asterisk/res/res_fax.c

4867 lines
156 KiB
C
Raw Normal View History

/*
* Asterisk -- An open source telephony toolkit.
*
* Copyright (C) 2008-2009, Digium, Inc.
*
* Dwayne M. Hubbard <dhubbard@digium.com>
* Kevin P. Fleming <kpfleming@digium.com>
* Matthew Nicholson <mnicholson@digium.com>
*
* Initial T.38-gateway code
* 2008, Daniel Ferenci <daniel.ferenci@nethemba.com>
* Created by Nethemba s.r.o. http://www.nethemba.com
* Sponsored by IPEX a.s. http://www.ipex.cz
*
* T.38-gateway integration into asterisk app_fax and rework
* 2008-2011, Gregory Hinton Nietsky <gregory@distrotech.co.za>
* dns Telecom http://www.dnstelecom.co.za
*
* Modified to make T.38-gateway compatible with Asterisk 1.6.2
* 2010, Anton Verevkin <mymail@verevkin.it>
* ViaNetTV http://www.vianettv.com
*
* Modified to make T.38-gateway work
* 2010, Klaus Darilion, IPCom GmbH, www.ipcom.at
*
* 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.
*/
/*** MODULEINFO
<support_level>core</support_level>
***/
/*! \file
*
* \brief Generic FAX Resource for FAX technology resource modules
*
* \author Dwayne M. Hubbard <dhubbard@digium.com>
* \author Kevin P. Fleming <kpfleming@digium.com>
* \author Matthew Nicholson <mnicholson@digium.com>
* \author Gregory H. Nietsky <gregory@distrotech.co.za>
*
* A generic FAX resource module that provides SendFAX and ReceiveFAX applications.
* This module requires FAX technology modules, like res_fax_spandsp, to register with it
* so it can use the technology modules to perform the actual FAX transmissions.
* \ingroup applications
*/
/*! \li \ref res_fax.c uses the configuration file \ref res_fax.conf
* \addtogroup configuration_file Configuration Files
*/
/*!
* \page res_fax.conf res_fax.conf
* \verbinclude res_fax.conf.sample
*/
#include "asterisk.h"
#include "asterisk/io.h"
#include "asterisk/file.h"
#include "asterisk/logger.h"
#include "asterisk/module.h"
#include "asterisk/app.h"
#include "asterisk/lock.h"
#include "asterisk/options.h"
#include "asterisk/strings.h"
#include "asterisk/cli.h"
#include "asterisk/utils.h"
#include "asterisk/config.h"
#include "asterisk/astobj2.h"
#include "asterisk/res_fax.h"
#include "asterisk/file.h"
#include "asterisk/channel.h"
#include "asterisk/pbx.h"
#include "asterisk/dsp.h"
#include "asterisk/indications.h"
#include "asterisk/ast_version.h"
#include "asterisk/translate.h"
#include "asterisk/stasis.h"
#include "asterisk/stasis_channels.h"
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
#include "asterisk/smoother.h"
#include "asterisk/format_cache.h"
/*** DOCUMENTATION
<application name="ReceiveFAX" language="en_US" module="res_fax">
<synopsis>
Receive a FAX and save as a TIFF/F file.
</synopsis>
<syntax>
<parameter name="filename" required="true" />
<parameter name="options">
<optionlist>
<option name="d">
<para>Enable FAX debugging.</para>
</option>
<option name="f">
<para>Allow audio fallback FAX transfer on T.38 capable channels.</para>
</option>
<option name="F">
<para>Force usage of audio mode on T.38 capable channels.</para>
</option>
<option name="s">
<para>Send progress Manager events (overrides statusevents setting in res_fax.conf).</para>
</option>
</optionlist>
</parameter>
</syntax>
<description>
<para>This application is provided by res_fax, which is a FAX technology agnostic module
that utilizes FAX technology resource modules to complete a FAX transmission.</para>
<para>Session arguments can be set by the FAXOPT function and to check results of the ReceiveFax() application.</para>
</description>
<see-also>
<ref type="function">FAXOPT</ref>
</see-also>
</application>
<application name="SendFAX" language="en_US" module="res_fax">
<synopsis>
Sends a specified TIFF/F file as a FAX.
</synopsis>
<syntax>
<parameter name="filename" required="true" argsep="&amp;">
<argument name="filename2" multiple="true">
<para>TIFF file to send as a FAX.</para>
</argument>
</parameter>
<parameter name="options">
<optionlist>
<option name="d">
<para>Enable FAX debugging.</para>
</option>
<option name="f">
<para>Allow audio fallback FAX transfer on T.38 capable channels.</para>
</option>
<option name="F">
<para>Force usage of audio mode on T.38 capable channels.</para>
</option>
<option name="s">
<para>Send progress Manager events (overrides statusevents setting in res_fax.conf).</para>
</option>
<option name="z">
<para>Initiate a T.38 reinvite on the channel if the remote end does not.</para>
</option>
</optionlist>
</parameter>
</syntax>
<description>
<para>This application is provided by res_fax, which is a FAX technology agnostic module
that utilizes FAX technology resource modules to complete a FAX transmission.</para>
<para>Session arguments can be set by the FAXOPT function and to check results of the SendFax() application.</para>
</description>
<see-also>
<ref type="function">FAXOPT</ref>
</see-also>
</application>
<function name="FAXOPT" language="en_US" module="res_fax">
<synopsis>
Gets/sets various pieces of information about a fax session.
</synopsis>
<syntax>
<parameter name="item" required="true">
<enumlist>
<enum name="ecm">
<para>R/W Error Correction Mode (ECM) enable with 'yes', disable with 'no'.</para>
</enum>
<enum name="error">
<para>R/O FAX transmission error code upon failure.</para>
</enum>
<enum name="filename">
<para>R/O Filename of the first file of the FAX transmission.</para>
</enum>
<enum name="filenames">
<para>R/O Filenames of all of the files in the FAX transmission (comma separated).</para>
</enum>
<enum name="headerinfo">
<para>R/W FAX header information.</para>
</enum>
<enum name="localstationid">
<para>R/W Local Station Identification.</para>
</enum>
<enum name="minrate">
<para>R/W Minimum transfer rate set before transmission.</para>
</enum>
<enum name="maxrate">
<para>R/W Maximum transfer rate set before transmission.</para>
</enum>
<enum name="modem">
<para>R/W Modem type (v17/v27/v29).</para>
</enum>
<enum name="gateway">
<para>R/W T38 fax gateway, with optional fax activity timeout in seconds (yes[,timeout]/no)</para>
</enum>
<enum name="faxdetect">
<para>R/W Enable FAX detect with optional timeout in seconds (yes,t38,cng[,timeout]/no)</para>
</enum>
<enum name="pages">
<para>R/O Number of pages transferred.</para>
</enum>
<enum name="rate">
<para>R/O Negotiated transmission rate.</para>
</enum>
<enum name="remotestationid">
<para>R/O Remote Station Identification after transmission.</para>
</enum>
<enum name="resolution">
<para>R/O Negotiated image resolution after transmission.</para>
</enum>
<enum name="sessionid">
<para>R/O Session ID of the FAX transmission.</para>
</enum>
<enum name="status">
<para>R/O Result Status of the FAX transmission.</para>
</enum>
<enum name="statusstr">
<para>R/O Verbose Result Status of the FAX transmission.</para>
</enum>
<enum name="t38timeout">
<para>R/W The timeout used for T.38 negotiation.</para>
</enum>
<enum name="negotiate_both">
<para>R/W Upon v21 detection allow gateway to send negotiation requests to both T.38 endpoints, and do not wait on the "other" side to initiate (yes|no)</para>
</enum>
</enumlist>
</parameter>
</syntax>
<description>
<para>FAXOPT can be used to override the settings for a FAX session listed in <filename>res_fax.conf</filename>,
it can also be used to retrieve information about a FAX session that has finished eg. pages/status.</para>
</description>
<see-also>
<ref type="application">ReceiveFax</ref>
<ref type="application">SendFax</ref>
</see-also>
</function>
<manager name="FAXSessions" language="en_US">
<synopsis>
Lists active FAX sessions
</synopsis>
<syntax>
<xi:include xpointer="xpointer(/docs/manager[@name='Login']/syntax/parameter[@name='ActionID'])" />
</syntax>
<description>
<para>Will generate a series of FAXSession events with information about each FAXSession. Closes with
a FAXSessionsComplete event which includes a count of the included FAX sessions. This action works in
the same manner as the CLI command 'fax show sessions'</para>
</description>
</manager>
<managerEvent language="en_US" name="FAXSessionsEntry">
<managerEventInstance class="EVENT_FLAG_REPORTING">
<synopsis>A single list item for the FAXSessions AMI command</synopsis>
<syntax>
<parameter name="ActionID" required="false"/>
<parameter name="Channel">
<para>Name of the channel responsible for the FAX session</para>
</parameter>
<parameter name="Technology">
<para>The FAX technology that the FAX session is using</para>
</parameter>
<parameter name="SessionNumber">
<para>The numerical identifier for this particular session</para>
</parameter>
<parameter name="SessionType">
<para>FAX session passthru/relay type</para>
<enumlist>
<enum name="G.711" />
<enum name="T.38" />
</enumlist>
</parameter>
<parameter name="Operation">
<para>FAX session operation type</para>
<enumlist>
<enum name="gateway" />
<enum name="V.21" />
<enum name="send" />
<enum name="receive" />
<enum name="none" />
</enumlist>
</parameter>
<parameter name="State">
<para>Current state of the FAX session</para>
<enumlist>
<enum name="Uninitialized" />
<enum name="Initialized" />
<enum name="Open" />
<enum name="Active" />
<enum name="Complete" />
<enum name="Reserved" />
<enum name="Inactive" />
<enum name="Unknown" />
</enumlist>
</parameter>
<parameter name="Files">
<para>File or list of files associated with this FAX session</para>
</parameter>
</syntax>
</managerEventInstance>
</managerEvent>
<managerEvent language="en_US" name="FAXSessionsComplete">
<managerEventInstance class="EVENT_FLAG_CALL">
<synopsis>Raised when all FAXSession events are completed for a FAXSessions command</synopsis>
<syntax>
<parameter name="ActionID" required="false"/>
<parameter name="Total">
<para>Count of FAXSession events sent in response to FAXSessions action</para>
</parameter>
</syntax>
</managerEventInstance>
</managerEvent>
<manager name="FAXSession" language="en_US">
<synopsis>
Responds with a detailed description of a single FAX session
</synopsis>
<syntax>
<xi:include xpointer="xpointer(/docs/manager[@name='Login']/syntax/parameter[@name='ActionID'])" />
<parameter name="SessionNumber" required="true">
<para>The session ID of the fax the user is interested in.</para>
</parameter>
</syntax>
<description>
<para>Provides details about a specific FAX session. The response will include a common subset of
the output from the CLI command 'fax show session &lt;session_number&gt;' for each technology. If the
FAX technolgy used by this session does not include a handler for FAXSession, then this action
will fail.</para>
</description>
</manager>
<managerEvent language="en_US" name="FAXSession">
<managerEventInstance class="EVENT_FLAG_REPORTING">
<synopsis>Raised in response to FAXSession manager command</synopsis>
<syntax>
<parameter name="ActionID" required="false"/>
<parameter name="SessionNumber">
<para>The numerical identifier for this particular session</para>
</parameter>
<xi:include xpointer="xpointer(/docs/managerEvent[@name='FAXSessionsEntry']/managerEventInstance/syntax/parameter[@name='Operation'])" />
<xi:include xpointer="xpointer(/docs/managerEvent[@name='FAXSessionsEntry']/managerEventInstance/syntax/parameter[@name='State'])" />
<parameter name="ErrorCorrectionMode" required="false">
<para>Whether error correcting mode is enabled for the FAX session. This field is not
