist or is not readable\n", name, value);
return -1;
}
ast_free(dtls_cfg->cafile);
dtls_cfg->cafile = ast_strdup(value);
} else if (!strcasecmp(name, "dtlscapath") || !strcasecmp(name, "dtlscadir")) {
if (!ast_strlen_zero(value) && !ast_file_is_readable(value)) {
ast_log(LOG_ERROR, "%s file %s does not exist or is not readable\n", name, value);
return -1;
}
ast_free(dtls_cfg->capath);
dtls_cfg->capath = ast_strdup(value);
} else if (!strcasecmp(name, "dtlssetup")) {
if (!strcasecmp(value, "active")) {
dtls_cfg->default_setup = AST_RTP_DTLS_SETUP_ACTIVE;
} else if (!strcasecmp(value, "passive")) {
dtls_cfg->default_setup = AST_RTP_DTLS_SETUP_PASSIVE;
} else if (!strcasecmp(value, "actpass")) {
dtls_cfg->default_setup = AST_RTP_DTLS_SETUP_ACTPASS;
}
} else if (!strcasecmp(name, "dtlsfingerprint")) {
if (!strcasecmp(value, "sha-256")) {
dtls_cfg->hash = AST_RTP_DTLS_HASH_SHA256;
} else if (!strcasecmp(value, "sha-1")) {
dtls_cfg->hash = AST_RTP_DTLS_HASH_SHA1;
}
} else {
return -1;
}
return 0;
}
int ast_rtp_dtls_cfg_validate(struct ast_rtp_dtls_cfg *dtls_cfg)
{
if (dtls_cfg->ephemeral_cert) {
if (!ast_strlen_zero(dtls_cfg->certfile)) {
ast_log(LOG_ERROR, "You cannot request automatically generated certificates"
" (dtls_auto_generate_cert) and also specify a certificate file"
" (dtls_cert_file) at the same time\n");
return -1;
} else if (!ast_strlen_zero(dtls_cfg->pvtfile)
|| !ast_strlen_zero(dtls_cfg->cafile)
|| !ast_strlen_zero(dtls_cfg->capath)) {
ast_log(LOG_NOTICE, "dtls_pvt_file, dtls_cafile, and dtls_ca_path are"
" ignored when dtls_auto_generate_cert is enabled\n");
}
}
return 0;
}
void ast_rtp_dtls_cfg_copy(const struct ast_rtp_dtls_cfg *src_cfg, struct ast_rtp_dtls_cfg *dst_cfg)
{
ast_rtp_dtls_cfg_free(dst_cfg); /* Prevent a double-call leaking memory via ast_strdup */
dst_cfg->enabled = src_cfg->enabled;
dst_cfg->verify = src_cfg->verify;
dst_cfg->rekey = src_cfg->rekey;
dst_cfg->suite = src_cfg->suite;
dst_cfg->hash = src_cfg->hash;
dst_cfg->ephemeral_cert = src_cfg->ephemeral_cert;
dst_cfg->certfile = ast_strdup(src_cfg->certfile);
dst_cfg->pvtfile = ast_strdup(src_cfg->pvtfile);
dst_cfg->cipher = ast_strdup(src_cfg->cipher);
dst_cfg->cafile = ast_strdup(src_cfg->cafile);
dst_cfg->capath = ast_strdup(src_cfg->capath);
dst_cfg->default_setup = src_cfg->default_setup;
}
void ast_rtp_dtls_cfg_free(struct ast_rtp_dtls_cfg *dtls_cfg)
{
ast_free(dtls_cfg->certfile);
dtls_cfg->certfile = NULL;
ast_free(dtls_cfg->pvtfile);
dtls_cfg->pvtfile = NULL;
ast_free(dtls_cfg->cipher);
dtls_cfg->cipher = NULL;
ast_free(dtls_cfg->cafile);
dtls_cfg->cafile = NULL;
ast_free(dtls_cfg->capath);
dtls_cfg->capath = NULL;
}
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
/*! \internal
* \brief Small helper routine that cleans up entry i in
* \c ast_rtp_mime_types.
*/
static void rtp_engine_mime_type_cleanup(int i)
{
ao2_cleanup(ast_rtp_mime_types[i].payload_type.format);
memset(&ast_rtp_mime_types[i], 0, sizeof(struct ast_rtp_mime_type));
}
static void set_next_mime_type(struct ast_format *format, int rtp_code, const char *type, const char *subtype, unsigned int sample_rate)
{
int x;
ast_rwlock_wrlock(&mime_types_lock);
x = mime_types_len;
if (ARRAY_LEN(ast_rtp_mime_types) <= x) {
ast_rwlock_unlock(&mime_types_lock);
return;
}
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
/* Make sure any previous value in ast_rtp_mime_types is cleaned up */
memset(&ast_rtp_mime_types[x], 0, sizeof(struct ast_rtp_mime_type));
if (format) {
ast_rtp_mime_types[x].payload_type.asterisk_format = 1;
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
ast_rtp_mime_types[x].payload_type.format = ao2_bump(format);
} else {
ast_rtp_mime_types[x].payload_type.rtp_code = rtp_code;
}
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
ast_copy_string(ast_rtp_mime_types[x].type, type, sizeof(ast_rtp_mime_types[x].type));
ast_copy_string(ast_rtp_mime_types[x].subtype, subtype, sizeof(ast_rtp_mime_types[x].subtype));
ast_rtp_mime_types[x].sample_rate = sample_rate;
mime_types_len++;
ast_rwlock_unlock(&mime_types_lock);
}
static void add_static_payload(int payload, struct ast_format *format, int rtp_code)
{
struct ast_rtp_payload_type *type;
/*
* ARRAY_LEN's result is cast to an int so 'map' is not autocast to a size_t,
* which if negative would cause an assertion.
*/
ast_assert(payload < (int)ARRAY_LEN(static_RTP_PT));
if (ast_option_rtpusedynamic && payload < 0) {
/*
* We're going to build dynamic payloads dynamically. An RTP code is
* required otherwise one will be dynamically allocated per instance.
*/
return;
}
/*
* Either the given payload is truly a static type, or Asterisk is
* globally storing the dynamic payloads in the static_RTP_PT object.
*/
ast_rwlock_wrlock(&static_RTP_PT_lock);
if (payload < 0) {
/*
* This is a dynamic payload that will be stored globally,
* so find the next available empty slot.
*/
payload = find_unused_payload(NULL);
if (payload < 0) {
ast_log(LOG_WARNING, "No dynamic RTP payload type values available "
"for %s - %d!\n", format ? ast_format_get_name(format) : "", rtp_code);
ast_rwlock_unlock(&static_RTP_PT_lock);
return;
}
}
type = rtp_payload_type_alloc(format, payload, rtp_code, 1);
if (type) {
ao2_cleanup(static_RTP_PT[payload]);
static_RTP_PT[payload] = type;
}
ast_rwlock_unlock(&static_RTP_PT_lock);
}
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
int ast_rtp_engine_load_format(struct ast_format *format)
{
char *codec_name = ast_strdupa(ast_format_get_codec_name(format));
codec_name = ast_str_to_upper(codec_name);
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
set_next_mime_type(format,
0,
ast_codec_media_type2str(ast_format_get_type(format)),
codec_name,
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
ast_format_get_sample_rate(format));
add_static_payload(-1, format, 0);
return 0;
}
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
int ast_rtp_engine_unload_format(struct ast_format *format)
{
int x;
int y = 0;
ast_rwlock_wrlock(&static_RTP_PT_lock);
/* remove everything pertaining to this format id from the lists */
for (x = 0; x < AST_RTP_MAX_PT; x++) {
if (static_RTP_PT[x]
&& ast_format_cmp(static_RTP_PT[x]->format, format) == AST_FORMAT_CMP_EQUAL) {
ao2_ref(static_RTP_PT[x], -1);
static_RTP_PT[x] = NULL;
}
}
ast_rwlock_unlock(&static_RTP_PT_lock);
ast_rwlock_wrlock(&mime_types_lock);
/* rebuild the list skipping the items matching this id */
for (x = 0; x < mime_types_len; x++) {
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
if (ast_format_cmp(ast_rtp_mime_types[x].payload_type.format, format) == AST_FORMAT_CMP_EQUAL) {
rtp_engine_mime_type_cleanup(x);
continue;
}
if (x != y) {
ast_rtp_mime_types[y] = ast_rtp_mime_types[x];
}
y++;
}
mime_types_len = y;
ast_rwlock_unlock(&mime_types_lock);
return 0;
}
/*!
* \internal
* \brief \ref stasis message payload for RTCP messages
*/
Refactor RTCP events over to Stasis; associate with channels This patch does the following: * It merges Jaco Kroon's patch from ASTERISK-20754, which provides channel information in the RTCP events. Because Stasis provides a cache, Jaco's patch was modified to pass the channel uniqueid to the RTP layer as opposed to a pointer to the channel. This has the following benefits: (1) It keeps the RTP engine 'clean' of references back to channels (2) It prevents circular dependencies and other potential ref counting issues * The RTP engine now allows any RTP implementation to raise RTCP messages. Potentially, other implementations (such as res_rtp_multicast) could also raise RTCP information. The engine provides structs to represent RTCP headers and RTCP SR/RR reports. * Some general refactoring in res_rtp_asterisk was done to try and tame the RTCP code. It isn't perfect - that's *way* beyond the scope of this work - but it does feel marginally better. * A few random bugs were fixed in the RTCP statistics. (Example: performing an assignment of a = a is probably not correct) * We now raise RTCP events for each SR/RR sent/received. Previously we wouldn't raise an event when we sent a RR report. Note that this work will be of use to others who want to monitor call quality or build modules that report call quality statistics. Since the events are now moving across the Stasis message bus, this is far easier to accomplish. It is also a first step (though by no means the last step) towards getting Olle's pinefrog work incorporated. Again: note that the patch by Jaco Kroon was modified slightly for this work; however, he did all of the hard work in finding the right places to set the channel in the RTP engine across the channel drivers. Much thanks goes to Jaco for his hard work here. Review: https://reviewboard.asterisk.org/r/2603/ (closes issue ASTERISK-20574) Reported by: Jaco Kroon patches: asterisk-rtcp-channel.patch uploaded by jkroon (License 5671) (closes issue ASTERISK-21471) Reported by: Matt Jordan git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@393740 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2013-07-05 17:33:33 +00:00
struct rtcp_message_payload {
struct ast_channel_snapshot *snapshot; /*< The channel snapshot, if available */
struct ast_rtp_rtcp_report *report; /*< The RTCP report */
struct ast_json *blob; /*< Extra JSON data to publish */
};
static void rtcp_message_payload_dtor(void *obj)
{
struct rtcp_message_payload *payload = obj;
ao2_cleanup(payload->report);
ao2_cleanup(payload->snapshot);
ast_json_unref(payload->blob);
}
static struct ast_manager_event_blob *rtcp_report_to_ami(struct stasis_message *msg)
{
struct rtcp_message_payload *payload = stasis_message_data(msg);
RAII_VAR(struct ast_str *, channel_string, NULL, ast_free);
RAII_VAR(struct ast_str *, packet_string, ast_str_create(512), ast_free);
unsigned int ssrc = payload->report->ssrc;
unsigned int type = payload->report->type;
unsigned int report_count = payload->report->reception_report_count;
int i;
if (!packet_string) {
return NULL;
}
if (payload->snapshot) {
channel_string = ast_manager_build_channel_state_string(payload->snapshot);
if (!channel_string) {
return NULL;
}
}
if (payload->blob) {
/* Optional data */
struct ast_json *to = ast_json_object_get(payload->blob, "to");
struct ast_json *from = ast_json_object_get(payload->blob, "from");
struct ast_json *rtt = ast_json_object_get(payload->blob, "rtt");
if (to) {
ast_str_append(&packet_string, 0, "To: %s\r\n", ast_json_string_get(to));
}
if (from) {
ast_str_append(&packet_string, 0, "From: %s\r\n", ast_json_string_get(from));
}
if (rtt) {
ast_str_append(&packet_string, 0, "RTT: %4.4f\r\n", ast_json_real_get(rtt));
}
}
ast_str_append(&packet_string, 0, "SSRC: 0x%.8x\r\n", ssrc);
ast_str_append(&packet_string, 0, "PT: %u(%s)\r\n", type, type== AST_RTP_RTCP_SR ? "SR" : "RR");
ast_str_append(&packet_string, 0, "ReportCount: %u\r\n", report_count);
if (type == AST_RTP_RTCP_SR) {
ast_str_append(&packet_string, 0, "SentNTP: %lu.%06lu\r\n",
(unsigned long)payload->report->sender_information.ntp_timestamp.tv_sec,
(unsigned long)payload->report->sender_information.ntp_timestamp.tv_usec);
Refactor RTCP events over to Stasis; associate with channels This patch does the following: * It merges Jaco Kroon's patch from ASTERISK-20754, which provides channel information in the RTCP events. Because Stasis provides a cache, Jaco's patch was modified to pass the channel uniqueid to the RTP layer as opposed to a pointer to the channel. This has the following benefits: (1) It keeps the RTP engine 'clean' of references back to channels (2) It prevents circular dependencies and other potential ref counting issues * The RTP engine now allows any RTP implementation to raise RTCP messages. Potentially, other implementations (such as res_rtp_multicast) could also raise RTCP information. The engine provides structs to represent RTCP headers and RTCP SR/RR reports. * Some general refactoring in res_rtp_asterisk was done to try and tame the RTCP code. It isn't perfect - that's *way* beyond the scope of this work - but it does feel marginally better. * A few random bugs were fixed in the RTCP statistics. (Example: performing an assignment of a = a is probably not correct) * We now raise RTCP events for each SR/RR sent/received. Previously we wouldn't raise an event when we sent a RR report. Note that this work will be of use to others who want to monitor call quality or build modules that report call quality statistics. Since the events are now moving across the Stasis message bus, this is far easier to accomplish. It is also a first step (though by no means the last step) towards getting Olle's pinefrog work incorporated. Again: note that the patch by Jaco Kroon was modified slightly for this work; however, he did all of the hard work in finding the right places to set the channel in the RTP engine across the channel drivers. Much thanks goes to Jaco for his hard work here. Review: https://reviewboard.asterisk.org/r/2603/ (closes issue ASTERISK-20574) Reported by: Jaco Kroon patches: asterisk-rtcp-channel.patch uploaded by jkroon (License 5671) (closes issue ASTERISK-21471) Reported by: Matt Jordan git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@393740 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2013-07-05 17:33:33 +00:00
ast_str_append(&packet_string, 0, "SentRTP: %u\r\n",
payload->report->sender_information.rtp_timestamp);
ast_str_append(&packet_string, 0, "SentPackets: %u\r\n",
payload->report->sender_information.packet_count);
ast_str_append(&packet_string, 0, "SentOctets: %u\r\n",
payload->report->sender_information.octet_count);
}
for (i = 0; i < report_count; i++) {
RAII_VAR(struct ast_str *, report_string, NULL, ast_free);
if (!payload->report->report_block[i]) {
break;
}
report_string = ast_str_create(256);
if (!report_string) {
return NULL;
}
ast_str_append(&report_string, 0, "Report%dSourceSSRC: 0x%.8x\r\n",
i, payload->report->report_block[i]->source_ssrc);
ast_str_append(&report_string, 0, "Report%dFractionLost: %d\r\n",
Refactor RTCP events over to Stasis; associate with channels This patch does the following: * It merges Jaco Kroon's patch from ASTERISK-20754, which provides channel information in the RTCP events. Because Stasis provides a cache, Jaco's patch was modified to pass the channel uniqueid to the RTP layer as opposed to a pointer to the channel. This has the following benefits: (1) It keeps the RTP engine 'clean' of references back to channels (2) It prevents circular dependencies and other potential ref counting issues * The RTP engine now allows any RTP implementation to raise RTCP messages. Potentially, other implementations (such as res_rtp_multicast) could also raise RTCP information. The engine provides structs to represent RTCP headers and RTCP SR/RR reports. * Some general refactoring in res_rtp_asterisk was done to try and tame the RTCP code. It isn't perfect - that's *way* beyond the scope of this work - but it does feel marginally better. * A few random bugs were fixed in the RTCP statistics. (Example: performing an assignment of a = a is probably not correct) * We now raise RTCP events for each SR/RR sent/received. Previously we wouldn't raise an event when we sent a RR report. Note that this work will be of use to others who want to monitor call quality or build modules that report call quality statistics. Since the events are now moving across the Stasis message bus, this is far easier to accomplish. It is also a first step (though by no means the last step) towards getting Olle's pinefrog work incorporated. Again: note that the patch by Jaco Kroon was modified slightly for this work; however, he did all of the hard work in finding the right places to set the channel in the RTP engine across the channel drivers. Much thanks goes to Jaco for his hard work here. Review: https://reviewboard.asterisk.org/r/2603/ (closes issue ASTERISK-20574) Reported by: Jaco Kroon patches: asterisk-rtcp-channel.patch uploaded by jkroon (License 5671) (closes issue ASTERISK-21471) Reported by: Matt Jordan git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@393740 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2013-07-05 17:33:33 +00:00
i, payload->report->report_block[i]->lost_count.fraction);
ast_str_append(&report_string, 0, "Report%dCumulativeLost: %u\r\n",
i, payload->report->report_block[i]->lost_count.packets);
ast_str_append(&report_string, 0, "Report%dHighestSequence: %u\r\n",
i, payload->report->report_block[i]->highest_seq_no & 0xffff);
ast_str_append(&report_string, 0, "Report%dSequenceNumberCycles: %u\r\n",
i, payload->report->report_block[i]->highest_seq_no >> 16);
ast_str_append(&report_string, 0, "Report%dIAJitter: %u\r\n",
i, payload->report->report_block[i]->ia_jitter);
ast_str_append(&report_string, 0, "Report%dLSR: %u\r\n",
i, payload->report->report_block[i]->lsr);
ast_str_append(&report_string, 0, "Report%dDLSR: %4.4f\r\n",
i, ((double)payload->report->report_block[i]->dlsr) / 65536);
ast_str_append(&packet_string, 0, "%s", ast_str_buffer(report_string));
}
return ast_manager_event_blob_create(EVENT_FLAG_REPORTING,
stasis_message_type(msg) == ast_rtp_rtcp_received_type() ? "RTCPReceived" : "RTCPSent",
"%s%s",
AS_OR(channel_string, ""),
ast_str_buffer(packet_string));
}
static struct ast_json *rtcp_report_to_json(struct stasis_message *msg,
const struct stasis_message_sanitizer *sanitize)
Refactor RTCP events over to Stasis; associate with channels This patch does the following: * It merges Jaco Kroon's patch from ASTERISK-20754, which provides channel information in the RTCP events. Because Stasis provides a cache, Jaco's patch was modified to pass the channel uniqueid to the RTP layer as opposed to a pointer to the channel. This has the following benefits: (1) It keeps the RTP engine 'clean' of references back to channels (2) It prevents circular dependencies and other potential ref counting issues * The RTP engine now allows any RTP implementation to raise RTCP messages. Potentially, other implementations (such as res_rtp_multicast) could also raise RTCP information. The engine provides structs to represent RTCP headers and RTCP SR/RR reports. * Some general refactoring in res_rtp_asterisk was done to try and tame the RTCP code. It isn't perfect - that's *way* beyond the scope of this work - but it does feel marginally better. * A few random bugs were fixed in the RTCP statistics. (Example: performing an assignment of a = a is probably not correct) * We now raise RTCP events for each SR/RR sent/received. Previously we wouldn't raise an event when we sent a RR report. Note that this work will be of use to others who want to monitor call quality or build modules that report call quality statistics. Since the events are now moving across the Stasis message bus, this is far easier to accomplish. It is also a first step (though by no means the last step) towards getting Olle's pinefrog work incorporated. Again: note that the patch by Jaco Kroon was modified slightly for this work; however, he did all of the hard work in finding the right places to set the channel in the RTP engine across the channel drivers. Much thanks goes to Jaco for his hard work here. Review: https://reviewboard.asterisk.org/r/2603/ (closes issue ASTERISK-20574) Reported by: Jaco Kroon patches: asterisk-rtcp-channel.patch uploaded by jkroon (License 5671) (closes issue ASTERISK-21471) Reported by: Matt Jordan git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@393740 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2013-07-05 17:33:33 +00:00
{
struct rtcp_message_payload *payload = stasis_message_data(msg);
struct ast_json *json_rtcp_report = NULL;
struct ast_json *json_rtcp_report_blocks;
struct ast_json *json_rtcp_sender_info = NULL;
struct ast_json *json_channel = NULL;