included when operation is 'V.21 Detect' or if operation is 'gateway' and state is
'Uninitialized'
</para>
<enumlist>
<enum name="yes" />
<enum name="no" />
</enumlist>
</parameter>
<parameter name="DataRate" required="false">
<para>Bit rate of the FAX. This field is not included when operation is 'V.21 Detect' or
if operation is 'gateway' and state is 'Uninitialized'.</para>
</parameter>
<parameter name="ImageResolution" required="false">
<para>Resolution of each page of the FAX. Will be in the format of X_RESxY_RES. This field
is not included if the operation is anything other than Receive/Transmit.</para>
</parameter>
<parameter name="PageNumber" required="false">
<para>Current number of pages transferred during this FAX session. May change as the FAX
progresses. This field is not included when operation is 'V.21 Detect' or if operation is
'gateway' and state is 'Uninitialized'.</para>
</parameter>
<parameter name="FileName" required="false">
<para>Filename of the image being sent/received for this FAX session. This field is not
included if Operation isn't 'send' or 'receive'.</para>
</parameter>
<parameter name="PagesTransmitted" required="false">
<para>Total number of pages sent during this session. This field is not included if
Operation isn't 'send' or 'receive'. Will always be 0 for 'receive'.</para>
</parameter>
<parameter name="PagesReceived" required="false">
<para>Total number of pages received during this session. This field is not included if
Operation is not 'send' or 'receive'. Will be 0 for 'send'.</para>
</parameter>
<parameter name="TotalBadLines" required="false">
<para>Total number of bad lines sent/received during this session. This field is not
included if Operation is not 'send' or 'received'.</para>
</parameter>
</syntax>
</managerEventInstance>
</managerEvent>
<manager name="FAXStats" language="en_US">
<synopsis>
Responds with fax statistics
</synopsis>
<syntax>
<xi:include xpointer="xpointer(/docs/manager[@name='Login']/syntax/parameter[@name='ActionID'])" />
</syntax>
<description>
<para>Provides FAX statistics including the number of active sessions, reserved sessions, completed
sessions, failed sessions, and the number of receive/transmit attempts. This command provides all
of the non-technology specific information provided by the CLI command 'fax show stats'</para>
</description>
</manager>
<managerEvent language="en_US" name="FAXStats">
<managerEventInstance class="EVENT_FLAG_REPORTING">
<synopsis>Raised in response to FAXStats manager command</synopsis>
<syntax>
<parameter name="ActionID" required="false"/>
<parameter name="CurrentSessions" required="true">
<para>Number of active FAX sessions</para>
</parameter>
<parameter name="ReservedSessions" required="true">
<para>Number of reserved FAX sessions</para>
</parameter>
<parameter name="TransmitAttempts" required="true">
<para>Total FAX sessions for which Asterisk is/was the transmitter</para>
</parameter>
<parameter name="ReceiveAttempts" required="true">
<para>Total FAX sessions for which Asterisk is/was the recipient</para>
</parameter>
<parameter name="CompletedFAXes" required="true">
<para>Total FAX sessions which have been completed successfully</para>
</parameter>
<parameter name="FailedFAXes" required="true">
<para>Total FAX sessions which failed to complete successfully</para>
</parameter>
</syntax>
</managerEventInstance>
</managerEvent>
***/
static const char app_receivefax[] = "ReceiveFAX";
static const char app_sendfax[] = "SendFAX";
struct debug_info_history {
unsigned int consec_frames;
unsigned int consec_ms;
unsigned char silence;
};
struct ast_fax_debug_info {
struct timeval base_tv;
struct debug_info_history c2s, s2c;
struct ast_dsp *dsp;
};
/*! \brief used for gateway framehook */
struct fax_gateway {
/*! \brief FAX Session */
struct ast_fax_session *s;
struct ast_fax_session *peer_v21_session;
struct ast_fax_session *chan_v21_session;
/*! \brief reserved fax session token */
struct ast_fax_tech_token *token;
/*! \brief the start of our timeout counter */
struct timeval timeout_start;
/*! \brief framehook used in gateway mode */
int framehook;
/*! \brief bridged */
int bridged:1;
/*! \brief 1 if a v21 preamble has been detected */
int detected_v21:1;
/*! \brief a flag to track the state of our negotiation */
enum ast_t38_state t38_state;
/*! \brief original audio formats */
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 *chan_read_format;
struct ast_format *chan_write_format;
struct ast_format *peer_read_format;
struct ast_format *peer_write_format;
};
/*! \brief used for fax detect framehook */
struct fax_detect {
/*! \brief the start of our timeout counter */
struct timeval timeout_start;
/*! \brief DSP Processor */
struct ast_dsp *dsp;
/*! \brief original audio formats */
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 *orig_format;
/*! \brief fax session details */
struct ast_fax_session_details *details;
/*! \brief mode */
int flags;
};
/*! \brief FAX Detect flags */
#define FAX_DETECT_MODE_CNG (1 << 0)
#define FAX_DETECT_MODE_T38 (1 << 1)
#define FAX_DETECT_MODE_BOTH (FAX_DETECT_MODE_CNG | FAX_DETECT_MODE_T38)
static int fax_logger_level = -1;
/*! \brief maximum buckets for res_fax ao2 containers */
#define FAX_MAXBUCKETS 10
#define RES_FAX_TIMEOUT 10000
#define FAX_GATEWAY_TIMEOUT RES_FAX_TIMEOUT
/*! \brief The faxregistry is used to manage information and statistics for all FAX sessions. */
static struct {
/*! The number of active FAX sessions */
int active_sessions;
/*! The number of reserved FAX sessions */
int reserved_sessions;
/*! active sessions are astobj2 objects */
struct ao2_container *container;
/*! Total number of Tx FAX attempts */
int fax_tx_attempts;
/*! Total number of Rx FAX attempts */
int fax_rx_attempts;
/*! Number of successful FAX transmissions */
int fax_complete;
/*! Number of failed FAX transmissions */
int fax_failures;
/*! the next unique session name */
int nextsessionname;
} faxregistry;
/*! \brief registered FAX technology modules are put into this list */
struct fax_module {
const struct ast_fax_tech *tech;
AST_RWLIST_ENTRY(fax_module) list;
};
static AST_RWLIST_HEAD_STATIC(faxmodules, fax_module);
#define RES_FAX_MINRATE 4800
#define RES_FAX_MAXRATE 14400
#define RES_FAX_STATUSEVENTS 0
#define RES_FAX_MODEM (AST_FAX_MODEM_V17 | AST_FAX_MODEM_V27TER | AST_FAX_MODEM_V29)
#define RES_FAX_T38TIMEOUT 5000
struct fax_options {
enum ast_fax_modems modems;
uint32_t statusevents:1;
uint32_t ecm:1;
unsigned int minrate;
unsigned int maxrate;
unsigned int t38timeout;
};
static struct fax_options general_options;
static const struct fax_options default_options = {
.minrate = RES_FAX_MINRATE,
.maxrate = RES_FAX_MAXRATE,
.statusevents = RES_FAX_STATUSEVENTS,
.modems = RES_FAX_MODEM,
.ecm = AST_FAX_OPTFLAG_TRUE,
.t38timeout = RES_FAX_T38TIMEOUT,
};
AST_RWLOCK_DEFINE_STATIC(options_lock);
static void get_general_options(struct fax_options* options);
static void set_general_options(const struct fax_options* options);
static const char *config = "res_fax.conf";
static int global_fax_debug = 0;
enum {
OPT_CALLEDMODE = (1 << 0),
OPT_CALLERMODE = (1 << 1),
OPT_DEBUG = (1 << 2),
OPT_STATUS = (1 << 3),
OPT_ALLOWAUDIO = (1 << 5),
OPT_REQUEST_T38 = (1 << 6),
OPT_FORCE_AUDIO = (1 << 7),
};
AST_APP_OPTIONS(fax_exec_options, BEGIN_OPTIONS
AST_APP_OPTION('a', OPT_CALLEDMODE),
AST_APP_OPTION('c', OPT_CALLERMODE),
AST_APP_OPTION('d', OPT_DEBUG),
AST_APP_OPTION('f', OPT_ALLOWAUDIO),
AST_APP_OPTION('F', OPT_FORCE_AUDIO),
AST_APP_OPTION('s', OPT_STATUS),
AST_APP_OPTION('z', OPT_REQUEST_T38),
END_OPTIONS);
static void debug_check_frame_for_silence(struct ast_fax_session *s, unsigned int c2s, struct ast_frame *frame)
{
struct debug_info_history *history = c2s ? &s->debug_info->c2s : &s->debug_info->s2c;
int dspsilence;
unsigned int last_consec_frames, last_consec_ms;
unsigned char wassil;
struct timeval diff;
diff = ast_tvsub(ast_tvnow(), s->debug_info->base_tv);
ast_dsp_reset(s->debug_info->dsp);
ast_dsp_silence(s->debug_info->dsp, frame, &dspsilence);
wassil = history->silence;
history->silence = (dspsilence != 0) ? 1 : 0;
if (history->silence != wassil) {
last_consec_frames = history->consec_frames;
last_consec_ms = history->consec_ms;
history->consec_frames = 0;
history->consec_ms = 0;
if ((last_consec_frames != 0)) {
Logger/CLI/etc.: Fix some aesthetic issues; reduce chatty verbose messages This patch addresses some aesthetic issues in Asterisk. These are all just minor tweaks to improve the look of the CLI when used in a variety of settings. Specifically: * A number of chatty verbose messages were removed or demoted to DEBUG messages. Verbose messages with a verbosity level of 5 or higher were - if kept as verbose messages - demoted to level 4. Several messages that were emitted at verbose level 3 were demoted to 4, as announcement of dialplan applications being executed occur at level 3 (and so the effects of those applications should generally be less). * Some verbose messages that only appear when their respective 'debug' options are enabled were bumped up to always be displayed. * Prefix/timestamping of verbose messages were moved to the verboser handlers. This was done to prevent duplication of prefixes when the timestamp option (-T) is used with the CLI. * Verbose magic is removed from messages before being emitted to non-verboser handlers. This prevents the magic in multi-line verbose messages (such as SIP debug traces or the output of DumpChan) from being written to files. * _Slightly_ better support for the "light background" option (-W) was added. This includes using ast_term_quit in the output of XML documentation help, as well as changing the "Asterisk Ready" prompt to bright green on the default background (which stands a better chance of being displayed properly than bright white). Review: https://reviewboard.asterisk.org/r/3547/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@414798 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2014-05-28 22:54:12 +00:00
ast_verb(0, "Channel '%s' fax session '%u', [ %.3ld.%.6ld ], %s sent %u frames (%u ms) of %s.\n",
s->channame, s->id, (long) diff.tv_sec, (long int) diff.tv_usec,
(c2s) ? "channel" : "stack", last_consec_frames, last_consec_ms,
(wassil) ? "silence" : "energy");
}
}
history->consec_frames++;
history->consec_ms += (frame->samples / 8);
}
static void destroy_callback(void *data)
{
if (data) {
ao2_ref(data, -1);
}
}
static void fixup_callback(void *data, struct ast_channel *old_chan, struct ast_channel *new_chan);
static const struct ast_datastore_info fax_datastore = {
.type = "res_fax",
.destroy = destroy_callback,
.chan_fixup = fixup_callback,
};
static int fax_gateway_attach(struct ast_channel *chan, struct ast_fax_session_details *details);
static int fax_detect_attach(struct ast_channel *chan, int timeout, int flags);
static struct ast_fax_session_details *find_or_create_details(struct ast_channel *chan);
static struct ast_fax_session *fax_v21_session_new (struct ast_channel *chan);
/*! \brief Copies fax detection and gateway framehooks during masquerades
*
* \note must be called with both old_chan and new_chan locked. Since this
* is only called by do_masquerade, that shouldn't be an issue.