Refactor RTCP events over to Stasis; associate with channels This patch does the following: * It merges Jaco Kroon's patch from ASTERISK-20754, which provides channel information in the RTCP events. Because Stasis provides a cache, Jaco's patch was modified to pass the channel uniqueid to the RTP layer as opposed to a pointer to the channel. This has the following benefits: (1) It keeps the RTP engine 'clean' of references back to channels (2) It prevents circular dependencies and other potential ref counting issues * The RTP engine now allows any RTP implementation to raise RTCP messages. Potentially, other implementations (such as res_rtp_multicast) could also raise RTCP information. The engine provides structs to represent RTCP headers and RTCP SR/RR reports. * Some general refactoring in res_rtp_asterisk was done to try and tame the RTCP code. It isn't perfect - that's *way* beyond the scope of this work - but it does feel marginally better. * A few random bugs were fixed in the RTCP statistics. (Example: performing an assignment of a = a is probably not correct) * We now raise RTCP events for each SR/RR sent/received. Previously we wouldn't raise an event when we sent a RR report. Note that this work will be of use to others who want to monitor call quality or build modules that report call quality statistics. Since the events are now moving across the Stasis message bus, this is far easier to accomplish. It is also a first step (though by no means the last step) towards getting Olle's pinefrog work incorporated. Again: note that the patch by Jaco Kroon was modified slightly for this work; however, he did all of the hard work in finding the right places to set the channel in the RTP engine across the channel drivers. Much thanks goes to Jaco for his hard work here. Review: https://reviewboard.asterisk.org/r/2603/ (closes issue ASTERISK-20574) Reported by: Jaco Kroon patches: asterisk-rtcp-channel.patch uploaded by jkroon (License 5671) (closes issue ASTERISK-21471) Reported by: Matt Jordan git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@393740 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2013-07-05 17:33:33 +00:00
int i;
json_rtcp_report_blocks = ast_json_array_create();
if (!json_rtcp_report_blocks) {
return NULL;
}
main/rtp_engine: Fix crash when processing more than one RTCP report info block Asterisk - in res_rtp_asterisk - only understands a single RTCP report info block. When the RTCP information was refactored in the RTP Engine to be pushed over the Stasis message bus, I put in the hooks into the engine to handle multiple RTCP report info blocks, in the hope that a future RTP implementation would be able to provide that data. Unfortunately, res_rtp_asterisk has a tendency to "lie": (1) It will send RTCP reports with a reception_report_count greater than 1 (which is pulled directly from the RTCP packet itself, so that part is correct) (2) It will only provide a single report block When the rtp_engine goes to convert this to a JSON blob, hilarity ensues as it looks for a report block that doesn't exist. This patch updates the rtp_engine to be a bit more skeptical about what it is presented with. While this could also be fixed in res_rtp_asterisk, this patch prefers to fix it in the engine for two reasons: (1) The engine is designed to work with multiple RTP implementation, and hence having it be more robust is a good thing (tm) (2) res_rtp_asterisk's handling of RTCP information is "fun". It should report the correct reception_report_count; ideally it should also be giving us all of the blocks - but it is *definitely* not designed to do that. Going down that road is a non-trivial effort. Review: https://reviewboard.asterisk.org/r/4158/ ASTERISK-24489 #close Reported by: Gregory Malsack Tested by: Gregory Malsack ASTERISK-24498 #close Reported by: Beppo Mazzucato Tested by: Beppo Maazucato ........ Merged revisions 427762 from http://svn.asterisk.org/svn/asterisk/branches/12 ........ Merged revisions 427763 from http://svn.asterisk.org/svn/asterisk/branches/13 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@427771 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-11-13 00:23:20 +00:00
for (i = 0; i < payload->report->reception_report_count && payload->report->report_block[i]; i++) {
Refactor RTCP events over to Stasis; associate with channels This patch does the following: * It merges Jaco Kroon's patch from ASTERISK-20754, which provides channel information in the RTCP events. Because Stasis provides a cache, Jaco's patch was modified to pass the channel uniqueid to the RTP layer as opposed to a pointer to the channel. This has the following benefits: (1) It keeps the RTP engine 'clean' of references back to channels (2) It prevents circular dependencies and other potential ref counting issues * The RTP engine now allows any RTP implementation to raise RTCP messages. Potentially, other implementations (such as res_rtp_multicast) could also raise RTCP information. The engine provides structs to represent RTCP headers and RTCP SR/RR reports. * Some general refactoring in res_rtp_asterisk was done to try and tame the RTCP code. It isn't perfect - that's *way* beyond the scope of this work - but it does feel marginally better. * A few random bugs were fixed in the RTCP statistics. (Example: performing an assignment of a = a is probably not correct) * We now raise RTCP events for each SR/RR sent/received. Previously we wouldn't raise an event when we sent a RR report. Note that this work will be of use to others who want to monitor call quality or build modules that report call quality statistics. Since the events are now moving across the Stasis message bus, this is far easier to accomplish. It is also a first step (though by no means the last step) towards getting Olle's pinefrog work incorporated. Again: note that the patch by Jaco Kroon was modified slightly for this work; however, he did all of the hard work in finding the right places to set the channel in the RTP engine across the channel drivers. Much thanks goes to Jaco for his hard work here. Review: https://reviewboard.asterisk.org/r/2603/ (closes issue ASTERISK-20574) Reported by: Jaco Kroon patches: asterisk-rtcp-channel.patch uploaded by jkroon (License 5671) (closes issue ASTERISK-21471) Reported by: Matt Jordan git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@393740 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2013-07-05 17:33:33 +00:00
struct ast_json *json_report_block;
res_hep_rtcp: Add module that sends RTCP information to a Homer Server This patch adds a new module to Asterisk, res_hep_rtcp. The module subscribes to the RTCP topics in Stasis and receives RTCP information back from the message bus. It encodes into HEPv3 packets and sends the information to the res_hep module for transmission. Using this, someone with a Homer server can get live call quality monitoring for all RTP-based channels in their Asterisk 12+ systems. In addition, there were a few bugs in the RTP engine, res_rtp_asterisk, and chan_pjsip that were uncovered by the tests written for the Asterisk Test Suite. This patch fixes the following: 1) chan_pjsip failed to set its channel unique ids on its RTP instance on outbound calls. It now does this in the appropriate location, in the serialized call callback. 2) The rtp_engine was overflowing some values when packed into JSON. Specifically, some longs and unsigned ints can't be be packed into integer values, for obvious reasons. Since libjansson only supports integers, floats, strings, booleans, and objects, we print these values into strings. 3) res_rtp_asterisk had a few problems: (a) it would emit a source IP address of 0.0.0.0 if bound to that IP address. We now use ast_find_ourip to get a better IP address, and properly marshal the result into an ast_strdupa'd string. (b) Reports can be generated with no report bodies. In particular, this occurs when a sender is transmitting information to a receiver (who will send no RTP back to the sender). As such, the sender has no report body for what it received. We now properly handle this case, and the sender will emit SR reports with no body. Likewise, if we receive an RTCP packet with no report body, we will still generate the appropriate events. ASTERISK-24119 #close ........ Merged revisions 419823 from http://svn.asterisk.org/svn/asterisk/branches/12 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419825 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-31 11:57:51 +00:00
char str_lsr[32];
res_hep_rtcp: Add module that sends RTCP information to a Homer Server This patch adds a new module to Asterisk, res_hep_rtcp. The module subscribes to the RTCP topics in Stasis and receives RTCP information back from the message bus. It encodes into HEPv3 packets and sends the information to the res_hep module for transmission. Using this, someone with a Homer server can get live call quality monitoring for all RTP-based channels in their Asterisk 12+ systems. In addition, there were a few bugs in the RTP engine, res_rtp_asterisk, and chan_pjsip that were uncovered by the tests written for the Asterisk Test Suite. This patch fixes the following: 1) chan_pjsip failed to set its channel unique ids on its RTP instance on outbound calls. It now does this in the appropriate location, in the serialized call callback. 2) The rtp_engine was overflowing some values when packed into JSON. Specifically, some longs and unsigned ints can't be be packed into integer values, for obvious reasons. Since libjansson only supports integers, floats, strings, booleans, and objects, we print these values into strings. 3) res_rtp_asterisk had a few problems: (a) it would emit a source IP address of 0.0.0.0 if bound to that IP address. We now use ast_find_ourip to get a better IP address, and properly marshal the result into an ast_strdupa'd string. (b) Reports can be generated with no report bodies. In particular, this occurs when a sender is transmitting information to a receiver (who will send no RTP back to the sender). As such, the sender has no report body for what it received. We now properly handle this case, and the sender will emit SR reports with no body. Likewise, if we receive an RTCP packet with no report body, we will still generate the appropriate events. ASTERISK-24119 #close ........ Merged revisions 419823 from http://svn.asterisk.org/svn/asterisk/branches/12 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419825 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-31 11:57:51 +00:00
snprintf(str_lsr, sizeof(str_lsr), "%u", payload->report->report_block[i]->lsr);
json_report_block = ast_json_pack("{s: I, s: I, s: I, s: I, s: I, s: s, s: I}",
"source_ssrc", (ast_json_int_t)payload->report->report_block[i]->source_ssrc,
"fraction_lost", (ast_json_int_t)payload->report->report_block[i]->lost_count.fraction,
"packets_lost", (ast_json_int_t)payload->report->report_block[i]->lost_count.packets,
"highest_seq_no", (ast_json_int_t)payload->report->report_block[i]->highest_seq_no,
"ia_jitter", (ast_json_int_t)payload->report->report_block[i]->ia_jitter,
"lsr", str_lsr,
"dlsr", (ast_json_int_t)payload->report->report_block[i]->dlsr);
if (!json_report_block
|| ast_json_array_append(json_rtcp_report_blocks, json_report_block)) {
ast_json_unref(json_rtcp_report_blocks);
Refactor RTCP events over to Stasis; associate with channels This patch does the following: * It merges Jaco Kroon's patch from ASTERISK-20754, which provides channel information in the RTCP events. Because Stasis provides a cache, Jaco's patch was modified to pass the channel uniqueid to the RTP layer as opposed to a pointer to the channel. This has the following benefits: (1) It keeps the RTP engine 'clean' of references back to channels (2) It prevents circular dependencies and other potential ref counting issues * The RTP engine now allows any RTP implementation to raise RTCP messages. Potentially, other implementations (such as res_rtp_multicast) could also raise RTCP information. The engine provides structs to represent RTCP headers and RTCP SR/RR reports. * Some general refactoring in res_rtp_asterisk was done to try and tame the RTCP code. It isn't perfect - that's *way* beyond the scope of this work - but it does feel marginally better. * A few random bugs were fixed in the RTCP statistics. (Example: performing an assignment of a = a is probably not correct) * We now raise RTCP events for each SR/RR sent/received. Previously we wouldn't raise an event when we sent a RR report. Note that this work will be of use to others who want to monitor call quality or build modules that report call quality statistics. Since the events are now moving across the Stasis message bus, this is far easier to accomplish. It is also a first step (though by no means the last step) towards getting Olle's pinefrog work incorporated. Again: note that the patch by Jaco Kroon was modified slightly for this work; however, he did all of the hard work in finding the right places to set the channel in the RTP engine across the channel drivers. Much thanks goes to Jaco for his hard work here. Review: https://reviewboard.asterisk.org/r/2603/ (closes issue ASTERISK-20574) Reported by: Jaco Kroon patches: asterisk-rtcp-channel.patch uploaded by jkroon (License 5671) (closes issue ASTERISK-21471) Reported by: Matt Jordan git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@393740 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2013-07-05 17:33:33 +00:00
return NULL;
}
}
if (payload->report->type == AST_RTP_RTCP_SR) {
res_hep_rtcp: Add module that sends RTCP information to a Homer Server This patch adds a new module to Asterisk, res_hep_rtcp. The module subscribes to the RTCP topics in Stasis and receives RTCP information back from the message bus. It encodes into HEPv3 packets and sends the information to the res_hep module for transmission. Using this, someone with a Homer server can get live call quality monitoring for all RTP-based channels in their Asterisk 12+ systems. In addition, there were a few bugs in the RTP engine, res_rtp_asterisk, and chan_pjsip that were uncovered by the tests written for the Asterisk Test Suite. This patch fixes the following: 1) chan_pjsip failed to set its channel unique ids on its RTP instance on outbound calls. It now does this in the appropriate location, in the serialized call callback. 2) The rtp_engine was overflowing some values when packed into JSON. Specifically, some longs and unsigned ints can't be be packed into integer values, for obvious reasons. Since libjansson only supports integers, floats, strings, booleans, and objects, we print these values into strings. 3) res_rtp_asterisk had a few problems: (a) it would emit a source IP address of 0.0.0.0 if bound to that IP address. We now use ast_find_ourip to get a better IP address, and properly marshal the result into an ast_strdupa'd string. (b) Reports can be generated with no report bodies. In particular, this occurs when a sender is transmitting information to a receiver (who will send no RTP back to the sender). As such, the sender has no report body for what it received. We now properly handle this case, and the sender will emit SR reports with no body. Likewise, if we receive an RTCP packet with no report body, we will still generate the appropriate events. ASTERISK-24119 #close ........ Merged revisions 419823 from http://svn.asterisk.org/svn/asterisk/branches/12 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419825 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-31 11:57:51 +00:00