*/
static void fixup_callback(void *data, struct ast_channel *old_chan, struct ast_channel *new_chan)
{
struct ast_fax_session_details *old_details = data;
struct ast_datastore *datastore = ast_channel_datastore_find(old_chan, &fax_datastore, NULL);
if (old_details->gateway_id >= 0) {
struct ast_fax_session_details *new_details = find_or_create_details(new_chan);
ast_framehook_detach(old_chan, old_details->gateway_id);
new_details->is_t38_negotiated = old_details->is_t38_negotiated;
fax_gateway_attach(new_chan, new_details);
ao2_cleanup(new_details);
}
if (old_details->faxdetect_id >= 0) {
ast_framehook_detach(old_chan, old_details->faxdetect_id);
fax_detect_attach(new_chan, old_details->faxdetect_timeout, old_details->faxdetect_flags);
}
if (datastore) {
ast_channel_datastore_remove(old_chan, datastore);
ast_datastore_free(datastore);
}
}
/*! \brief returns a reference counted pointer to a fax datastore, if it exists */
static struct ast_fax_session_details *find_details(struct ast_channel *chan)
{
struct ast_fax_session_details *details;
struct ast_datastore *datastore;
ast_channel_lock(chan);
if (!(datastore = ast_channel_datastore_find(chan, &fax_datastore, NULL))) {
ast_channel_unlock(chan);
return NULL;
}
if (!(details = datastore->data)) {
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, "Huh? channel '%s' has a FAX datastore without data!\n", ast_channel_name(chan));
ast_channel_unlock(chan);
return NULL;
}
ao2_ref(details, 1);
ast_channel_unlock(chan);
return details;
}
/*! \brief destroy a FAX session details structure */
static void destroy_session_details(void *details)
{
struct ast_fax_session_details *d = details;
struct ast_fax_document *doc;
while ((doc = AST_LIST_REMOVE_HEAD(&d->documents, next))) {
ast_free(doc);
}
ast_string_field_free_memory(d);
}
/*! \brief create a FAX session details structure */
static struct ast_fax_session_details *session_details_new(void)
{
struct ast_fax_session_details *d;
struct fax_options options;
if (!(d = ao2_alloc(sizeof(*d), destroy_session_details))) {
return NULL;
}
if (ast_string_field_init(d, 512)) {
ao2_ref(d, -1);
return NULL;
}
get_general_options(&options);
AST_LIST_HEAD_INIT_NOLOCK(&d->documents);
/* These options need to be set to the configured default and may be overridden by
* SendFAX, ReceiveFAX, or FAXOPT */
d->option.request_t38 = AST_FAX_OPTFLAG_FALSE;
d->option.send_cng = AST_FAX_OPTFLAG_FALSE;
d->option.send_ced = AST_FAX_OPTFLAG_FALSE;
d->option.ecm = options.ecm;
d->option.statusevents = options.statusevents;
d->modems = options.modems;
d->minrate = options.minrate;
d->maxrate = options.maxrate;
d->t38timeout = options.t38timeout;
d->gateway_id = -1;
d->faxdetect_id = -1;
d->gateway_timeout = 0;
d->negotiate_both = 0;
return d;
}
static struct ast_control_t38_parameters our_t38_parameters = {
.version = 0,
.max_ifp = 400,
.rate = AST_T38_RATE_14400,
.rate_management = AST_T38_RATE_MANAGEMENT_TRANSFERRED_TCF,
};
static void t38_parameters_ast_to_fax(struct ast_fax_t38_parameters *dst, const struct ast_control_t38_parameters *src)
{
dst->version = src->version;
dst->max_ifp = src->max_ifp;
dst->rate = src->rate;
dst->rate_management = src->rate_management;
dst->fill_bit_removal = src->fill_bit_removal;
dst->transcoding_mmr = src->transcoding_mmr;
dst->transcoding_jbig = src->transcoding_jbig;
}
static void t38_parameters_fax_to_ast(struct ast_control_t38_parameters *dst, const struct ast_fax_t38_parameters *src)
{
dst->version = src->version;
dst->max_ifp = src->max_ifp;
dst->rate = src->rate;
dst->rate_management = src->rate_management;
dst->fill_bit_removal = src->fill_bit_removal;
dst->transcoding_mmr = src->transcoding_mmr;
dst->transcoding_jbig = src->transcoding_jbig;
}
/*! \brief returns a reference counted details structure from the channel's fax datastore. If the datastore
* does not exist it will be created */
static struct ast_fax_session_details *find_or_create_details(struct ast_channel *chan)
{
struct ast_fax_session_details *details;
struct ast_datastore *datastore;
if ((details = find_details(chan))) {
return details;
}
/* channel does not have one so we must create one */
if (!(details = session_details_new())) {
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' can't get a FAX details structure for the datastore!\n", ast_channel_name(chan));
return NULL;
}
if (!(datastore = ast_datastore_alloc(&fax_datastore, NULL))) {
ao2_ref(details, -1);
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' can't get a datastore!\n", ast_channel_name(chan));
return NULL;
}
/* add the datastore to the channel and increment the refcount */
datastore->data = details;
/* initialize default T.38 parameters */
t38_parameters_ast_to_fax(&details->our_t38_parameters, &our_t38_parameters);
t38_parameters_ast_to_fax(&details->their_t38_parameters, &our_t38_parameters);
ao2_ref(details, 1);
ast_channel_lock(chan);
ast_channel_datastore_add(chan, datastore);
ast_channel_unlock(chan);
return details;
}
unsigned int ast_fax_maxrate(void)
{
struct fax_options options;
get_general_options(&options);
return options.maxrate;
}
unsigned int ast_fax_minrate(void)
{
struct fax_options options;
get_general_options(&options);
return options.minrate;
}
static int update_modem_bits(enum ast_fax_modems *bits, const char *value)
{
char *m[5], *tok, *v = (char *) value, *rest;
int i = 0, j;
if (!strchr(v, ',')) {
m[i++] = v;
m[i] = NULL;
} else {
tok = strtok_r(v, ", ", &rest);
while (tok && i < ARRAY_LEN(m) - 1) {
m[i++] = tok;
tok = strtok_r(NULL, ", ", &rest);
}
m[i] = NULL;
}
*bits = 0;
for (j = 0; j < i; j++) {
if (!strcasecmp(m[j], "v17")) {
*bits |= AST_FAX_MODEM_V17;
} else if (!strcasecmp(m[j], "v27")) {
*bits |= AST_FAX_MODEM_V27TER;
} else if (!strcasecmp(m[j], "v29")) {
*bits |= AST_FAX_MODEM_V29;
} else if (!strcasecmp(m[j], "v34")) {
*bits |= AST_FAX_MODEM_V34;
} else {
ast_log(LOG_WARNING, "ignoring invalid modem setting: '%s', valid options {v17 | v27 | v29 | v34}\n", m[j]);
}
}
return 0;
}
static char *ast_fax_caps_to_str(enum ast_fax_capabilities caps, char *buf, size_t bufsize)
{
char *out = buf;
size_t size = bufsize;
int first = 1;
if (caps & AST_FAX_TECH_SEND) {
if (!first) {
ast_build_string(&buf, &size, ",");
}
ast_build_string(&buf, &size, "SEND");
first = 0;
}
if (caps & AST_FAX_TECH_RECEIVE) {
if (!first) {
ast_build_string(&buf, &size, ",");
}
ast_build_string(&buf, &size, "RECEIVE");
first = 0;
}
if (caps & AST_FAX_TECH_AUDIO) {
if (!first) {
ast_build_string(&buf, &size, ",");
}
ast_build_string(&buf, &size, "AUDIO");
first = 0;
}
if (caps & AST_FAX_TECH_T38) {
if (!first) {
ast_build_string(&buf, &size, ",");
}
ast_build_string(&buf, &size, "T38");
first = 0;
}
if (caps & AST_FAX_TECH_MULTI_DOC) {
if (!first) {
ast_build_string(&buf, &size, ",");
}
ast_build_string(&buf, &size, "MULTI_DOC");
first = 0;
}
if (caps & AST_FAX_TECH_GATEWAY) {
if (!first) {
ast_build_string(&buf, &size, ",");
}
ast_build_string(&buf, &size, "GATEWAY");
first = 0;
}
if (caps & AST_FAX_TECH_V21_DETECT) {
if (!first) {
ast_build_string(&buf, &size, ",");
}
ast_build_string(&buf, &size, "V21");
first = 0;
}
return out;
}
static int ast_fax_modem_to_str(enum ast_fax_modems bits, char *tbuf, size_t bufsize)
{
int count = 0;
if (bits & AST_FAX_MODEM_V17) {
strcat(tbuf, "V17");
count++;
}
if (bits & AST_FAX_MODEM_V27TER) {
if (count) {
strcat(tbuf, ",");
}
strcat(tbuf, "V27");
count++;
}
if (bits & AST_FAX_MODEM_V29) {
if (count) {
strcat(tbuf, ",");
}
strcat(tbuf, "V29");
count++;
}
if (bits & AST_FAX_MODEM_V34) {
if (count) {
strcat(tbuf, ",");
}
strcat(tbuf, "V34");
count++;
}
return 0;
}
static int check_modem_rate(enum ast_fax_modems modems, unsigned int rate)
{
switch (rate) {
case 2400:
case 4800:
if (!(modems & (AST_FAX_MODEM_V27TER | AST_FAX_MODEM_V34))) {
return 1;
}
break;
case 7200:
case 9600:
if (!(modems & (AST_FAX_MODEM_V17 | AST_FAX_MODEM_V29 | AST_FAX_MODEM_V34))) {
return 1;
}
break;
case 12000:
case 14400:
if (!(modems & (AST_FAX_MODEM_V17 | AST_FAX_MODEM_V34))) {
return 1;
}
break;
case 28800:
case 33600:
if (!(modems & AST_FAX_MODEM_V34)) {
return 1;
}
break;
default:
/* this should never happen */
return 1;
}
return 0;
}
/*! \brief register a FAX technology module */
int ast_fax_tech_register(struct ast_fax_tech *tech)
{
struct fax_module *fax;
if (!(fax = ast_calloc(1, sizeof(*fax)))) {
return -1;
}
fax->tech = tech;
AST_RWLIST_WRLOCK(&faxmodules);
AST_RWLIST_INSERT_TAIL(&faxmodules, fax, list);
AST_RWLIST_UNLOCK(&faxmodules);
ast_verb(3, "Registered handler for '%s' (%s)\n", fax->tech->type, fax->tech->description);
return 0;
}
/*! \brief unregister a FAX technology module */
void ast_fax_tech_unregister(struct ast_fax_tech *tech)
{
struct fax_module *fax;
ast_verb(3, "Unregistering FAX module type '%s'\n", tech->type);
AST_RWLIST_WRLOCK(&faxmodules);
AST_RWLIST_TRAVERSE_SAFE_BEGIN(&faxmodules, fax, list) {
if (fax->tech != tech) {
continue;
}
AST_RWLIST_REMOVE_CURRENT(list);
ast_free(fax);
ast_verb(4, "Unregistered FAX module type '%s'\n", tech->type);
break;
}
AST_RWLIST_TRAVERSE_SAFE_END;
AST_RWLIST_UNLOCK(&faxmodules);
}
/*! \brief convert a ast_fax_state to a string */
const char *ast_fax_state_to_str(enum ast_fax_state state)
{
switch (state) {
case AST_FAX_STATE_UNINITIALIZED:
return "Uninitialized";
case AST_FAX_STATE_INITIALIZED:
return "Initialized";
case AST_FAX_STATE_OPEN:
return "Open";
case AST_FAX_STATE_ACTIVE:
return "Active";
case AST_FAX_STATE_COMPLETE:
return "Complete";
case AST_FAX_STATE_RESERVED:
return "Reserved";
case AST_FAX_STATE_INACTIVE:
return "Inactive";
default:
ast_log(LOG_WARNING, "unhandled FAX state: %u\n", state);
return "Unknown";
}
}
void ast_fax_log(int level, const char *file, const int line, const char *function, const char *msg)
{
if (fax_logger_level != -1) {
ast_log_dynamic_level(fax_logger_level, "%s", msg);
} else {
ast_log(level, file, line, function, "%s", msg);
}
}
/*! \brief convert a rate string to a rate */
static unsigned int fax_rate_str_to_int(const char *ratestr)
{
int rate;
if (sscanf(ratestr, "%d", &rate) != 1) {
ast_log(LOG_ERROR, "failed to sscanf '%s' to rate\n", ratestr);
return 0;
}
switch (rate) {
case 2400:
case 4800:
case 7200:
case 9600:
case 12000:
case 14400:
case 28800:
case 33600:
return rate;
default:
ast_log(LOG_WARNING, "ignoring invalid rate '%s'. Valid options are {2400 | 4800 | 7200 | 9600 | 12000 | 14400 | 28800 | 33600}\n", ratestr);
return 0;
}
}
/*! \brief Release a session token.