char sec[32];
char usec[32];
snprintf(sec, sizeof(sec), "%lu", (unsigned long)payload->report->sender_information.ntp_timestamp.tv_sec);
snprintf(usec, sizeof(usec), "%lu", (unsigned long)payload->report->sender_information.ntp_timestamp.tv_usec);
json_rtcp_sender_info = ast_json_pack("{s: s, s: s, s: I, s: I, s: I}",
"ntp_timestamp_sec", sec,
"ntp_timestamp_usec", usec,
"rtp_timestamp", (ast_json_int_t)payload->report->sender_information.rtp_timestamp,
"packets", (ast_json_int_t)payload->report->sender_information.packet_count,
"octets", (ast_json_int_t)payload->report->sender_information.octet_count);
Refactor RTCP events over to Stasis; associate with channels This patch does the following: * It merges Jaco Kroon's patch from ASTERISK-20754, which provides channel information in the RTCP events. Because Stasis provides a cache, Jaco's patch was modified to pass the channel uniqueid to the RTP layer as opposed to a pointer to the channel. This has the following benefits: (1) It keeps the RTP engine 'clean' of references back to channels (2) It prevents circular dependencies and other potential ref counting issues * The RTP engine now allows any RTP implementation to raise RTCP messages. Potentially, other implementations (such as res_rtp_multicast) could also raise RTCP information. The engine provides structs to represent RTCP headers and RTCP SR/RR reports. * Some general refactoring in res_rtp_asterisk was done to try and tame the RTCP code. It isn't perfect - that's *way* beyond the scope of this work - but it does feel marginally better. * A few random bugs were fixed in the RTCP statistics. (Example: performing an assignment of a = a is probably not correct) * We now raise RTCP events for each SR/RR sent/received. Previously we wouldn't raise an event when we sent a RR report. Note that this work will be of use to others who want to monitor call quality or build modules that report call quality statistics. Since the events are now moving across the Stasis message bus, this is far easier to accomplish. It is also a first step (though by no means the last step) towards getting Olle's pinefrog work incorporated. Again: note that the patch by Jaco Kroon was modified slightly for this work; however, he did all of the hard work in finding the right places to set the channel in the RTP engine across the channel drivers. Much thanks goes to Jaco for his hard work here. Review: https://reviewboard.asterisk.org/r/2603/ (closes issue ASTERISK-20574) Reported by: Jaco Kroon patches: asterisk-rtcp-channel.patch uploaded by jkroon (License 5671) (closes issue ASTERISK-21471) Reported by: Matt Jordan git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@393740 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2013-07-05 17:33:33 +00:00
if (!json_rtcp_sender_info) {
ast_json_unref(json_rtcp_report_blocks);
Refactor RTCP events over to Stasis; associate with channels This patch does the following: * It merges Jaco Kroon's patch from ASTERISK-20754, which provides channel information in the RTCP events. Because Stasis provides a cache, Jaco's patch was modified to pass the channel uniqueid to the RTP layer as opposed to a pointer to the channel. This has the following benefits: (1) It keeps the RTP engine 'clean' of references back to channels (2) It prevents circular dependencies and other potential ref counting issues * The RTP engine now allows any RTP implementation to raise RTCP messages. Potentially, other implementations (such as res_rtp_multicast) could also raise RTCP information. The engine provides structs to represent RTCP headers and RTCP SR/RR reports. * Some general refactoring in res_rtp_asterisk was done to try and tame the RTCP code. It isn't perfect - that's *way* beyond the scope of this work - but it does feel marginally better. * A few random bugs were fixed in the RTCP statistics. (Example: performing an assignment of a = a is probably not correct) * We now raise RTCP events for each SR/RR sent/received. Previously we wouldn't raise an event when we sent a RR report. Note that this work will be of use to others who want to monitor call quality or build modules that report call quality statistics. Since the events are now moving across the Stasis message bus, this is far easier to accomplish. It is also a first step (though by no means the last step) towards getting Olle's pinefrog work incorporated. Again: note that the patch by Jaco Kroon was modified slightly for this work; however, he did all of the hard work in finding the right places to set the channel in the RTP engine across the channel drivers. Much thanks goes to Jaco for his hard work here. Review: https://reviewboard.asterisk.org/r/2603/ (closes issue ASTERISK-20574) Reported by: Jaco Kroon patches: asterisk-rtcp-channel.patch uploaded by jkroon (License 5671) (closes issue ASTERISK-21471) Reported by: Matt Jordan git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@393740 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2013-07-05 17:33:33 +00:00
return NULL;
}
}
json_rtcp_report = ast_json_pack("{s: I, s: I, s: i, s: o, s: o}",
"ssrc", (ast_json_int_t)payload->report->ssrc,
"type", (ast_json_int_t)payload->report->type,
"report_count", payload->report->reception_report_count,
"sender_information", json_rtcp_sender_info ?: ast_json_null(),
"report_blocks", json_rtcp_report_blocks);
Refactor RTCP events over to Stasis; associate with channels This patch does the following: * It merges Jaco Kroon's patch from ASTERISK-20754, which provides channel information in the RTCP events. Because Stasis provides a cache, Jaco's patch was modified to pass the channel uniqueid to the RTP layer as opposed to a pointer to the channel. This has the following benefits: (1) It keeps the RTP engine 'clean' of references back to channels (2) It prevents circular dependencies and other potential ref counting issues * The RTP engine now allows any RTP implementation to raise RTCP messages. Potentially, other implementations (such as res_rtp_multicast) could also raise RTCP information. The engine provides structs to represent RTCP headers and RTCP SR/RR reports. * Some general refactoring in res_rtp_asterisk was done to try and tame the RTCP code. It isn't perfect - that's *way* beyond the scope of this work - but it does feel marginally better. * A few random bugs were fixed in the RTCP statistics. (Example: performing an assignment of a = a is probably not correct) * We now raise RTCP events for each SR/RR sent/received. Previously we wouldn't raise an event when we sent a RR report. Note that this work will be of use to others who want to monitor call quality or build modules that report call quality statistics. Since the events are now moving across the Stasis message bus, this is far easier to accomplish. It is also a first step (though by no means the last step) towards getting Olle's pinefrog work incorporated. Again: note that the patch by Jaco Kroon was modified slightly for this work; however, he did all of the hard work in finding the right places to set the channel in the RTP engine across the channel drivers. Much thanks goes to Jaco for his hard work here. Review: https://reviewboard.asterisk.org/r/2603/ (closes issue ASTERISK-20574) Reported by: Jaco Kroon patches: asterisk-rtcp-channel.patch uploaded by jkroon (License 5671) (closes issue ASTERISK-21471) Reported by: Matt Jordan git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@393740 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2013-07-05 17:33:33 +00:00
if (!json_rtcp_report) {
return NULL;
}
if (payload->snapshot) {
json_channel = ast_channel_snapshot_to_json(payload->snapshot, sanitize);
if (!json_channel) {
ast_json_unref(json_rtcp_report);
return NULL;
}
}
return ast_json_pack("{s: s, s: o?, s: o, s: O?}",
"type", stasis_message_type(msg) == ast_rtp_rtcp_received_type() ? "RTCPReceived" : "RTCPSent",
"channel", json_channel,
"rtcp_report", json_rtcp_report,
"blob", payload->blob);
Refactor RTCP events over to Stasis; associate with channels This patch does the following: * It merges Jaco Kroon's patch from ASTERISK-20754, which provides channel information in the RTCP events. Because Stasis provides a cache, Jaco's patch was modified to pass the channel uniqueid to the RTP layer as opposed to a pointer to the channel. This has the following benefits: (1) It keeps the RTP engine 'clean' of references back to channels (2) It prevents circular dependencies and other potential ref counting issues * The RTP engine now allows any RTP implementation to raise RTCP messages. Potentially, other implementations (such as res_rtp_multicast) could also raise RTCP information. The engine provides structs to represent RTCP headers and RTCP SR/RR reports. * Some general refactoring in res_rtp_asterisk was done to try and tame the RTCP code. It isn't perfect - that's *way* beyond the scope of this work - but it does feel marginally better. * A few random bugs were fixed in the RTCP statistics. (Example: performing an assignment of a = a is probably not correct) * We now raise RTCP events for each SR/RR sent/received. Previously we wouldn't raise an event when we sent a RR report. Note that this work will be of use to others who want to monitor call quality or build modules that report call quality statistics. Since the events are now moving across the Stasis message bus, this is far easier to accomplish. It is also a first step (though by no means the last step) towards getting Olle's pinefrog work incorporated. Again: note that the patch by Jaco Kroon was modified slightly for this work; however, he did all of the hard work in finding the right places to set the channel in the RTP engine across the channel drivers. Much thanks goes to Jaco for his hard work here. Review: https://reviewboard.asterisk.org/r/2603/ (closes issue ASTERISK-20574) Reported by: Jaco Kroon patches: asterisk-rtcp-channel.patch uploaded by jkroon (License 5671) (closes issue ASTERISK-21471) Reported by: Matt Jordan git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@393740 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2013-07-05 17:33:33 +00:00
}
static void rtp_rtcp_report_dtor(void *obj)
{
int i;
struct ast_rtp_rtcp_report *rtcp_report = obj;
for (i = 0; i < rtcp_report->reception_report_count; i++) {
ast_free(rtcp_report->report_block[i]);
}
}
struct ast_rtp_rtcp_report *ast_rtp_rtcp_report_alloc(unsigned int report_blocks)
{
struct ast_rtp_rtcp_report *rtcp_report;
/* Size of object is sizeof the report + the number of report_blocks * sizeof pointer */
rtcp_report = ao2_alloc((sizeof(*rtcp_report) + report_blocks * sizeof(struct ast_rtp_rtcp_report_block *)),
rtp_rtcp_report_dtor);
return rtcp_report;
}
void ast_rtp_publish_rtcp_message(struct ast_rtp_instance *rtp,
struct stasis_message_type *message_type,
struct ast_rtp_rtcp_report *report,
struct ast_json *blob)
{
RAII_VAR(struct rtcp_message_payload *, payload, NULL, ao2_cleanup);
Refactor RTCP events over to Stasis; associate with channels This patch does the following: * It merges Jaco Kroon's patch from ASTERISK-20754, which provides channel information in the RTCP events. Because Stasis provides a cache, Jaco's patch was modified to pass the channel uniqueid to the RTP layer as opposed to a pointer to the channel. This has the following benefits: (1) It keeps the RTP engine 'clean' of references back to channels (2) It prevents circular dependencies and other potential ref counting issues * The RTP engine now allows any RTP implementation to raise RTCP messages. Potentially, other implementations (such as res_rtp_multicast) could also raise RTCP information. The engine provides structs to represent RTCP headers and RTCP SR/RR reports. * Some general refactoring in res_rtp_asterisk was done to try and tame the RTCP code. It isn't perfect - that's *way* beyond the scope of this work - but it does feel marginally better. * A few random bugs were fixed in the RTCP statistics. (Example: performing an assignment of a = a is probably not correct) * We now raise RTCP events for each SR/RR sent/received. Previously we wouldn't raise an event when we sent a RR report. Note that this work will be of use to others who want to monitor call quality or build modules that report call quality statistics. Since the events are now moving across the Stasis message bus, this is far easier to accomplish. It is also a first step (though by no means the last step) towards getting Olle's pinefrog work incorporated. Again: note that the patch by Jaco Kroon was modified slightly for this work; however, he did all of the hard work in finding the right places to set the channel in the RTP engine across the channel drivers. Much thanks goes to Jaco for his hard work here. Review: https://reviewboard.asterisk.org/r/2603/ (closes issue ASTERISK-20574) Reported by: Jaco Kroon patches: asterisk-rtcp-channel.patch uploaded by jkroon (License 5671) (closes issue ASTERISK-21471) Reported by: Matt Jordan git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@393740 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2013-07-05 17:33:33 +00:00
RAII_VAR(struct stasis_message *, message, NULL, ao2_cleanup);
if (!message_type) {
return;
}