* \param s a session returned from fax_session_reserve()
* \param token a token generated from fax_session_reserve()
*
* This function releases the given token and marks the given session as no
* longer reserved. It is safe to call on a session that is not actually
* reserved and with a NULL token. This is so that sessions returned by
* technologies that do not support reserved sessions don't require extra logic
* to handle.
*
* \note This function DOES NOT release the given fax session, only the given
* token.
*/
static void fax_session_release(struct ast_fax_session *s, struct ast_fax_tech_token *token)
{
if (token) {
s->tech->release_token(token);
}
if (s->state == AST_FAX_STATE_RESERVED) {
ast_atomic_fetchadd_int(&faxregistry.reserved_sessions, -1);
s->state = AST_FAX_STATE_INACTIVE;
}
}
/*! \brief destroy a FAX session structure */
static void destroy_session(void *session)
{
struct ast_fax_session *s = session;
if (s->tech) {
fax_session_release(s, NULL);
if (s->tech_pvt) {
s->tech->destroy_session(s);
}
ast_module_unref(s->tech->module);
}
if (s->details) {
if (s->details->caps & AST_FAX_TECH_GATEWAY) {
s->details->caps &= ~AST_FAX_TECH_GATEWAY;
}
ao2_ref(s->details, -1);
s->details = NULL;
}
if (s->debug_info) {
ast_dsp_free(s->debug_info->dsp);
ast_free(s->debug_info);
}
if (s->smoother) {
ast_smoother_free(s->smoother);
}
if (s->state != AST_FAX_STATE_INACTIVE) {
ast_atomic_fetchadd_int(&faxregistry.active_sessions, -1);
}
ast_free(s->channame);
ast_free(s->chan_uniqueid);
}
/*! \brief Reserve a fax session.
* \param details the fax session details
* \param token a pointer to a place to store a token to be passed to fax_session_new() later
*
* This function reserves a fax session for use later. If the selected fax
* technology does not support reserving sessions a session will still be
* returned but token will not be set.
*
* \note The reference returned by this function does not get consumed by
* fax_session_new() and must always be dereferenced separately.
*
* \return NULL or an uninitialized and possibly reserved session
*/
static struct ast_fax_session *fax_session_reserve(struct ast_fax_session_details *details, struct ast_fax_tech_token **token)
{
struct ast_fax_session *s;
struct fax_module *faxmod;
if (!(s = ao2_alloc(sizeof(*s), destroy_session))) {
return NULL;
}
s->state = AST_FAX_STATE_INACTIVE;
s->details = details;
ao2_ref(s->details, 1);
/* locate a FAX technology module that can handle said requirements
* Note: the requirements have not yet been finalized as T.38
* negotiation has not yet occured. */
AST_RWLIST_RDLOCK(&faxmodules);
AST_RWLIST_TRAVERSE(&faxmodules, faxmod, list) {
if ((faxmod->tech->caps & details->caps) != details->caps) {
continue;
}
if (!ast_module_running_ref(faxmod->tech->module)) {
continue;
}
ast_debug(4, "Reserving a FAX session from '%s'.\n", faxmod->tech->description);
s->tech = faxmod->tech;
break;
}
AST_RWLIST_UNLOCK(&faxmodules);
if (!faxmod) {
char caps[128] = "";
ast_log(LOG_ERROR, "Could not locate a FAX technology module with capabilities (%s)\n", ast_fax_caps_to_str(details->caps, caps, sizeof(caps)));
ao2_ref(s, -1);
return NULL;
}
if (!s->tech->reserve_session) {
ast_debug(1, "Selected FAX technology module (%s) does not support reserving sessions.\n", s->tech->description);
return s;
}
if (!(*token = s->tech->reserve_session(s))) {
ao2_ref(s, -1);
return NULL;
}
s->state = AST_FAX_STATE_RESERVED;
ast_atomic_fetchadd_int(&faxregistry.reserved_sessions, 1);
return s;
}
/*! \brief create a FAX session
*
* \param details details for the session
* \param chan the channel the session will run on
* \param reserved a reserved session to base this session on (can be NULL)
* \param token the token for a reserved session (can be NULL)
*
* Create a new fax session based on the given details structure.
*
* \note The given token is always consumed (by tech->new_session() or by
* fax_session_release() in the event of a failure). The given reference to a
* reserved session is never consumed and must be dereferenced separately from
* the reference returned by this function.
*
* \return NULL or a reference to a new fax session
*/
static struct ast_fax_session *fax_session_new(struct ast_fax_session_details *details, struct ast_channel *chan, struct ast_fax_session *reserved, struct ast_fax_tech_token *token)
{
struct ast_fax_session *s = NULL;
struct fax_module *faxmod;
if (reserved) {
s = reserved;
ao2_ref(reserved, +1);
ao2_unlink(faxregistry.container, reserved);
/* NOTE: we don't consume the reference to the reserved
* session. The session returned from fax_session_new() is a
* new reference and must be derefed in addition to the
* reserved session.
*/
if (s->state == AST_FAX_STATE_RESERVED) {
ast_atomic_fetchadd_int(&faxregistry.reserved_sessions, -1);
s->state = AST_FAX_STATE_UNINITIALIZED;
}
}
if (!s && !(s = ao2_alloc(sizeof(*s), destroy_session))) {
return NULL;
}
ast_atomic_fetchadd_int(&faxregistry.active_sessions, 1);
s->state = AST_FAX_STATE_UNINITIALIZED;
if (details->option.debug && (details->caps & AST_FAX_TECH_AUDIO)) {
if (!(s->debug_info = ast_calloc(1, sizeof(*(s->debug_info))))) {
fax_session_release(s, token);
ao2_ref(s, -1);
return NULL;
}
if (!(s->debug_info->dsp = ast_dsp_new())) {
ast_free(s->debug_info);
s->debug_info = NULL;
fax_session_release(s, token);
ao2_ref(s, -1);
return NULL;
}
ast_dsp_set_threshold(s->debug_info->dsp, 128);
}
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
if (!(s->channame = ast_strdup(ast_channel_name(chan)))) {
fax_session_release(s, token);
ao2_ref(s, -1);
return NULL;
}
if (!(s->chan_uniqueid = ast_strdup(ast_channel_uniqueid(chan)))) {
fax_session_release(s, token);
ao2_ref(s, -1);
return NULL;
}
s->chan = chan;
if (!s->details) {
s->details = details;
ao2_ref(s->details, 1);
}
details->id = s->id = ast_atomic_fetchadd_int(&faxregistry.nextsessionname, 1);
if (!token) {
/* locate a FAX technology module that can handle said requirements */
AST_RWLIST_RDLOCK(&faxmodules);
AST_RWLIST_TRAVERSE(&faxmodules, faxmod, list) {
if ((faxmod->tech->caps & details->caps) != details->caps) {
continue;
}
if (!ast_module_running_ref(faxmod->tech->module)) {
continue;
}
ast_debug(4, "Requesting a new FAX session from '%s'.\n", faxmod->tech->description);
if (reserved) {
/* Balance module ref from reserved session */
ast_module_unref(reserved->tech->module);
}
s->tech = faxmod->tech;
break;
}
AST_RWLIST_UNLOCK(&faxmodules);
if (!faxmod) {
char caps[128] = "";
ast_log(LOG_ERROR, "Could not locate a FAX technology module with capabilities (%s)\n", ast_fax_caps_to_str(details->caps, caps, sizeof(caps)));
ao2_ref(s, -1);
return NULL;
}
}
if (!(s->tech_pvt = s->tech->new_session(s, token))) {
ast_log(LOG_ERROR, "FAX session failed to initialize.\n");
ao2_ref(s, -1);
return NULL;
}
/* link the session to the session container */
if (!(ao2_link(faxregistry.container, s))) {
ast_log(LOG_ERROR, "failed to add FAX session '%u' to container.\n", s->id);
ao2_ref(s, -1);
return NULL;
}
ast_debug(4, "channel '%s' using FAX session '%u'\n", s->channame, s->id);
return s;
}
/*!
* \internal
* \brief Convert the filenames in a fax session into a JSON array
* \retval NULL on error
* \retval A \ref ast_json array on success
*/
static struct ast_json *generate_filenames_json(struct ast_fax_session_details *details)
{
RAII_VAR(struct ast_json *, json_array, ast_json_array_create(), ast_json_unref);
struct ast_fax_document *doc;
if (!details || !json_array) {
return NULL;
}
/* don't process empty lists */
if (AST_LIST_EMPTY(&details->documents)) {
return NULL;
}
AST_LIST_TRAVERSE(&details->documents, doc, next) {
struct ast_json *entry = ast_json_string_create(doc->filename);
if (!entry) {
return NULL;
}
if (ast_json_array_append(json_array, entry)) {
return NULL;
}
}
ast_json_ref(json_array);
return json_array;
}
/*!
* \brief Generate a string of filenames using the given prefix and separator.
* \param details the fax session details
* \param prefix the prefix to each filename
* \param separator the separator between filenames
*
* This function generates a string of filenames from the given details
* structure and using the given prefix and separator.
*
* \retval NULL there was an error generating the string
* \return the string generated string
*/
static char *generate_filenames_string(struct ast_fax_session_details *details, char *prefix, char *separator)
{
char *filenames, *c;
size_t size = 0;
int first = 1;
struct ast_fax_document *doc;
/* don't process empty lists */
if (AST_LIST_EMPTY(&details->documents)) {
return ast_strdup("");
}
/* Calculate the total length of all of the file names */
AST_LIST_TRAVERSE(&details->documents, doc, next) {
size += strlen(separator) + strlen(prefix) + strlen(doc->filename);
}
size += 1; /* add space for the terminating null */
if (!(filenames = ast_malloc(size))) {
return NULL;
}
c = filenames;
ast_build_string(&c, &size, "%s%s", prefix, AST_LIST_FIRST(&details->documents)->filename);
AST_LIST_TRAVERSE(&details->documents, doc, next) {
if (first) {
first = 0;
continue;
}
ast_build_string(&c, &size, "%s%s%s", separator, prefix, doc->filename);
}
return filenames;
}
/*! \brief send a FAX status manager event */
static int report_fax_status(struct ast_channel *chan, struct ast_fax_session_details *details, const char *status)
{
RAII_VAR(struct ast_json *, json_object, NULL, ast_json_unref);
RAII_VAR(struct stasis_message *, message, NULL, ao2_cleanup);
struct ast_json *json_filenames = NULL;
if (!details->option.statusevents) {
return 0;
}
json_filenames = generate_filenames_json(details);
if (!json_filenames) {
return -1;
}
json_object = ast_json_pack("{s: s, s: s, s: s, s: s, s: o}",
"type", "status",
"operation", (details->caps & AST_FAX_TECH_GATEWAY)
? "gateway"
: (details->caps & AST_FAX_TECH_RECEIVE) ? "receive" : "send",
"status", status,
"local_station_id", AST_JSON_UTF8_VALIDATE(details->localstationid),
"filenames", json_filenames);
if (!json_object) {
return -1;
}
{
SCOPED_CHANNELLOCK(lock, chan);
message = ast_channel_blob_create_from_cache(ast_channel_uniqueid(chan), ast_channel_fax_type(), json_object);
if (!message) {
return -1;
}
stasis_publish(ast_channel_topic(chan), message);
}
return 0;
}
/*! \brief Set fax related channel variables. */
static void set_channel_variables(struct ast_channel *chan, struct ast_fax_session_details *details)
{
char buf[10];
pbx_builtin_setvar_helper(chan, "FAXSTATUS", S_OR(details->result, NULL));
pbx_builtin_setvar_helper(chan, "FAXERROR", S_OR(details->error, NULL));
pbx_builtin_setvar_helper(chan, "FAXSTATUSSTRING", S_OR(details->resultstr, NULL));
pbx_builtin_setvar_helper(chan, "REMOTESTATIONID", AST_JSON_UTF8_VALIDATE(details->remotestationid));
pbx_builtin_setvar_helper(chan, "LOCALSTATIONID", AST_JSON_UTF8_VALIDATE(details->localstationid));
pbx_builtin_setvar_helper(chan, "FAXBITRATE", S_OR(details->transfer_rate, NULL));
pbx_builtin_setvar_helper(chan, "FAXRESOLUTION", S_OR(details->resolution, NULL));
if (details->is_t38_negotiated) {
pbx_builtin_setvar_helper(chan, "FAXMODE", "T38");
} else {
pbx_builtin_setvar_helper(chan, "FAXMODE", "audio");
}
snprintf(buf, sizeof(buf), "%u", details->pages_transferred);
pbx_builtin_setvar_helper(chan, "FAXPAGES", buf);
}
#define GENERIC_FAX_EXEC_SET_VARS(fax, chan, errorstr, reason) \
do { \
if (ast_strlen_zero(fax->details->result)) \
ast_string_field_set(fax->details, result, "FAILED"); \
if (ast_strlen_zero(fax->details->resultstr)) \
ast_string_field_set(fax->details, resultstr, reason); \
if (ast_strlen_zero(fax->details->error)) \
ast_string_field_set(fax->details, error, errorstr); \
set_channel_variables(chan, fax->details); \
} while (0)
#define GENERIC_FAX_EXEC_ERROR_QUIET(fax, chan, errorstr, reason) \
do { \
GENERIC_FAX_EXEC_SET_VARS(fax, chan, errorstr, reason); \
} while (0)
#define GENERIC_FAX_EXEC_ERROR(fax, chan, errorstr, reason) \
do { \
ast_log(LOG_ERROR, "channel '%s' FAX session '%u' failure, reason: '%s' (%s)\n", ast_channel_name(chan), fax->id, reason, errorstr); \
GENERIC_FAX_EXEC_ERROR_QUIET(fax, chan, errorstr, reason); \
} while (0)
static int set_fax_t38_caps(struct ast_channel *chan, struct ast_fax_session_details *details)
{
switch (ast_channel_get_t38_state(chan)) {
Improve handling of T.38 re-INVITEs that arrive before a T.38-capable application is executing on a channel. This patch addresses an issue found during working with end-users using res_fax. If an incoming call is answered in the dialplan, or jumps to the 'fax' extension due to reception of a CNG tone (with faxdetect enabled), and then the remote endpoint sends a T.38 re-INVITE, it is possible for the channel's T.38 state to be 'T38_STATE_NEGOTIATING' when the application starts up. Unfortunately, even if the application wants to use T.38, it can't respond to the peer's negotiation request, because the AST_CONTROL_T38_PARAMETERS control frame that chan_sip sent originally has been lost, and the application needs the content of that frame to be able to formulate a reply. This patch adds a new 'request' type to AST_CONTROL_T38_PARAMETERS, AST_T38_REQUEST_PARMS. If the application sends this request, chan_sip will re-send the original control frame (with AST_T38_REQUEST_NEGOTIATE as the request type), and the application can respond as normal. If this occurs within the five second timeout in chan_sip, the automatic cancellation of the peer reinvite will be stopped, and the application will 'own' the negotiation process from that point onwards. This also improves the code path in chan_sip to allow sip_indicate(), when called for AST_CONTROL_T38_PARAMETERS, to be able to return a non-zero response, which should have been in place before since the control frame *can* fail to be processed properly. It also modifies ast_indicate() to return whatever result the channel driver returned for this control frame, rather than converting all non-zero results into '-1'. Finally, the new request type intentionally returns a positive value, so that an application that sends AST_T38_REQUEST_PARMS can know for certain whether the channel driver accepted it and will be replying with a control frame of its own, or whether it was ignored (if the sip_indicate()/ast_indicate() path had properly supported failure responses before, this would not be necessary). This patch also modifies res_fax to take advantage of the new request. In addition, this patch makes sip_t38_abort() actually lock the private structure before doing its work... bad programmer, no donut. This patch also enhances chan_sip's 'faxdetect' support to allow triggering on T.38 re-INVITEs received as well as CNG tone detection. Review: https://reviewboard.asterisk.org/r/556/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@254450 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2010-03-25 15:27:31 +00:00
case T38_STATE_UNKNOWN:
details->caps |= AST_FAX_TECH_T38;
break;
case T38_STATE_REJECTED:
Improve handling of T.38 re-INVITEs that arrive before a T.38-capable application is executing on a channel. This patch addresses an issue found during working with end-users using res_fax. If an incoming call is answered in the dialplan, or jumps to the 'fax' extension due to reception of a CNG tone (with faxdetect enabled), and then the remote endpoint sends a T.38 re-INVITE, it is possible for the channel's T.38 state to be 'T38_STATE_NEGOTIATING' when the application starts up. Unfortunately, even if the application wants to use T.38, it can't respond to the peer's negotiation request, because the AST_CONTROL_T38_PARAMETERS control frame that chan_sip sent originally has been lost, and the application needs the content of that frame to be able to formulate a reply. This patch adds a new 'request' type to AST_CONTROL_T38_PARAMETERS, AST_T38_REQUEST_PARMS. If the application sends this request, chan_sip will re-send the original control frame (with AST_T38_REQUEST_NEGOTIATE as the request type), and the application can respond as normal. If this occurs within the five second timeout in chan_sip, the automatic cancellation of the peer reinvite will be stopped, and the application will 'own' the negotiation process from that point onwards. This also improves the code path in chan_sip to allow sip_indicate(), when called for AST_CONTROL_T38_PARAMETERS, to be able to return a non-zero response, which should have been in place before since the control frame *can* fail to be processed properly. It also modifies ast_indicate() to return whatever result the channel driver returned for this control frame, rather than converting all non-zero results into '-1'. Finally, the new request type intentionally returns a positive value, so that an application that sends AST_T38_REQUEST_PARMS can know for certain whether the channel driver accepted it and will be replying with a control frame of its own, or whether it was ignored (if the sip_indicate()/ast_indicate() path had properly supported failure responses before, this would not be necessary). This patch also modifies res_fax to take advantage of the new request. In addition, this patch makes sip_t38_abort() actually lock the private structure before doing its work... bad programmer, no donut. This patch also enhances chan_sip's 'faxdetect' support to allow triggering on T.38 re-INVITEs received as well as CNG tone detection. Review: https://reviewboard.asterisk.org/r/556/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@254450 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2010-03-25 15:27:31 +00:00
case T38_STATE_UNAVAILABLE:
details->caps |= AST_FAX_TECH_AUDIO;
break;
case T38_STATE_NEGOTIATED:
/* already in T.38 mode? This should not happen. */
Improve handling of T.38 re-INVITEs that arrive before a T.38-capable application is executing on a channel. This patch addresses an issue found during working with end-users using res_fax. If an incoming call is answered in the dialplan, or jumps to the 'fax' extension due to reception of a CNG tone (with faxdetect enabled), and then the remote endpoint sends a T.38 re-INVITE, it is possible for the channel's T.38 state to be 'T38_STATE_NEGOTIATING' when the application starts up. Unfortunately, even if the application wants to use T.38, it can't respond to the peer's negotiation request, because the AST_CONTROL_T38_PARAMETERS control frame that chan_sip sent originally has been lost, and the application needs the content of that frame to be able to formulate a reply. This patch adds a new 'request' type to AST_CONTROL_T38_PARAMETERS, AST_T38_REQUEST_PARMS. If the application sends this request, chan_sip will re-send the original control frame (with AST_T38_REQUEST_NEGOTIATE as the request type), and the application can respond as normal. If this occurs within the five second timeout in chan_sip, the automatic cancellation of the peer reinvite will be stopped, and the application will 'own' the negotiation process from that point onwards. This also improves the code path in chan_sip to allow sip_indicate(), when called for AST_CONTROL_T38_PARAMETERS, to be able to return a non-zero response, which should have been in place before since the control frame *can* fail to be processed properly. It also modifies ast_indicate() to return whatever result the channel driver returned for this control frame, rather than converting all non-zero results into '-1'. Finally, the new request type intentionally returns a positive value, so that an application that sends AST_T38_REQUEST_PARMS can know for certain whether the channel driver accepted it and will be replying with a control frame of its own, or whether it was ignored (if the sip_indicate()/ast_indicate() path had properly supported failure responses before, this would not be necessary). This patch also modifies res_fax to take advantage of the new request. In addition, this patch makes sip_t38_abort() actually lock the private structure before doing its work... bad programmer, no donut. This patch also enhances chan_sip's 'faxdetect' support to allow triggering on T.38 re-INVITEs received as well as CNG tone detection. Review: https://reviewboard.asterisk.org/r/556/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@254450 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2010-03-25 15:27:31 +00:00
case T38_STATE_NEGOTIATING: {
/* the other end already sent us a T.38 reinvite, so we need to prod the channel
* driver into resending their parameters to us if it supports doing so... if
* not, we can't proceed, because we can't create a proper reply without them.