payload = ao2_alloc(sizeof(*payload), rtcp_message_payload_dtor);
Refactor RTCP events over to Stasis; associate with channels This patch does the following: * It merges Jaco Kroon's patch from ASTERISK-20754, which provides channel information in the RTCP events. Because Stasis provides a cache, Jaco's patch was modified to pass the channel uniqueid to the RTP layer as opposed to a pointer to the channel. This has the following benefits: (1) It keeps the RTP engine 'clean' of references back to channels (2) It prevents circular dependencies and other potential ref counting issues * The RTP engine now allows any RTP implementation to raise RTCP messages. Potentially, other implementations (such as res_rtp_multicast) could also raise RTCP information. The engine provides structs to represent RTCP headers and RTCP SR/RR reports. * Some general refactoring in res_rtp_asterisk was done to try and tame the RTCP code. It isn't perfect - that's *way* beyond the scope of this work - but it does feel marginally better. * A few random bugs were fixed in the RTCP statistics. (Example: performing an assignment of a = a is probably not correct) * We now raise RTCP events for each SR/RR sent/received. Previously we wouldn't raise an event when we sent a RR report. Note that this work will be of use to others who want to monitor call quality or build modules that report call quality statistics. Since the events are now moving across the Stasis message bus, this is far easier to accomplish. It is also a first step (though by no means the last step) towards getting Olle's pinefrog work incorporated. Again: note that the patch by Jaco Kroon was modified slightly for this work; however, he did all of the hard work in finding the right places to set the channel in the RTP engine across the channel drivers. Much thanks goes to Jaco for his hard work here. Review: https://reviewboard.asterisk.org/r/2603/ (closes issue ASTERISK-20574) Reported by: Jaco Kroon patches: asterisk-rtcp-channel.patch uploaded by jkroon (License 5671) (closes issue ASTERISK-21471) Reported by: Matt Jordan git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@393740 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2013-07-05 17:33:33 +00:00
if (!payload || !report) {
return;
}
if (!ast_strlen_zero(rtp->channel_uniqueid)) {
payload->snapshot = ast_channel_snapshot_get_latest(rtp->channel_uniqueid);
Refactor RTCP events over to Stasis; associate with channels This patch does the following: * It merges Jaco Kroon's patch from ASTERISK-20754, which provides channel information in the RTCP events. Because Stasis provides a cache, Jaco's patch was modified to pass the channel uniqueid to the RTP layer as opposed to a pointer to the channel. This has the following benefits: (1) It keeps the RTP engine 'clean' of references back to channels (2) It prevents circular dependencies and other potential ref counting issues * The RTP engine now allows any RTP implementation to raise RTCP messages. Potentially, other implementations (such as res_rtp_multicast) could also raise RTCP information. The engine provides structs to represent RTCP headers and RTCP SR/RR reports. * Some general refactoring in res_rtp_asterisk was done to try and tame the RTCP code. It isn't perfect - that's *way* beyond the scope of this work - but it does feel marginally better. * A few random bugs were fixed in the RTCP statistics. (Example: performing an assignment of a = a is probably not correct) * We now raise RTCP events for each SR/RR sent/received. Previously we wouldn't raise an event when we sent a RR report. Note that this work will be of use to others who want to monitor call quality or build modules that report call quality statistics. Since the events are now moving across the Stasis message bus, this is far easier to accomplish. It is also a first step (though by no means the last step) towards getting Olle's pinefrog work incorporated. Again: note that the patch by Jaco Kroon was modified slightly for this work; however, he did all of the hard work in finding the right places to set the channel in the RTP engine across the channel drivers. Much thanks goes to Jaco for his hard work here. Review: https://reviewboard.asterisk.org/r/2603/ (closes issue ASTERISK-20574) Reported by: Jaco Kroon patches: asterisk-rtcp-channel.patch uploaded by jkroon (License 5671) (closes issue ASTERISK-21471) Reported by: Matt Jordan git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@393740 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2013-07-05 17:33:33 +00:00
}
if (blob) {
payload->blob = blob;
Refactor RTCP events over to Stasis; associate with channels This patch does the following: * It merges Jaco Kroon's patch from ASTERISK-20754, which provides channel information in the RTCP events. Because Stasis provides a cache, Jaco's patch was modified to pass the channel uniqueid to the RTP layer as opposed to a pointer to the channel. This has the following benefits: (1) It keeps the RTP engine 'clean' of references back to channels (2) It prevents circular dependencies and other potential ref counting issues * The RTP engine now allows any RTP implementation to raise RTCP messages. Potentially, other implementations (such as res_rtp_multicast) could also raise RTCP information. The engine provides structs to represent RTCP headers and RTCP SR/RR reports. * Some general refactoring in res_rtp_asterisk was done to try and tame the RTCP code. It isn't perfect - that's *way* beyond the scope of this work - but it does feel marginally better. * A few random bugs were fixed in the RTCP statistics. (Example: performing an assignment of a = a is probably not correct) * We now raise RTCP events for each SR/RR sent/received. Previously we wouldn't raise an event when we sent a RR report. Note that this work will be of use to others who want to monitor call quality or build modules that report call quality statistics. Since the events are now moving across the Stasis message bus, this is far easier to accomplish. It is also a first step (though by no means the last step) towards getting Olle's pinefrog work incorporated. Again: note that the patch by Jaco Kroon was modified slightly for this work; however, he did all of the hard work in finding the right places to set the channel in the RTP engine across the channel drivers. Much thanks goes to Jaco for his hard work here. Review: https://reviewboard.asterisk.org/r/2603/ (closes issue ASTERISK-20574) Reported by: Jaco Kroon patches: asterisk-rtcp-channel.patch uploaded by jkroon (License 5671) (closes issue ASTERISK-21471) Reported by: Matt Jordan git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@393740 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2013-07-05 17:33:33 +00:00
ast_json_ref(blob);
}
ao2_ref(report, +1);
Refactor RTCP events over to Stasis; associate with channels This patch does the following: * It merges Jaco Kroon's patch from ASTERISK-20754, which provides channel information in the RTCP events. Because Stasis provides a cache, Jaco's patch was modified to pass the channel uniqueid to the RTP layer as opposed to a pointer to the channel. This has the following benefits: (1) It keeps the RTP engine 'clean' of references back to channels (2) It prevents circular dependencies and other potential ref counting issues * The RTP engine now allows any RTP implementation to raise RTCP messages. Potentially, other implementations (such as res_rtp_multicast) could also raise RTCP information. The engine provides structs to represent RTCP headers and RTCP SR/RR reports. * Some general refactoring in res_rtp_asterisk was done to try and tame the RTCP code. It isn't perfect - that's *way* beyond the scope of this work - but it does feel marginally better. * A few random bugs were fixed in the RTCP statistics. (Example: performing an assignment of a = a is probably not correct) * We now raise RTCP events for each SR/RR sent/received. Previously we wouldn't raise an event when we sent a RR report. Note that this work will be of use to others who want to monitor call quality or build modules that report call quality statistics. Since the events are now moving across the Stasis message bus, this is far easier to accomplish. It is also a first step (though by no means the last step) towards getting Olle's pinefrog work incorporated. Again: note that the patch by Jaco Kroon was modified slightly for this work; however, he did all of the hard work in finding the right places to set the channel in the RTP engine across the channel drivers. Much thanks goes to Jaco for his hard work here. Review: https://reviewboard.asterisk.org/r/2603/ (closes issue ASTERISK-20574) Reported by: Jaco Kroon patches: asterisk-rtcp-channel.patch uploaded by jkroon (License 5671) (closes issue ASTERISK-21471) Reported by: Matt Jordan git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@393740 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2013-07-05 17:33:33 +00:00
payload->report = report;
message = stasis_message_create(message_type, payload);
if (!message) {
return;
}
stasis_publish(ast_rtp_topic(), message);
}
/*!
* @{ \brief Define RTCP/RTP message types.
*/
STASIS_MESSAGE_TYPE_DEFN(ast_rtp_rtcp_sent_type,
.to_ami = rtcp_report_to_ami,
.to_json = rtcp_report_to_json,);
STASIS_MESSAGE_TYPE_DEFN(ast_rtp_rtcp_received_type,
.to_ami = rtcp_report_to_ami,
.to_json = rtcp_report_to_json,);
/*! @} */
struct stasis_topic *ast_rtp_topic(void)
{
return rtp_topic;
}
static uintmax_t debug_category_rtp_id;
uintmax_t ast_debug_category_rtp_id(void)
{
return debug_category_rtp_id;
}
static uintmax_t debug_category_rtp_packet_id;
uintmax_t ast_debug_category_rtp_packet_id(void)
{
return debug_category_rtp_packet_id;
}
static uintmax_t debug_category_rtcp_id;
uintmax_t ast_debug_category_rtcp_id(void)
{
return debug_category_rtcp_id;
}
static uintmax_t debug_category_rtcp_packet_id;
uintmax_t ast_debug_category_rtcp_packet_id(void)
{
return debug_category_rtcp_packet_id;
}
static uintmax_t debug_category_dtls_id;
uintmax_t ast_debug_category_dtls_id(void)
{
return debug_category_dtls_id;
}
static uintmax_t debug_category_dtls_packet_id;
uintmax_t ast_debug_category_dtls_packet_id(void)
{
return debug_category_dtls_packet_id;
}
static uintmax_t debug_category_ice_id;
uintmax_t ast_debug_category_ice_id(void)
{
return debug_category_ice_id;
}
Refactor RTCP events over to Stasis; associate with channels This patch does the following: * It merges Jaco Kroon's patch from ASTERISK-20754, which provides channel information in the RTCP events. Because Stasis provides a cache, Jaco's patch was modified to pass the channel uniqueid to the RTP layer as opposed to a pointer to the channel. This has the following benefits: (1) It keeps the RTP engine 'clean' of references back to channels (2) It prevents circular dependencies and other potential ref counting issues * The RTP engine now allows any RTP implementation to raise RTCP messages. Potentially, other implementations (such as res_rtp_multicast) could also raise RTCP information. The engine provides structs to represent RTCP headers and RTCP SR/RR reports. * Some general refactoring in res_rtp_asterisk was done to try and tame the RTCP code. It isn't perfect - that's *way* beyond the scope of this work - but it does feel marginally better. * A few random bugs were fixed in the RTCP statistics. (Example: performing an assignment of a = a is probably not correct) * We now raise RTCP events for each SR/RR sent/received. Previously we wouldn't raise an event when we sent a RR report. Note that this work will be of use to others who want to monitor call quality or build modules that report call quality statistics. Since the events are now moving across the Stasis message bus, this is far easier to accomplish. It is also a first step (though by no means the last step) towards getting Olle's pinefrog work incorporated. Again: note that the patch by Jaco Kroon was modified slightly for this work; however, he did all of the hard work in finding the right places to set the channel in the RTP engine across the channel drivers. Much thanks goes to Jaco for his hard work here. Review: https://reviewboard.asterisk.org/r/2603/ (closes issue ASTERISK-20574) Reported by: Jaco Kroon patches: asterisk-rtcp-channel.patch uploaded by jkroon (License 5671) (closes issue ASTERISK-21471) Reported by: Matt Jordan git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@393740 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2013-07-05 17:33:33 +00:00
static void rtp_engine_shutdown(void)
{
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
int x;
Refactor RTCP events over to Stasis; associate with channels This patch does the following: * It merges Jaco Kroon's patch from ASTERISK-20754, which provides channel information in the RTCP events. Because Stasis provides a cache, Jaco's patch was modified to pass the channel uniqueid to the RTP layer as opposed to a pointer to the channel. This has the following benefits: (1) It keeps the RTP engine 'clean' of references back to channels (2) It prevents circular dependencies and other potential ref counting issues * The RTP engine now allows any RTP implementation to raise RTCP messages. Potentially, other implementations (such as res_rtp_multicast) could also raise RTCP information. The engine provides structs to represent RTCP headers and RTCP SR/RR reports. * Some general refactoring in res_rtp_asterisk was done to try and tame the RTCP code. It isn't perfect - that's *way* beyond the scope of this work - but it does feel marginally better. * A few random bugs were fixed in the RTCP statistics. (Example: performing an assignment of a = a is probably not correct) * We now raise RTCP events for each SR/RR sent/received. Previously we wouldn't raise an event when we sent a RR report. Note that this work will be of use to others who want to monitor call quality or build modules that report call quality statistics. Since the events are now moving across the Stasis message bus, this is far easier to accomplish. It is also a first step (though by no means the last step) towards getting Olle's pinefrog work incorporated. Again: note that the patch by Jaco Kroon was modified slightly for this work; however, he did all of the hard work in finding the right places to set the channel in the RTP engine across the channel drivers. Much thanks goes to Jaco for his hard work here. Review: https://reviewboard.asterisk.org/r/2603/ (closes issue ASTERISK-20574) Reported by: Jaco Kroon patches: asterisk-rtcp-channel.patch uploaded by jkroon (License 5671) (closes issue ASTERISK-21471) Reported by: Matt Jordan git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@393740 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2013-07-05 17:33:33 +00:00