* if it does work, the channel driver will send an AST_CONTROL_T38_PARAMETERS
* with a request of AST_T38_REQUEST_NEGOTIATE, which will be read by the function
* that gets called after this one completes
*/
struct ast_control_t38_parameters parameters = { .request_response = AST_T38_REQUEST_PARMS, };
if (ast_indicate_data(chan, AST_CONTROL_T38_PARAMETERS, &parameters, sizeof(parameters)) != AST_T38_REQUEST_PARMS) {
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_ERROR, "channel '%s' is in an unsupported T.38 negotiation state, cannot continue.\n", ast_channel_name(chan));
Improve handling of T.38 re-INVITEs that arrive before a T.38-capable application is executing on a channel. This patch addresses an issue found during working with end-users using res_fax. If an incoming call is answered in the dialplan, or jumps to the 'fax' extension due to reception of a CNG tone (with faxdetect enabled), and then the remote endpoint sends a T.38 re-INVITE, it is possible for the channel's T.38 state to be 'T38_STATE_NEGOTIATING' when the application starts up. Unfortunately, even if the application wants to use T.38, it can't respond to the peer's negotiation request, because the AST_CONTROL_T38_PARAMETERS control frame that chan_sip sent originally has been lost, and the application needs the content of that frame to be able to formulate a reply. This patch adds a new 'request' type to AST_CONTROL_T38_PARAMETERS, AST_T38_REQUEST_PARMS. If the application sends this request, chan_sip will re-send the original control frame (with AST_T38_REQUEST_NEGOTIATE as the request type), and the application can respond as normal. If this occurs within the five second timeout in chan_sip, the automatic cancellation of the peer reinvite will be stopped, and the application will 'own' the negotiation process from that point onwards. This also improves the code path in chan_sip to allow sip_indicate(), when called for AST_CONTROL_T38_PARAMETERS, to be able to return a non-zero response, which should have been in place before since the control frame *can* fail to be processed properly. It also modifies ast_indicate() to return whatever result the channel driver returned for this control frame, rather than converting all non-zero results into '-1'. Finally, the new request type intentionally returns a positive value, so that an application that sends AST_T38_REQUEST_PARMS can know for certain whether the channel driver accepted it and will be replying with a control frame of its own, or whether it was ignored (if the sip_indicate()/ast_indicate() path had properly supported failure responses before, this would not be necessary). This patch also modifies res_fax to take advantage of the new request. In addition, this patch makes sip_t38_abort() actually lock the private structure before doing its work... bad programmer, no donut. This patch also enhances chan_sip's 'faxdetect' support to allow triggering on T.38 re-INVITEs received as well as CNG tone detection. Review: https://reviewboard.asterisk.org/r/556/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@254450 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2010-03-25 15:27:31 +00:00
return -1;
}
details->caps |= AST_FAX_TECH_T38;
break;
}
default:
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_ERROR, "channel '%s' is in an unsupported T.38 negotiation state, cannot continue.\n", ast_channel_name(chan));
Improve handling of T.38 re-INVITEs that arrive before a T.38-capable application is executing on a channel. This patch addresses an issue found during working with end-users using res_fax. If an incoming call is answered in the dialplan, or jumps to the 'fax' extension due to reception of a CNG tone (with faxdetect enabled), and then the remote endpoint sends a T.38 re-INVITE, it is possible for the channel's T.38 state to be 'T38_STATE_NEGOTIATING' when the application starts up. Unfortunately, even if the application wants to use T.38, it can't respond to the peer's negotiation request, because the AST_CONTROL_T38_PARAMETERS control frame that chan_sip sent originally has been lost, and the application needs the content of that frame to be able to formulate a reply. This patch adds a new 'request' type to AST_CONTROL_T38_PARAMETERS, AST_T38_REQUEST_PARMS. If the application sends this request, chan_sip will re-send the original control frame (with AST_T38_REQUEST_NEGOTIATE as the request type), and the application can respond as normal. If this occurs within the five second timeout in chan_sip, the automatic cancellation of the peer reinvite will be stopped, and the application will 'own' the negotiation process from that point onwards. This also improves the code path in chan_sip to allow sip_indicate(), when called for AST_CONTROL_T38_PARAMETERS, to be able to return a non-zero response, which should have been in place before since the control frame *can* fail to be processed properly. It also modifies ast_indicate() to return whatever result the channel driver returned for this control frame, rather than converting all non-zero results into '-1'. Finally, the new request type intentionally returns a positive value, so that an application that sends AST_T38_REQUEST_PARMS can know for certain whether the channel driver accepted it and will be replying with a control frame of its own, or whether it was ignored (if the sip_indicate()/ast_indicate() path had properly supported failure responses before, this would not be necessary). This patch also modifies res_fax to take advantage of the new request. In addition, this patch makes sip_t38_abort() actually lock the private structure before doing its work... bad programmer, no donut. This patch also enhances chan_sip's 'faxdetect' support to allow triggering on T.38 re-INVITEs received as well as CNG tone detection. Review: https://reviewboard.asterisk.org/r/556/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@254450 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2010-03-25 15:27:31 +00:00
return -1;
}
return 0;
}
static int disable_t38(struct ast_channel *chan)
{
Multiple revisions 375993-375994 ........ r375993 | mmichelson | 2012-11-07 11:01:13 -0600 (Wed, 07 Nov 2012) | 30 lines Fix misuses of timeouts throughout the code. Prior to this change, a common method for determining if a timeout was reached was to call a function such as ast_waitfor_n() and inspect the out parameter that told how many milliseconds were left, then use that as the input to ast_waitfor_n() on the next go-around. The problem with this is that in some cases, submillisecond timeouts can occur, resulting in the out parameter not decreasing any. When this happens thousands of times, the result is that the timeout takes much longer than intended to be reached. As an example, I had a situation where a 3 second timeout took multiple days to finally end since most wakeups from ast_waitfor_n() were under a millisecond. This patch seeks to fix this pattern throughout the code. Now we log the time when an operation began and find the difference in wall clock time between now and when the event started. This means that sub-millisecond timeouts now cannot play havoc when trying to determine if something has timed out. Part of this fix also includes changing the function ast_waitfor() so that it is possible for it to return less than zero when a negative timeout is given to it. This makes it actually possible to detect errors in ast_waitfor() when there is no timeout. (closes issue ASTERISK-20414) reported by David M. Lee Review: https://reviewboard.asterisk.org/r/2135/ ........ r375994 | mmichelson | 2012-11-07 11:08:44 -0600 (Wed, 07 Nov 2012) | 3 lines Remove some debugging that accidentally made it in the last commit. ........ Merged revisions 375993-375994 from http://svn.asterisk.org/svn/asterisk/branches/1.8 ........ Merged revisions 375995 from http://svn.asterisk.org/svn/asterisk/branches/10 ........ Merged revisions 376014 from http://svn.asterisk.org/svn/asterisk/branches/11 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@376015 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-11-07 19:15:26 +00:00
int timeout_ms;
struct ast_frame *frame = NULL;
struct ast_control_t38_parameters t38_parameters = { .request_response = AST_T38_REQUEST_TERMINATE, };
Multiple revisions 375993-375994 ........ r375993 | mmichelson | 2012-11-07 11:01:13 -0600 (Wed, 07 Nov 2012) | 30 lines Fix misuses of timeouts throughout the code. Prior to this change, a common method for determining if a timeout was reached was to call a function such as ast_waitfor_n() and inspect the out parameter that told how many milliseconds were left, then use that as the input to ast_waitfor_n() on the next go-around. The problem with this is that in some cases, submillisecond timeouts can occur, resulting in the out parameter not decreasing any. When this happens thousands of times, the result is that the timeout takes much longer than intended to be reached. As an example, I had a situation where a 3 second timeout took multiple days to finally end since most wakeups from ast_waitfor_n() were under a millisecond. This patch seeks to fix this pattern throughout the code. Now we log the time when an operation began and find the difference in wall clock time between now and when the event started. This means that sub-millisecond timeouts now cannot play havoc when trying to determine if something has timed out. Part of this fix also includes changing the function ast_waitfor() so that it is possible for it to return less than zero when a negative timeout is given to it. This makes it actually possible to detect errors in ast_waitfor() when there is no timeout. (closes issue ASTERISK-20414) reported by David M. Lee Review: https://reviewboard.asterisk.org/r/2135/ ........ r375994 | mmichelson | 2012-11-07 11:08:44 -0600 (Wed, 07 Nov 2012) | 3 lines Remove some debugging that accidentally made it in the last commit. ........ Merged revisions 375993-375994 from http://svn.asterisk.org/svn/asterisk/branches/1.8 ........ Merged revisions 375995 from http://svn.asterisk.org/svn/asterisk/branches/10 ........ Merged revisions 376014 from http://svn.asterisk.org/svn/asterisk/branches/11 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@376015 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-11-07 19:15:26 +00:00
struct timeval start;
int ms;
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_${fieldname}' and then writing setters/getters and converting the existing code to using them. When all fields are done, we can move ast_channel to a C file from channel.h and lop off the '__do_not_use_'. This patch sets up main/channel_interal_api.c to be the only file that actually accesses the ast_channel's fields directly. The intent would be for any API functions in channel.c to use the accessor functions. No more monkeying around with channel internals. We should use our own APIs. The interesting changes in this patch are the addition of channel_internal_api.c, the moving of the AST_DATA stuff from channel.c to channel_internal_api.c (note: the AST_DATA stuff will have to be reworked to use accessor functions when ast_channel is really opaque), and some re-working of the way channel iterators/callbacks are handled so as to avoid creating fake ast_channels on the stack to pass in matching data by directly accessing fields (since "name" is a stringfield and the fake channel doesn't init the stringfields, you can't use the ast_channel_name_set() function). I went with ast_channel_name(chan) for a getter, and ast_channel_name_set(chan, name) for a setter. The majority of the grunt-work for this change was done by writing a semantic patch using Coccinelle ( http://coccinelle.lip6.fr/ ). Review: https://reviewboard.asterisk.org/r/1655/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@350223 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-01-09 22:15:50 +00:00
ast_debug(1, "Shutting down T.38 on %s\n", ast_channel_name(chan));
if (ast_indicate_data(chan, AST_CONTROL_T38_PARAMETERS, &t38_parameters, sizeof(t38_parameters)) != 0) {
Replace direct access to channel name with accessor functions There are many benefits to making the ast_channel an opaque handle, from increasing maintainability to presenting ways to kill masquerades. This patch kicks things off by taking things a field at a time, renaming the field to '__do_not_use_${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_debug(1, "error while disabling T.38 on channel '%s'\n", ast_channel_name(chan));
return -1;
}
/* wait up to five seconds for negotiation to complete */
Multiple revisions 375993-375994 ........ r375993 | mmichelson | 2012-11-07 11:01:13 -0600 (Wed, 07 Nov 2012) | 30 lines Fix misuses of timeouts throughout the code. Prior to this change, a common method for determining if a timeout was reached was to call a function such as ast_waitfor_n() and inspect the out parameter that told how many milliseconds were left, then use that as the input to ast_waitfor_n() on the next go-around. The problem with this is that in some cases, submillisecond timeouts can occur, resulting in the out parameter not decreasing any. When this happens thousands of times, the result is that the timeout takes much longer than intended to be reached. As an example, I had a situation where a 3 second timeout took multiple days to finally end since most wakeups from ast_waitfor_n() were under a millisecond. This patch seeks to fix this pattern throughout the code. Now we log the time when an operation began and find the difference in wall clock time between now and when the event started. This means that sub-millisecond timeouts now cannot play havoc when trying to determine if something has timed out. Part of this fix also includes changing the function ast_waitfor() so that it is possible for it to return less than zero when a negative timeout is given to it. This makes it actually possible to detect errors in ast_waitfor() when there is no timeout. (closes issue ASTERISK-20414) reported by David M. Lee Review: https://reviewboard.asterisk.org/r/2135/ ........ r375994 | mmichelson | 2012-11-07 11:08:44 -0600 (Wed, 07 Nov 2012) | 3 lines Remove some debugging that accidentally made it in the last commit. ........ Merged revisions 375993-375994 from http://svn.asterisk.org/svn/asterisk/branches/1.8 ........ Merged revisions 375995 from http://svn.asterisk.org/svn/asterisk/branches/10 ........ Merged revisions 376014 from http://svn.asterisk.org/svn/asterisk/branches/11 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@376015 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-11-07 19:15:26 +00:00