ao2_cleanup(rtp_topic);
rtp_topic = NULL;
STASIS_MESSAGE_TYPE_CLEANUP(ast_rtp_rtcp_received_type);
STASIS_MESSAGE_TYPE_CLEANUP(ast_rtp_rtcp_sent_type);
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
ast_rwlock_wrlock(&static_RTP_PT_lock);
for (x = 0; x < AST_RTP_MAX_PT; x++) {
ao2_cleanup(static_RTP_PT[x]);
static_RTP_PT[x] = NULL;
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
}
ast_rwlock_unlock(&static_RTP_PT_lock);
ast_rwlock_wrlock(&mime_types_lock);
for (x = 0; x < mime_types_len; x++) {
if (ast_rtp_mime_types[x].payload_type.format) {
rtp_engine_mime_type_cleanup(x);
}
}
mime_types_len = 0;
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
ast_rwlock_unlock(&mime_types_lock);
ast_debug_category_unregister(AST_LOG_CATEGORY_ICE);
ast_debug_category_unregister(AST_LOG_CATEGORY_DTLS_PACKET);
ast_debug_category_unregister(AST_LOG_CATEGORY_DTLS);
ast_debug_category_unregister(AST_LOG_CATEGORY_RTCP_PACKET);
ast_debug_category_unregister(AST_LOG_CATEGORY_RTCP);
ast_debug_category_unregister(AST_LOG_CATEGORY_RTP_PACKET);
ast_debug_category_unregister(AST_LOG_CATEGORY_RTP);
Refactor RTCP events over to Stasis; associate with channels This patch does the following: * It merges Jaco Kroon's patch from ASTERISK-20754, which provides channel information in the RTCP events. Because Stasis provides a cache, Jaco's patch was modified to pass the channel uniqueid to the RTP layer as opposed to a pointer to the channel. This has the following benefits: (1) It keeps the RTP engine 'clean' of references back to channels (2) It prevents circular dependencies and other potential ref counting issues * The RTP engine now allows any RTP implementation to raise RTCP messages. Potentially, other implementations (such as res_rtp_multicast) could also raise RTCP information. The engine provides structs to represent RTCP headers and RTCP SR/RR reports. * Some general refactoring in res_rtp_asterisk was done to try and tame the RTCP code. It isn't perfect - that's *way* beyond the scope of this work - but it does feel marginally better. * A few random bugs were fixed in the RTCP statistics. (Example: performing an assignment of a = a is probably not correct) * We now raise RTCP events for each SR/RR sent/received. Previously we wouldn't raise an event when we sent a RR report. Note that this work will be of use to others who want to monitor call quality or build modules that report call quality statistics. Since the events are now moving across the Stasis message bus, this is far easier to accomplish. It is also a first step (though by no means the last step) towards getting Olle's pinefrog work incorporated. Again: note that the patch by Jaco Kroon was modified slightly for this work; however, he did all of the hard work in finding the right places to set the channel in the RTP engine across the channel drivers. Much thanks goes to Jaco for his hard work here. Review: https://reviewboard.asterisk.org/r/2603/ (closes issue ASTERISK-20574) Reported by: Jaco Kroon patches: asterisk-rtcp-channel.patch uploaded by jkroon (License 5671) (closes issue ASTERISK-21471) Reported by: Matt Jordan git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@393740 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2013-07-05 17:33:33 +00:00
}
int ast_rtp_engine_init(void)
{
ast_rwlock_init(&mime_types_lock);
ast_rwlock_init(&static_RTP_PT_lock);
rtp_topic = stasis_topic_create("rtp:all");
Refactor RTCP events over to Stasis; associate with channels This patch does the following: * It merges Jaco Kroon's patch from ASTERISK-20754, which provides channel information in the RTCP events. Because Stasis provides a cache, Jaco's patch was modified to pass the channel uniqueid to the RTP layer as opposed to a pointer to the channel. This has the following benefits: (1) It keeps the RTP engine 'clean' of references back to channels (2) It prevents circular dependencies and other potential ref counting issues * The RTP engine now allows any RTP implementation to raise RTCP messages. Potentially, other implementations (such as res_rtp_multicast) could also raise RTCP information. The engine provides structs to represent RTCP headers and RTCP SR/RR reports. * Some general refactoring in res_rtp_asterisk was done to try and tame the RTCP code. It isn't perfect - that's *way* beyond the scope of this work - but it does feel marginally better. * A few random bugs were fixed in the RTCP statistics. (Example: performing an assignment of a = a is probably not correct) * We now raise RTCP events for each SR/RR sent/received. Previously we wouldn't raise an event when we sent a RR report. Note that this work will be of use to others who want to monitor call quality or build modules that report call quality statistics. Since the events are now moving across the Stasis message bus, this is far easier to accomplish. It is also a first step (though by no means the last step) towards getting Olle's pinefrog work incorporated. Again: note that the patch by Jaco Kroon was modified slightly for this work; however, he did all of the hard work in finding the right places to set the channel in the RTP engine across the channel drivers. Much thanks goes to Jaco for his hard work here. Review: https://reviewboard.asterisk.org/r/2603/ (closes issue ASTERISK-20574) Reported by: Jaco Kroon patches: asterisk-rtcp-channel.patch uploaded by jkroon (License 5671) (closes issue ASTERISK-21471) Reported by: Matt Jordan git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@393740 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2013-07-05 17:33:33 +00:00
if (!rtp_topic) {
return -1;
}
STASIS_MESSAGE_TYPE_INIT(ast_rtp_rtcp_sent_type);
STASIS_MESSAGE_TYPE_INIT(ast_rtp_rtcp_received_type);
ast_register_cleanup(rtp_engine_shutdown);
Refactor RTCP events over to Stasis; associate with channels This patch does the following: * It merges Jaco Kroon's patch from ASTERISK-20754, which provides channel information in the RTCP events. Because Stasis provides a cache, Jaco's patch was modified to pass the channel uniqueid to the RTP layer as opposed to a pointer to the channel. This has the following benefits: (1) It keeps the RTP engine 'clean' of references back to channels (2) It prevents circular dependencies and other potential ref counting issues * The RTP engine now allows any RTP implementation to raise RTCP messages. Potentially, other implementations (such as res_rtp_multicast) could also raise RTCP information. The engine provides structs to represent RTCP headers and RTCP SR/RR reports. * Some general refactoring in res_rtp_asterisk was done to try and tame the RTCP code. It isn't perfect - that's *way* beyond the scope of this work - but it does feel marginally better. * A few random bugs were fixed in the RTCP statistics. (Example: performing an assignment of a = a is probably not correct) * We now raise RTCP events for each SR/RR sent/received. Previously we wouldn't raise an event when we sent a RR report. Note that this work will be of use to others who want to monitor call quality or build modules that report call quality statistics. Since the events are now moving across the Stasis message bus, this is far easier to accomplish. It is also a first step (though by no means the last step) towards getting Olle's pinefrog work incorporated. Again: note that the patch by Jaco Kroon was modified slightly for this work; however, he did all of the hard work in finding the right places to set the channel in the RTP engine across the channel drivers. Much thanks goes to Jaco for his hard work here. Review: https://reviewboard.asterisk.org/r/2603/ (closes issue ASTERISK-20574) Reported by: Jaco Kroon patches: asterisk-rtcp-channel.patch uploaded by jkroon (License 5671) (closes issue ASTERISK-21471) Reported by: Matt Jordan git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@393740 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2013-07-05 17:33:33 +00:00
/* Define all the RTP mime types available */
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
set_next_mime_type(ast_format_g723, 0, "audio", "G723", 8000);
set_next_mime_type(ast_format_gsm, 0, "audio", "GSM", 8000);
set_next_mime_type(ast_format_ulaw, 0, "audio", "PCMU", 8000);
set_next_mime_type(ast_format_ulaw, 0, "audio", "G711U", 8000);
set_next_mime_type(ast_format_alaw, 0, "audio", "PCMA", 8000);
set_next_mime_type(ast_format_alaw, 0, "audio", "G711A", 8000);
set_next_mime_type(ast_format_g726, 0, "audio", "G726-32", 8000);
set_next_mime_type(ast_format_adpcm, 0, "audio", "DVI4", 8000);
set_next_mime_type(ast_format_slin, 0, "audio", "L16", 8000);
set_next_mime_type(ast_format_slin16, 0, "audio", "L16", 16000);
set_next_mime_type(ast_format_slin16, 0, "audio", "L16-256", 16000);
set_next_mime_type(ast_format_slin12, 0, "audio", "L16", 12000);
set_next_mime_type(ast_format_slin24, 0, "audio", "L16", 24000);
set_next_mime_type(ast_format_slin32, 0, "audio", "L16", 32000);
set_next_mime_type(ast_format_slin44, 0, "audio", "L16", 44000);
set_next_mime_type(ast_format_slin48, 0, "audio", "L16", 48000);
set_next_mime_type(ast_format_slin96, 0, "audio", "L16", 96000);
set_next_mime_type(ast_format_slin192, 0, "audio", "L16", 192000);
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
set_next_mime_type(ast_format_lpc10, 0, "audio", "LPC", 8000);
set_next_mime_type(ast_format_g729, 0, "audio", "G729", 8000);
set_next_mime_type(ast_format_g729, 0, "audio", "G729A", 8000);
set_next_mime_type(ast_format_g729, 0, "audio", "G.729", 8000);
set_next_mime_type(ast_format_speex, 0, "audio", "speex", 8000);
set_next_mime_type(ast_format_speex16, 0, "audio", "speex", 16000);
set_next_mime_type(ast_format_speex32, 0, "audio", "speex", 32000);
set_next_mime_type(ast_format_ilbc, 0, "audio", "iLBC", 8000);
/* this is the sample rate listed in the RTP profile for the G.722 codec, *NOT* the actual sample rate of the media stream */
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
set_next_mime_type(ast_format_g722, 0, "audio", "G722", 8000);
set_next_mime_type(ast_format_g726_aal2, 0, "audio", "AAL2-G726-32", 8000);
set_next_mime_type(NULL, AST_RTP_DTMF, "audio", "telephone-event", 8000);
set_next_mime_type(NULL, AST_RTP_CISCO_DTMF, "audio", "cisco-telephone-event", 8000);
set_next_mime_type(NULL, AST_RTP_CN, "audio", "CN", 8000);
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
set_next_mime_type(ast_format_jpeg, 0, "video", "JPEG", 90000);
set_next_mime_type(ast_format_png, 0, "video", "PNG", 90000);
set_next_mime_type(ast_format_h261, 0, "video", "H261", 90000);
set_next_mime_type(ast_format_h263, 0, "video", "H263", 90000);
set_next_mime_type(ast_format_h263p, 0, "video", "h263-1998", 90000);
set_next_mime_type(ast_format_h264, 0, "video", "H264", 90000);
set_next_mime_type(ast_format_h265, 0, "video", "H265", 90000);
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
set_next_mime_type(ast_format_mp4, 0, "video", "MP4V-ES", 90000);
set_next_mime_type(ast_format_t140_red, 0, "text", "RED", 1000);
set_next_mime_type(ast_format_t140, 0, "text", "T140", 1000);
set_next_mime_type(ast_format_siren7, 0, "audio", "G7221", 16000);
set_next_mime_type(ast_format_siren14, 0, "audio", "G7221", 32000);
set_next_mime_type(ast_format_g719, 0, "audio", "G719", 48000);
/* Opus, VP8, and VP9 */
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
set_next_mime_type(ast_format_opus, 0, "audio", "opus", 48000);
set_next_mime_type(ast_format_vp8, 0, "video", "VP8", 90000);
set_next_mime_type(ast_format_vp9, 0, "video", "VP9", 90000);
/* Define the static rtp payload mappings */
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
add_static_payload(0, ast_format_ulaw, 0);
#ifdef USE_DEPRECATED_G726
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
add_static_payload(2, ast_format_g726, 0);/* Technically this is G.721, but if Cisco can do it, so can we... */
#endif
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
add_static_payload(3, ast_format_gsm, 0);
add_static_payload(4, ast_format_g723, 0);
add_static_payload(5, ast_format_adpcm, 0);/* 8 kHz */
add_static_payload(6, ast_format_adpcm, 0); /* 16 kHz */
add_static_payload(7, ast_format_lpc10, 0);
add_static_payload(8, ast_format_alaw, 0);
add_static_payload(9, ast_format_g722, 0);
add_static_payload(10, ast_format_slin, 0); /* 2 channels */
add_static_payload(11, ast_format_slin, 0); /* 1 channel */
add_static_payload(13, NULL, AST_RTP_CN);
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
add_static_payload(16, ast_format_adpcm, 0); /* 11.025 kHz */
add_static_payload(17, ast_format_adpcm, 0); /* 22.050 kHz */
add_static_payload(18, ast_format_g729, 0);
add_static_payload(19, NULL, AST_RTP_CN); /* Also used for CN */
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
add_static_payload(26, ast_format_jpeg, 0);
add_static_payload(31, ast_format_h261, 0);
add_static_payload(34, ast_format_h263, 0);
/*
* Dynamic payload types - Even when dynamically assigning them we'll fall
* back to using the statically declared values as the default number.
*/
add_static_payload(96, ast_format_slin192, 0);
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
add_static_payload(97, ast_format_ilbc, 0);
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
add_static_payload(99, ast_format_h264, 0);
add_static_payload(100, ast_format_vp8, 0);
add_static_payload(101, NULL, AST_RTP_DTMF);
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
add_static_payload(102, ast_format_siren7, 0);
add_static_payload(103, ast_format_h263p, 0);
add_static_payload(104, ast_format_mp4, 0);
add_static_payload(105, ast_format_t140_red, 0); /* Real time text chat (with redundancy encoding) */
add_static_payload(106, ast_format_t140, 0); /* Real time text chat */
add_static_payload(107, ast_format_opus, 0);
add_static_payload(108, ast_format_vp9, 0);
add_static_payload(109, ast_format_h265, 0);
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
add_static_payload(110, ast_format_speex, 0);
add_static_payload(111, ast_format_g726, 0);
add_static_payload(112, ast_format_g726_aal2, 0);
media formats: re-architect handling of media for performance improvements In the old times media formats were represented using a bit field. This was fast but had a few limitations. 1. Asterisk was limited in how many formats it could handle. 2. Formats, being a bit field, could not include any attribute information. A format was strictly its type, e.g., "this is ulaw". This was changed in Asterisk 10 (see https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for notes on that work) which led to the creation of the ast_format structure. This structure allowed Asterisk to handle attributes and bundle information with a format. Additionally, ast_format_cap was created to act as a container for multiple formats that, together, formed the capability of some entity. Another mechanism was added to allow logic to be registered which performed format attribute negotiation. Everywhere throughout the codebase Asterisk was changed to use this strategy. Unfortunately, in software, there is no free lunch. These new capabilities came at a cost. Performance analysis and profiling showed that we spend an inordinate amount of time comparing, copying, and generally manipulating formats and their related structures. Basic prototyping has shown that a reasonably large performance improvement could be made in this area. This patch is the result of that project, which overhauled the media format architecture and its usage in Asterisk to improve performance. Generally, the new philosophy for handling formats is as follows: * The ast_format structure is reference counted. This removed a large amount of the memory allocations and copying that was done in prior versions. * In order to prevent race conditions while keeping things performant, the ast_format structure is immutable by convention and lock-free. Violate this tenet at your peril! * Because formats are reference counted, codecs are also reference counted. The Asterisk core generally provides built-in codecs and caches the ast_format structures created to represent them. Generally, to prevent inordinate amounts of module reference bumping, codecs and formats can be added at run-time but cannot be removed. * All compatibility with the bit field representation of codecs/formats has been moved to a compatibility API. The primary user of this representation is chan_iax2, which must continue to maintain its bit-field usage of formats for interoperability concerns. * When a format is negotiated with attributes, or when a format cannot be represented by one of the cached formats, a new format object is created or cloned from an existing format. That format may have the same codec underlying it, but is a different format than a version of the format with different attributes or without attributes. * While formats are reference counted objects, the reference count maintained on the format should be manipulated with care. Formats are generally cached and will persist for the lifetime of Asterisk and do not explicitly need to have their lifetime modified. An exception to this is when the user of a format does not know where the format came from *and* the user may outlive the provider of the format. This occurs, for example, when a format is read from a channel: the channel may have a format with attributes (hence, non-cached) and the user of the format may last longer than the channel (if the reference to the channel is released prior to the format's reference). For more information on this work, see the API design notes: https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite Finally, this work was the culmination of a large number of developer's efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the work in the Asterisk core, chan_sip, and was an invaluable resource in peer reviews throughout this project. There were a substantial number of patches contributed during this work; the following issues/patch names simply reflect some of the work (and will cause the release scripts to give attribution to the individuals who work on them). Reviews: https://reviewboard.asterisk.org/r/3814 https://reviewboard.asterisk.org/r/3808 https://reviewboard.asterisk.org/r/3805 https://reviewboard.asterisk.org/r/3803 https://reviewboard.asterisk.org/r/3801 https://reviewboard.asterisk.org/r/3798 https://reviewboard.asterisk.org/r/3800 https://reviewboard.asterisk.org/r/3794 https://reviewboard.asterisk.org/r/3793 https://reviewboard.asterisk.org/r/3792 https://reviewboard.asterisk.org/r/3791 https://reviewboard.asterisk.org/r/3790 https://reviewboard.asterisk.org/r/3789 https://reviewboard.asterisk.org/r/3788 https://reviewboard.asterisk.org/r/3787 https://reviewboard.asterisk.org/r/3786 https://reviewboard.asterisk.org/r/3784 https://reviewboard.asterisk.org/r/3783 https://reviewboard.asterisk.org/r/3778 https://reviewboard.asterisk.org/r/3774 https://reviewboard.asterisk.org/r/3775 https://reviewboard.asterisk.org/r/3772 https://reviewboard.asterisk.org/r/3761 https://reviewboard.asterisk.org/r/3754 https://reviewboard.asterisk.org/r/3753 https://reviewboard.asterisk.org/r/3751 https://reviewboard.asterisk.org/r/3750 https://reviewboard.asterisk.org/r/3748 https://reviewboard.asterisk.org/r/3747 https://reviewboard.asterisk.org/r/3746 https://reviewboard.asterisk.org/r/3742 https://reviewboard.asterisk.org/r/3740 https://reviewboard.asterisk.org/r/3739 https://reviewboard.asterisk.org/r/3738 https://reviewboard.asterisk.org/r/3737 https://reviewboard.asterisk.org/r/3736 https://reviewboard.asterisk.org/r/3734 https://reviewboard.asterisk.org/r/3722 https://reviewboard.asterisk.org/r/3713 https://reviewboard.asterisk.org/r/3703 https://reviewboard.asterisk.org/r/3689 https://reviewboard.asterisk.org/r/3687 https://reviewboard.asterisk.org/r/3674 https://reviewboard.asterisk.org/r/3671 https://reviewboard.asterisk.org/r/3667 https://reviewboard.asterisk.org/r/3665 https://reviewboard.asterisk.org/r/3625 https://reviewboard.asterisk.org/r/3602 https://reviewboard.asterisk.org/r/3519 https://reviewboard.asterisk.org/r/3518 https://reviewboard.asterisk.org/r/3516 https://reviewboard.asterisk.org/r/3515 https://reviewboard.asterisk.org/r/3512 https://reviewboard.asterisk.org/r/3506 https://reviewboard.asterisk.org/r/3413 https://reviewboard.asterisk.org/r/3410 https://reviewboard.asterisk.org/r/3387 https://reviewboard.asterisk.org/r/3388 https://reviewboard.asterisk.org/r/3389 https://reviewboard.asterisk.org/r/3390 https://reviewboard.asterisk.org/r/3321 https://reviewboard.asterisk.org/r/3320 https://reviewboard.asterisk.org/r/3319 https://reviewboard.asterisk.org/r/3318 https://reviewboard.asterisk.org/r/3266 https://reviewboard.asterisk.org/r/3265 https://reviewboard.asterisk.org/r/3234 https://reviewboard.asterisk.org/r/3178 ASTERISK-23114 #close Reported by: mjordan media_formats_translation_core.diff uploaded by kharwell (License 6464) rb3506.diff uploaded by mjordan (License 6283) media_format_app_file.diff uploaded by kharwell (License 6464) misc-2.diff uploaded by file (License 5000) chan_mild-3.diff uploaded by file (License 5000) chan_obscure.diff uploaded by file (License 5000) jingle.diff uploaded by file (License 5000) funcs.diff uploaded by file (License 5000) formats.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) bridges.diff uploaded by file (License 5000) mf-codecs-2.diff uploaded by file (License 5000) mf-app_fax.diff uploaded by file (License 5000) mf-apps-3.diff uploaded by file (License 5000) media-formats-3.diff uploaded by file (License 5000) ASTERISK-23715 rb3713.patch uploaded by coreyfarrell (License 5909) rb3689.patch uploaded by mjordan (License 6283) ASTERISK-23957 rb3722.patch uploaded by mjordan (License 6283) mf-attributes-3.diff uploaded by file (License 5000) ASTERISK-23958 Tested by: jrose rb3822.patch uploaded by coreyfarrell (License 5909) rb3800.patch uploaded by jrose (License 6182) chan_sip.diff uploaded by mjordan (License 6283) rb3747.patch uploaded by jrose (License 6182) ASTERISK-23959 #close Tested by: sgriepentrog, mjordan, coreyfarrell sip_cleanup.diff uploaded by opticron (License 6273) chan_sip_caps.diff uploaded by mjordan (License 6283) rb3751.patch uploaded by coreyfarrell (License 5909) chan_sip-3.diff uploaded by file (License 5000) ASTERISK-23960 #close Tested by: opticron direct_media.diff uploaded by opticron (License 6273) pjsip-direct-media.diff uploaded by file (License 5000) format_cap_remove.diff uploaded by opticron (License 6273) media_format_fixes.diff uploaded by opticron (License 6273) chan_pjsip-2.diff uploaded by file (License 5000) ASTERISK-23966 #close Tested by: rmudgett rb3803.patch uploaded by rmudgetti (License 5621) chan_dahdi.diff uploaded by file (License 5000) ASTERISK-24064 #close Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose rb3814.patch uploaded by rmudgett (License 5621) moh_cleanup.diff uploaded by opticron (License 6273) bridge_leak.diff uploaded by opticron (License 6273) translate.diff uploaded by file (License 5000) rb3795.patch uploaded by rmudgett (License 5621) tls_fix.diff uploaded by mjordan (License 6283) fax-mf-fix-2.diff uploaded by file (License 5000) rtp_transfer_stuff uploaded by mjordan (License 6283) rb3787.patch uploaded by rmudgett (License 5621) media-formats-explicit-translate-format-3.diff uploaded by file (License 5000) format_cache_case_fix.diff uploaded by opticron (License 6273) rb3774.patch uploaded by rmudgett (License 5621) rb3775.patch uploaded by rmudgett (License 5621) rtp_engine_fix.diff uploaded by opticron (License 6273) rtp_crash_fix.diff uploaded by opticron (License 6273) rb3753.patch uploaded by mjordan (License 6283) rb3750.patch uploaded by mjordan (License 6283) rb3748.patch uploaded by rmudgett (License 5621) media_format_fixes.diff uploaded by opticron (License 6273) rb3740.patch uploaded by mjordan (License 6283) rb3739.patch uploaded by mjordan (License 6283) rb3734.patch uploaded by mjordan (License 6283) rb3689.patch uploaded by mjordan (License 6283) rb3674.patch uploaded by coreyfarrell (License 5909) rb3671.patch uploaded by coreyfarrell (License 5909) rb3667.patch uploaded by coreyfarrell (License 5909) rb3665.patch uploaded by mjordan (License 6283) rb3625.patch uploaded by coreyfarrell (License 5909) rb3602.patch uploaded by coreyfarrell (License 5909) format_compatibility-2.diff uploaded by file (License 5000) core.diff uploaded by file (License 5000) git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-07-20 22:06:33 +00:00