timeout_ms = 5000;
start = ast_tvnow();
while ((ms = ast_remaining_ms(start, timeout_ms))) {
ms = ast_waitfor(chan, ms);
Multiple revisions 375993-375994 ........ r375993 | mmichelson | 2012-11-07 11:01:13 -0600 (Wed, 07 Nov 2012) | 30 lines Fix misuses of timeouts throughout the code. Prior to this change, a common method for determining if a timeout was reached was to call a function such as ast_waitfor_n() and inspect the out parameter that told how many milliseconds were left, then use that as the input to ast_waitfor_n() on the next go-around. The problem with this is that in some cases, submillisecond timeouts can occur, resulting in the out parameter not decreasing any. When this happens thousands of times, the result is that the timeout takes much longer than intended to be reached. As an example, I had a situation where a 3 second timeout took multiple days to finally end since most wakeups from ast_waitfor_n() were under a millisecond. This patch seeks to fix this pattern throughout the code. Now we log the time when an operation began and find the difference in wall clock time between now and when the event started. This means that sub-millisecond timeouts now cannot play havoc when trying to determine if something has timed out. Part of this fix also includes changing the function ast_waitfor() so that it is possible for it to return less than zero when a negative timeout is given to it. This makes it actually possible to detect errors in ast_waitfor() when there is no timeout. (closes issue ASTERISK-20414) reported by David M. Lee Review: https://reviewboard.asterisk.org/r/2135/ ........ r375994 | mmichelson | 2012-11-07 11:08:44 -0600 (Wed, 07 Nov 2012) | 3 lines Remove some debugging that accidentally made it in the last commit. ........ Merged revisions 375993-375994 from http://svn.asterisk.org/svn/asterisk/branches/1.8 ........ Merged revisions 375995 from http://svn.asterisk.org/svn/asterisk/branches/10 ........ Merged revisions 376014 from http://svn.asterisk.org/svn/asterisk/branches/11 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@376015 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-11-07 19:15:26 +00:00
if (ms == 0) {
break;
}
if (ms < 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_debug(1, "error while disabling T.38 on channel '%s'\n", ast_channel_name(chan));
return -1;
}
if (!(frame = ast_read(chan))) {
return -1;
}
if ((frame->frametype == AST_FRAME_CONTROL) &&
(frame->subclass.integer == AST_CONTROL_T38_PARAMETERS) &&
(frame->datalen == sizeof(t38_parameters))) {
struct ast_control_t38_parameters *parameters = frame->data.ptr;
switch (parameters->request_response) {
case AST_T38_TERMINATED:
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_debug(1, "Shut down T.38 on %s\n", ast_channel_name(chan));
break;
case AST_T38_REFUSED:
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_${fieldname}' and then writing setters/getters and converting the existing code to using them. When all fields are done, we can move ast_channel to a C file from channel.h and lop off the '__do_not_use_'. This patch sets up main/channel_interal_api.c to be the only file that actually accesses the ast_channel's fields directly. The intent would be for any API functions in channel.c to use the accessor functions. No more monkeying around with channel internals. We should use our own APIs. The interesting changes in this patch are the addition of channel_internal_api.c, the moving of the AST_DATA stuff from channel.c to channel_internal_api.c (note: the AST_DATA stuff will have to be reworked to use accessor functions when ast_channel is really opaque), and some re-working of the way channel iterators/callbacks are handled so as to avoid creating fake ast_channels on the stack to pass in matching data by directly accessing fields (since "name" is a stringfield and the fake channel doesn't init the stringfields, you can't use the ast_channel_name_set() function). I went with ast_channel_name(chan) for a getter, and ast_channel_name_set(chan, name) for a setter. The majority of the grunt-work for this change was done by writing a semantic patch using Coccinelle ( http://coccinelle.lip6.fr/ ). Review: https://reviewboard.asterisk.org/r/1655/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@350223 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-01-09 22:15:50 +00:00
ast_log(LOG_WARNING, "channel '%s' refused to disable T.38\n", ast_channel_name(chan));
ast_frfree(frame);
return -1;
default:
Replace direct access to channel name with accessor functions There are many benefits to making the ast_channel an opaque handle, from increasing maintainability to presenting ways to kill masquerades. This patch kicks things off by taking things a field at a time, renaming the field to '__do_not_use_${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_ERROR, "channel '%s' failed to disable T.38\n", ast_channel_name(chan));
ast_frfree(frame);
return -1;
}
ast_frfree(frame);
break;
}
ast_frfree(frame);
}
Multiple revisions 375993-375994 ........ r375993 | mmichelson | 2012-11-07 11:01:13 -0600 (Wed, 07 Nov 2012) | 30 lines Fix misuses of timeouts throughout the code. Prior to this change, a common method for determining if a timeout was reached was to call a function such as ast_waitfor_n() and inspect the out parameter that told how many milliseconds were left, then use that as the input to ast_waitfor_n() on the next go-around. The problem with this is that in some cases, submillisecond timeouts can occur, resulting in the out parameter not decreasing any. When this happens thousands of times, the result is that the timeout takes much longer than intended to be reached. As an example, I had a situation where a 3 second timeout took multiple days to finally end since most wakeups from ast_waitfor_n() were under a millisecond. This patch seeks to fix this pattern throughout the code. Now we log the time when an operation began and find the difference in wall clock time between now and when the event started. This means that sub-millisecond timeouts now cannot play havoc when trying to determine if something has timed out. Part of this fix also includes changing the function ast_waitfor() so that it is possible for it to return less than zero when a negative timeout is given to it. This makes it actually possible to detect errors in ast_waitfor() when there is no timeout. (closes issue ASTERISK-20414) reported by David M. Lee Review: https://reviewboard.asterisk.org/r/2135/ ........ r375994 | mmichelson | 2012-11-07 11:08:44 -0600 (Wed, 07 Nov 2012) | 3 lines Remove some debugging that accidentally made it in the last commit. ........ Merged revisions 375993-375994 from http://svn.asterisk.org/svn/asterisk/branches/1.8 ........ Merged revisions 375995 from http://svn.asterisk.org/svn/asterisk/branches/10 ........ Merged revisions 376014 from http://svn.asterisk.org/svn/asterisk/branches/11 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@376015 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-11-07 19:15:26 +00:00
if (ms == 0) { /* all done, nothing happened */
ast_debug(1, "channel '%s' timed-out during T.38 shutdown\n", ast_channel_name(chan));
}
return 0;
}
/*! \brief this is the generic FAX session handling function */
static int generic_fax_exec(struct ast_channel *chan, struct ast_fax_session_details *details, struct ast_fax_session *reserved, struct ast_fax_tech_token *token)
{
int ms;
int timeout = RES_FAX_TIMEOUT;
int chancount;
unsigned int expected_frametype = -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
struct ast_frame_subclass expected_framesubclass = { .integer = 0, };
unsigned int t38negotiated = (ast_channel_get_t38_state(chan) == T38_STATE_NEGOTIATED);
struct ast_control_t38_parameters t38_parameters;
const char *tempvar;
struct ast_fax_session *fax = NULL;
struct ast_frame *frame = NULL;
struct ast_channel *c = chan;
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
RAII_VAR(struct ast_format *, orig_write_format, NULL, ao2_cleanup);
RAII_VAR(struct ast_format *, orig_read_format, NULL, ao2_cleanup);
Multiple revisions 375993-375994 ........ r375993 | mmichelson | 2012-11-07 11:01:13 -0600 (Wed, 07 Nov 2012) | 30 lines Fix misuses of timeouts throughout the code. Prior to this change, a common method for determining if a timeout was reached was to call a function such as ast_waitfor_n() and inspect the out parameter that told how many milliseconds were left, then use that as the input to ast_waitfor_n() on the next go-around. The problem with this is that in some cases, submillisecond timeouts can occur, resulting in the out parameter not decreasing any. When this happens thousands of times, the result is that the timeout takes much longer than intended to be reached. As an example, I had a situation where a 3 second timeout took multiple days to finally end since most wakeups from ast_waitfor_n() were under a millisecond. This patch seeks to fix this pattern throughout the code. Now we log the time when an operation began and find the difference in wall clock time between now and when the event started. This means that sub-millisecond timeouts now cannot play havoc when trying to determine if something has timed out. Part of this fix also includes changing the function ast_waitfor() so that it is possible for it to return less than zero when a negative timeout is given to it. This makes it actually possible to detect errors in ast_waitfor() when there is no timeout. (closes issue ASTERISK-20414) reported by David M. Lee Review: https://reviewboard.asterisk.org/r/2135/ ........ r375994 | mmichelson | 2012-11-07 11:08:44 -0600 (Wed, 07 Nov 2012) | 3 lines Remove some debugging that accidentally made it in the last commit. ........ Merged revisions 375993-375994 from http://svn.asterisk.org/svn/asterisk/branches/1.8 ........ Merged revisions 375995 from http://svn.asterisk.org/svn/asterisk/branches/10 ........ Merged revisions 376014 from http://svn.asterisk.org/svn/asterisk/branches/11 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@376015 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-11-07 19:15:26 +00:00
int remaining_time;
struct timeval start;
chancount = 1;
/* Make sure one or the other is set to avoid race condition */
if (t38negotiated) {
details->caps |= AST_FAX_TECH_T38;
} else {
details->caps |= AST_FAX_TECH_AUDIO;
}
/* create the FAX session */
if (!(fax = fax_session_new(details, chan, reserved, token))) {
ast_log(LOG_ERROR, "Can't create a FAX session, FAX attempt failed.\n");
report_fax_status(chan, details, "No Available Resource");
return -1;
}
ast_channel_lock(chan);
/* update session details */
if (ast_strlen_zero(details->headerinfo) && (tempvar = pbx_builtin_getvar_helper(chan, "LOCALHEADERINFO"))) {
ast_string_field_set(details, headerinfo, tempvar);
}
if (ast_strlen_zero(details->localstationid)) {
tempvar = pbx_builtin_getvar_helper(chan, "LOCALSTATIONID");
ast_string_field_set(details, localstationid, tempvar ? tempvar : "unknown");
}
ast_channel_unlock(chan);
report_fax_status(chan, details, "Allocating Resources");
if (details->caps & AST_FAX_TECH_AUDIO) {
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
expected_frametype = AST_FRAME_VOICE;
expected_framesubclass.format = ast_format_slin;
orig_write_format = ao2_bump(ast_channel_writeformat(chan));
if (ast_set_write_format(chan, ast_format_slin) < 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_ERROR, "channel '%s' failed to set write format to signed linear'.\n", ast_channel_name(chan));
ao2_unlink(faxregistry.container, fax);
ao2_ref(fax, -1);
return -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
orig_read_format = ao2_bump(ast_channel_readformat(chan));
if (ast_set_read_format(chan, ast_format_slin) < 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_ERROR, "channel '%s' failed to set read format to signed linear.\n", ast_channel_name(chan));
ao2_unlink(faxregistry.container, fax);
ao2_ref(fax, -1);
return -1;
}
if (fax->smoother) {
ast_smoother_free(fax->smoother);
fax->smoother = NULL;
}
if (!(fax->smoother = ast_smoother_new(320))) {
ast_log(LOG_WARNING, "Channel '%s' FAX session '%u' failed to obtain a smoother.\n", ast_channel_name(chan), fax->id);
}
} else {
expected_frametype = AST_FRAME_MODEM;
expected_framesubclass.integer = AST_MODEM_T38;
}
if (fax->debug_info) {
fax->debug_info->base_tv = ast_tvnow();
}
/* reset our result fields just in case the fax tech driver wants to
* set custom error messages */
ast_string_field_set(details, result, "");
ast_string_field_set(details, resultstr, "");
ast_string_field_set(details, error, "");
details->is_t38_negotiated = t38negotiated;
set_channel_variables(chan, details);
if (fax->tech->start_session(fax) < 0) {
GENERIC_FAX_EXEC_ERROR(fax, chan, "INIT_ERROR", "failed to start FAX session");
}
report_fax_status(chan, details, "FAX Transmission In Progress");
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_debug(5, "channel %s will wait on FAX fd %d\n", ast_channel_name(chan), fax->fd);
/* handle frames for the session */