add_static_payload(115, ast_format_siren14, 0);
add_static_payload(116, ast_format_g719, 0);
add_static_payload(117, ast_format_speex16, 0);
add_static_payload(118, ast_format_slin16, 0); /* 16 Khz signed linear */
add_static_payload(119, ast_format_speex32, 0);
add_static_payload(121, NULL, AST_RTP_CISCO_DTMF); /* Must be type 121 */
add_static_payload(122, ast_format_slin12, 0);
add_static_payload(123, ast_format_slin24, 0);
add_static_payload(124, ast_format_slin32, 0);
add_static_payload(125, ast_format_slin44, 0);
add_static_payload(126, ast_format_slin48, 0);
add_static_payload(127, ast_format_slin96, 0);
/* payload types above 127 are not valid */
debug_category_rtp_id = ast_debug_category_register(AST_LOG_CATEGORY_RTP);
debug_category_rtp_packet_id = ast_debug_category_register(AST_LOG_CATEGORY_RTP_PACKET);
debug_category_rtcp_id = ast_debug_category_register(AST_LOG_CATEGORY_RTCP);
debug_category_rtcp_packet_id = ast_debug_category_register(AST_LOG_CATEGORY_RTCP_PACKET);
debug_category_dtls_id = ast_debug_category_register(AST_LOG_CATEGORY_DTLS);
debug_category_dtls_packet_id = ast_debug_category_register(AST_LOG_CATEGORY_DTLS_PACKET);
debug_category_ice_id = ast_debug_category_register(AST_LOG_CATEGORY_ICE);
return 0;
}
time_t ast_rtp_instance_get_last_tx(const struct ast_rtp_instance *rtp)
{
return rtp->last_tx;
}
void ast_rtp_instance_set_last_tx(struct ast_rtp_instance *rtp, time_t time)
{
rtp->last_tx = time;
}
time_t ast_rtp_instance_get_last_rx(const struct ast_rtp_instance *rtp)
{
return rtp->last_rx;
}
void ast_rtp_instance_set_last_rx(struct ast_rtp_instance *rtp, time_t time)
{
rtp->last_rx = time;
}
unsigned int ast_rtp_instance_get_ssrc(struct ast_rtp_instance *rtp)
{
unsigned int ssrc = 0;
ao2_lock(rtp);
if (rtp->engine->ssrc_get) {
ssrc = rtp->engine->ssrc_get(rtp);
}
ao2_unlock(rtp);
return ssrc;
}
const char *ast_rtp_instance_get_cname(struct ast_rtp_instance *rtp)
{
const char *cname = "";
ao2_lock(rtp);
if (rtp->engine->cname_get) {
cname = rtp->engine->cname_get(rtp);
}
ao2_unlock(rtp);
return cname;
}
int ast_rtp_instance_bundle(struct ast_rtp_instance *child, struct ast_rtp_instance *parent)
{
int res = -1;
if (parent && (child->engine != parent->engine)) {
return -1;
}
ao2_lock(child);
if (child->engine->bundle) {
res = child->engine->bundle(child, parent);
}
ao2_unlock(child);
return res;
}
void ast_rtp_instance_set_remote_ssrc(struct ast_rtp_instance *rtp, unsigned int ssrc)
{
ao2_lock(rtp);
if (rtp->engine->set_remote_ssrc) {
rtp->engine->set_remote_ssrc(rtp, ssrc);
}
ao2_unlock(rtp);
}
void ast_rtp_instance_set_stream_num(struct ast_rtp_instance *rtp, int stream_num)
{
ao2_lock(rtp);
if (rtp->engine->set_stream_num) {
rtp->engine->set_stream_num(rtp, stream_num);
}
ao2_unlock(rtp);
}
#ifdef TEST_FRAMEWORK
size_t ast_rtp_instance_get_recv_buffer_max(struct ast_rtp_instance *instance)
{
size_t res;
struct ast_rtp_engine_test *test = ast_rtp_instance_get_test(instance);
if (!test) {
ast_log(LOG_ERROR, "There is no test engine set up!\n");
return 0;
}
ao2_lock(instance);
res = test->recv_buffer_max(instance);
ao2_unlock(instance);
return res;
}
size_t ast_rtp_instance_get_recv_buffer_count(struct ast_rtp_instance *instance)
{
size_t res;
struct ast_rtp_engine_test *test = ast_rtp_instance_get_test(instance);
if (!test) {
ast_log(LOG_ERROR, "There is no test engine set up!\n");
return 0;
}
ao2_lock(instance);
res = test->recv_buffer_count(instance);
ao2_unlock(instance);
return res;
}
size_t ast_rtp_instance_get_send_buffer_count(struct ast_rtp_instance *instance)
{
size_t res;
struct ast_rtp_engine_test *test = ast_rtp_instance_get_test(instance);
if (!test) {
ast_log(LOG_ERROR, "There is no test engine set up!\n");
return 0;
}
ao2_lock(instance);
res = test->send_buffer_count(instance);
ao2_unlock(instance);
return res;
}
void ast_rtp_instance_set_schedid(struct ast_rtp_instance *instance, int id)
{
struct ast_rtp_engine_test *test = ast_rtp_instance_get_test(instance);
if (!test) {
ast_log(LOG_ERROR, "There is no test engine set up!\n");
return;
}
ao2_lock(instance);
test->set_schedid(instance, id);
ao2_unlock(instance);
}
void ast_rtp_instance_drop_packets(struct ast_rtp_instance *instance, int num)
{
struct ast_rtp_engine_test *test = ast_rtp_instance_get_test(instance);
if (!test) {
ast_log(LOG_ERROR, "There is no test engine set up!\n");
return;
}
test->packets_to_drop = num;
}
void ast_rtp_instance_queue_report(struct ast_rtp_instance *instance)
{
struct ast_rtp_engine_test *test = ast_rtp_instance_get_test(instance);
if (!test) {
ast_log(LOG_ERROR, "There is no test engine set up!\n");
return;
}
test->send_report = 1;
}
int ast_rtp_instance_get_sdes_received(struct ast_rtp_instance *instance)
{
struct ast_rtp_engine_test *test = ast_rtp_instance_get_test(instance);
if (!test) {
ast_log(LOG_ERROR, "There is no test engine set up!\n");
return 0;
}
return test->sdes_received;
}
void ast_rtp_instance_reset_test_engine(struct ast_rtp_instance *instance)
{
struct ast_rtp_engine_test *test = ast_rtp_instance_get_test(instance);
if (!test) {
ast_log(LOG_ERROR, "There is no test engine set up!\n");
return;
}
test->packets_to_drop = 0;
test->send_report = 0;
test->sdes_received = 0;
}
#endif
struct ast_json *ast_rtp_convert_stats_json(const struct ast_rtp_instance_stats *stats)
{
struct ast_json *j_res;
int ret;
j_res = ast_json_object_create();
if (!j_res) {
return NULL;
}
/* set mandatory items */
ret = ast_json_object_set(j_res, "txcount", ast_json_integer_create(stats->txcount));
ret |= ast_json_object_set(j_res, "rxcount", ast_json_integer_create(stats->rxcount));
ret |= ast_json_object_set(j_res, "txploss", ast_json_integer_create(stats->txploss));
ret |= ast_json_object_set(j_res, "rxploss", ast_json_integer_create(stats->rxploss));
ret |= ast_json_object_set(j_res, "local_ssrc", ast_json_integer_create(stats->local_ssrc));
ret |= ast_json_object_set(j_res, "remote_ssrc", ast_json_integer_create(stats->remote_ssrc));
ret |= ast_json_object_set(j_res, "txoctetcount", ast_json_integer_create(stats->txoctetcount));
ret |= ast_json_object_set(j_res, "rxoctetcount", ast_json_integer_create(stats->rxoctetcount));
ret |= ast_json_object_set(j_res, "channel_uniqueid", ast_json_string_create(stats->channel_uniqueid));
if (ret) {
ast_log(LOG_WARNING, "Could not create rtp statistics info. channel: %s\n", stats->channel_uniqueid);
ast_json_unref(j_res);
return NULL;
}
/* set other items */
SET_AST_JSON_OBJ(j_res, "txjitter", ast_json_real_create(stats->txjitter));
SET_AST_JSON_OBJ(j_res, "rxjitter", ast_json_real_create(stats->rxjitter));
SET_AST_JSON_OBJ(j_res, "remote_maxjitter", ast_json_real_create(stats->remote_maxjitter));
SET_AST_JSON_OBJ(j_res, "remote_minjitter", ast_json_real_create(stats->remote_minjitter));
SET_AST_JSON_OBJ(j_res, "remote_normdevjitter", ast_json_real_create(stats->remote_normdevjitter));
SET_AST_JSON_OBJ(j_res, "remote_stdevjitter", ast_json_real_create(stats->remote_stdevjitter));
SET_AST_JSON_OBJ(j_res, "local_maxjitter", ast_json_real_create(stats->local_maxjitter));
SET_AST_JSON_OBJ(j_res, "local_minjitter", ast_json_real_create(stats->local_minjitter));
SET_AST_JSON_OBJ(j_res, "local_normdevjitter", ast_json_real_create(stats->local_normdevjitter));
SET_AST_JSON_OBJ(j_res, "local_stdevjitter", ast_json_real_create(stats->local_stdevjitter));
SET_AST_JSON_OBJ(j_res, "remote_maxrxploss", ast_json_real_create(stats->remote_maxrxploss));
SET_AST_JSON_OBJ(j_res, "remote_minrxploss", ast_json_real_create(stats->remote_minrxploss));
SET_AST_JSON_OBJ(j_res, "remote_normdevrxploss", ast_json_real_create(stats->remote_normdevrxploss));
SET_AST_JSON_OBJ(j_res, "remote_stdevrxploss", ast_json_real_create(stats->remote_stdevrxploss));
SET_AST_JSON_OBJ(j_res, "local_maxrxploss", ast_json_real_create(stats->local_maxrxploss));
SET_AST_JSON_OBJ(j_res, "local_minrxploss", ast_json_real_create(stats->local_minrxploss));
SET_AST_JSON_OBJ(j_res, "local_normdevrxploss", ast_json_real_create(stats->local_normdevrxploss));
SET_AST_JSON_OBJ(j_res, "local_stdevrxploss", ast_json_real_create(stats->local_stdevrxploss));
SET_AST_JSON_OBJ(j_res, "rtt", ast_json_real_create(stats->rtt));
SET_AST_JSON_OBJ(j_res, "maxrtt", ast_json_real_create(stats->maxrtt));
SET_AST_JSON_OBJ(j_res, "minrtt", ast_json_real_create(stats->minrtt));
SET_AST_JSON_OBJ(j_res, "normdevrtt", ast_json_real_create(stats->normdevrtt));
SET_AST_JSON_OBJ(j_res, "stdevrtt", ast_json_real_create(stats->stdevrtt));
return j_res;
}
struct ast_json *ast_rtp_instance_get_stats_all_json(struct ast_rtp_instance *instance)
{
struct ast_rtp_instance_stats stats = {0,};
if(ast_rtp_instance_get_stats(instance, &stats, AST_RTP_INSTANCE_STAT_ALL)) {
return NULL;
}
return ast_rtp_convert_stats_json(&stats);
}
int ast_rtp_get_rate(const struct ast_format *format)
{
/* For those wondering: due to a fluke in RFC publication, G.722 is advertised
* as having a sample rate of 8kHz, while implementations must know that its
* real rate is 16kHz. Seriously.
*/
return (ast_format_cmp(format, ast_format_g722) == AST_FORMAT_CMP_EQUAL) ? 8000 : (int)ast_format_get_sample_rate(format);
}