Multiple revisions 375993-375994 ........ r375993 | mmichelson | 2012-11-07 11:01:13 -0600 (Wed, 07 Nov 2012) | 30 lines Fix misuses of timeouts throughout the code. Prior to this change, a common method for determining if a timeout was reached was to call a function such as ast_waitfor_n() and inspect the out parameter that told how many milliseconds were left, then use that as the input to ast_waitfor_n() on the next go-around. The problem with this is that in some cases, submillisecond timeouts can occur, resulting in the out parameter not decreasing any. When this happens thousands of times, the result is that the timeout takes much longer than intended to be reached. As an example, I had a situation where a 3 second timeout took multiple days to finally end since most wakeups from ast_waitfor_n() were under a millisecond. This patch seeks to fix this pattern throughout the code. Now we log the time when an operation began and find the difference in wall clock time between now and when the event started. This means that sub-millisecond timeouts now cannot play havoc when trying to determine if something has timed out. Part of this fix also includes changing the function ast_waitfor() so that it is possible for it to return less than zero when a negative timeout is given to it. This makes it actually possible to detect errors in ast_waitfor() when there is no timeout. (closes issue ASTERISK-20414) reported by David M. Lee Review: https://reviewboard.asterisk.org/r/2135/ ........ r375994 | mmichelson | 2012-11-07 11:08:44 -0600 (Wed, 07 Nov 2012) | 3 lines Remove some debugging that accidentally made it in the last commit. ........ Merged revisions 375993-375994 from http://svn.asterisk.org/svn/asterisk/branches/1.8 ........ Merged revisions 375995 from http://svn.asterisk.org/svn/asterisk/branches/10 ........ Merged revisions 376014 from http://svn.asterisk.org/svn/asterisk/branches/11 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@376015 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-11-07 19:15:26 +00:00
remaining_time = timeout;
start = ast_tvnow();
while (remaining_time > 0) {
struct ast_channel *ready_chan;
int ofd, exception;
ms = 1000;
errno = 0;
ready_chan = ast_waitfor_nandfds(&c, chancount, &fax->fd, 1, &exception, &ofd, &ms);
if (ready_chan) {
if (!(frame = ast_read(chan))) {
/* the channel is probably gone, so lets stop polling on it and let the
* FAX session complete before we exit the application. if needed,
* send the FAX stack silence so the modems can finish their session without
* any problems */
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_debug(1, "Channel '%s' did not return a frame; probably hung up.\n", ast_channel_name(chan));
GENERIC_FAX_EXEC_SET_VARS(fax, chan, "HANGUP", "remote channel hungup");
c = NULL;
chancount = 0;
Multiple revisions 375993-375994 ........ r375993 | mmichelson | 2012-11-07 11:01:13 -0600 (Wed, 07 Nov 2012) | 30 lines Fix misuses of timeouts throughout the code. Prior to this change, a common method for determining if a timeout was reached was to call a function such as ast_waitfor_n() and inspect the out parameter that told how many milliseconds were left, then use that as the input to ast_waitfor_n() on the next go-around. The problem with this is that in some cases, submillisecond timeouts can occur, resulting in the out parameter not decreasing any. When this happens thousands of times, the result is that the timeout takes much longer than intended to be reached. As an example, I had a situation where a 3 second timeout took multiple days to finally end since most wakeups from ast_waitfor_n() were under a millisecond. This patch seeks to fix this pattern throughout the code. Now we log the time when an operation began and find the difference in wall clock time between now and when the event started. This means that sub-millisecond timeouts now cannot play havoc when trying to determine if something has timed out. Part of this fix also includes changing the function ast_waitfor() so that it is possible for it to return less than zero when a negative timeout is given to it. This makes it actually possible to detect errors in ast_waitfor() when there is no timeout. (closes issue ASTERISK-20414) reported by David M. Lee Review: https://reviewboard.asterisk.org/r/2135/ ........ r375994 | mmichelson | 2012-11-07 11:08:44 -0600 (Wed, 07 Nov 2012) | 3 lines Remove some debugging that accidentally made it in the last commit. ........ Merged revisions 375993-375994 from http://svn.asterisk.org/svn/asterisk/branches/1.8 ........ Merged revisions 375995 from http://svn.asterisk.org/svn/asterisk/branches/10 ........ Merged revisions 376014 from http://svn.asterisk.org/svn/asterisk/branches/11 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@376015 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-11-07 19:15:26 +00:00
remaining_time = ast_remaining_ms(start, timeout);
fax->tech->cancel_session(fax);
if (fax->tech->generate_silence) {
fax->tech->generate_silence(fax);
}
continue;
}
if ((frame->frametype == AST_FRAME_CONTROL) &&
(frame->subclass.integer == AST_CONTROL_T38_PARAMETERS) &&
(frame->datalen == sizeof(t38_parameters))) {
unsigned int was_t38 = t38negotiated;
struct ast_control_t38_parameters *parameters = frame->data.ptr;
switch (parameters->request_response) {
case AST_T38_REQUEST_NEGOTIATE:
/* the other end has requested a switch to T.38, so reply that we are willing, if we can
* do T.38 as well
*/
t38_parameters_fax_to_ast(&t38_parameters, &details->our_t38_parameters);
if (details->caps & AST_FAX_TECH_T38) {
details->is_t38_negotiated = 1;
t38_parameters.request_response = AST_T38_NEGOTIATED;
} else {
t38_parameters.request_response = AST_T38_REFUSED;
}
ast_indicate_data(chan, AST_CONTROL_T38_PARAMETERS, &t38_parameters, sizeof(t38_parameters));
break;
case AST_T38_NEGOTIATED:
t38_parameters_ast_to_fax(&details->their_t38_parameters, parameters);
t38negotiated = 1;
details->is_t38_negotiated = 1;
break;
default:
break;
}
if (t38negotiated && !was_t38) {
if (fax->tech->switch_to_t38(fax)) {
GENERIC_FAX_EXEC_ERROR(fax, chan, "UNKNOWN", "T.38 switch failed");
break;
}
details->caps &= ~AST_FAX_TECH_AUDIO;
expected_frametype = AST_FRAME_MODEM;
expected_framesubclass.integer = AST_MODEM_T38;
if (fax->smoother) {
ast_smoother_free(fax->smoother);
fax->smoother = NULL;
}
report_fax_status(chan, details, "T.38 Negotiated");
ast_verb(3, "Channel '%s' switched to T.38 FAX session '%u'.\n", ast_channel_name(chan), fax->id);
}
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 if ((frame->frametype == expected_frametype) && (expected_framesubclass.integer == frame->subclass.integer) &&
((!frame->subclass.format && !expected_framesubclass.format) ||
(frame->subclass.format && expected_framesubclass.format &&
(ast_format_cmp(frame->subclass.format, expected_framesubclass.format) != AST_FORMAT_CMP_NOT_EQUAL)))) {
struct ast_frame *f;
if (fax->smoother) {
/* push the frame into a smoother */
if (ast_smoother_feed(fax->smoother, frame) < 0) {
GENERIC_FAX_EXEC_ERROR(fax, chan, "UNKNOWN", "Failed to feed the smoother");
}
while ((f = ast_smoother_read(fax->smoother)) && (f->data.ptr)) {
if (fax->debug_info) {
debug_check_frame_for_silence(fax, 1, f);
}
/* write the frame to the FAX stack */
fax->tech->write(fax, f);
fax->frames_received++;
if (f != frame) {
ast_frfree(f);
}
}
} else {
/* write the frame to the FAX stack */
fax->tech->write(fax, frame);
fax->frames_received++;
}
Multiple revisions 375993-375994 ........ r375993 | mmichelson | 2012-11-07 11:01:13 -0600 (Wed, 07 Nov 2012) | 30 lines Fix misuses of timeouts throughout the code. Prior to this change, a common method for determining if a timeout was reached was to call a function such as ast_waitfor_n() and inspect the out parameter that told how many milliseconds were left, then use that as the input to ast_waitfor_n() on the next go-around. The problem with this is that in some cases, submillisecond timeouts can occur, resulting in the out parameter not decreasing any. When this happens thousands of times, the result is that the timeout takes much longer than intended to be reached. As an example, I had a situation where a 3 second timeout took multiple days to finally end since most wakeups from ast_waitfor_n() were under a millisecond. This patch seeks to fix this pattern throughout the code. Now we log the time when an operation began and find the difference in wall clock time between now and when the event started. This means that sub-millisecond timeouts now cannot play havoc when trying to determine if something has timed out. Part of this fix also includes changing the function ast_waitfor() so that it is possible for it to return less than zero when a negative timeout is given to it. This makes it actually possible to detect errors in ast_waitfor() when there is no timeout. (closes issue ASTERISK-20414) reported by David M. Lee Review: https://reviewboard.asterisk.org/r/2135/ ........ r375994 | mmichelson | 2012-11-07 11:08:44 -0600 (Wed, 07 Nov 2012) | 3 lines Remove some debugging that accidentally made it in the last commit. ........ Merged revisions 375993-375994 from http://svn.asterisk.org/svn/asterisk/branches/1.8 ........ Merged revisions 375995 from http://svn.asterisk.org/svn/asterisk/branches/10 ........ Merged revisions 376014 from http://svn.asterisk.org/svn/asterisk/branches/11 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@376015 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-11-07 19:15:26 +00:00
start = ast_tvnow();
}
ast_frfree(frame);
} else if (ofd == fax->fd) {
/* read a frame from the FAX stack and send it out the channel.
* the FAX stack will return a NULL if the FAX session has already completed */
if (!(frame = fax->tech->read(fax))) {
break;
}
if (fax->debug_info && (frame->frametype == AST_FRAME_VOICE)) {
debug_check_frame_for_silence(fax, 0, frame);
}
ast_write(chan, frame);
fax->frames_sent++;
ast_frfree(frame);
Multiple revisions 375993-375994 ........ r375993 | mmichelson | 2012-11-07 11:01:13 -0600 (Wed, 07 Nov 2012) | 30 lines Fix misuses of timeouts throughout the code. Prior to this change, a common method for determining if a timeout was reached was to call a function such as ast_waitfor_n() and inspect the out parameter that told how many milliseconds were left, then use that as the input to ast_waitfor_n() on the next go-around. The problem with this is that in some cases, submillisecond timeouts can occur, resulting in the out parameter not decreasing any. When this happens thousands of times, the result is that the timeout takes much longer than intended to be reached. As an example, I had a situation where a 3 second timeout took multiple days to finally end since most wakeups from ast_waitfor_n() were under a millisecond. This patch seeks to fix this pattern throughout the code. Now we log the time when an operation began and find the difference in wall clock time between now and when the event started. This means that sub-millisecond timeouts now cannot play havoc when trying to determine if something has timed out. Part of this fix also includes changing the function ast_waitfor() so that it is possible for it to return less than zero when a negative timeout is given to it. This makes it actually possible to detect errors in ast_waitfor() when there is no timeout. (closes issue ASTERISK-20414) reported by David M. Lee Review: https://reviewboard.asterisk.org/r/2135/ ........ r375994 | mmichelson | 2012-11-07 11:08:44 -0600 (Wed, 07 Nov 2012) | 3 lines Remove some debugging that accidentally made it in the last commit. ........ Merged revisions 375993-375994 from http://svn.asterisk.org/svn/asterisk/branches/1.8 ........ Merged revisions 375995 from http://svn.asterisk.org/svn/asterisk/branches/10 ........ Merged revisions 376014 from http://svn.asterisk.org/svn/asterisk/branches/11 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@376015 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-11-07 19:15:26 +00:00
start = ast_tvnow();
} else {
if (ms && (ofd < 0)) {
if ((errno == 0) || (errno == EINTR)) {
Multiple revisions 375993-375994 ........ r375993 | mmichelson | 2012-11-07 11:01:13 -0600 (Wed, 07 Nov 2012) | 30 lines Fix misuses of timeouts throughout the code. Prior to this change, a common method for determining if a timeout was reached was to call a function such as ast_waitfor_n() and inspect the out parameter that told how many milliseconds were left, then use that as the input to ast_waitfor_n() on the next go-around. The problem with this is that in some cases, submillisecond timeouts can occur, resulting in the out parameter not decreasing any. When this happens thousands of times, the result is that the timeout takes much longer than intended to be reached. As an example, I had a situation where a 3 second timeout took multiple days to finally end since most wakeups from ast_waitfor_n() were under a millisecond. This patch seeks to fix this pattern throughout the code. Now we log the time when an operation began and find the difference in wall clock time between now and when the event started. This means that sub-millisecond timeouts now cannot play havoc when trying to determine if something has timed out. Part of this fix also includes changing the function ast_waitfor() so that it is possible for it to return less than zero when a negative timeout is given to it. This makes it actually possible to detect errors in ast_waitfor() when there is no timeout. (closes issue ASTERISK-20414) reported by David M. Lee Review: https://reviewboard.asterisk.org/r/2135/ ........ r375994 | mmichelson | 2012-11-07 11:08:44 -0600 (Wed, 07 Nov 2012) | 3 lines Remove some debugging that accidentally made it in the last commit. ........ Merged revisions 375993-375994 from http://svn.asterisk.org/svn/asterisk/branches/1.8 ........ Merged revisions 375995 from http://svn.asterisk.org/svn/asterisk/branches/10 ........ Merged revisions 376014 from http://svn.asterisk.org/svn/asterisk/branches/11 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@376015 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-11-07 19:15:26 +00:00
remaining_time = ast_remaining_ms(start, timeout);
if (remaining_time <= 0)
GENERIC_FAX_EXEC_ERROR(fax, chan, "TIMEOUT", "fax session timed-out");
continue;
} else {
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, "something bad happened while channel '%s' was polling.\n", ast_channel_name(chan));
GENERIC_FAX_EXEC_ERROR(fax, chan, "UNKNOWN", "error polling data");
break;
}
} else {
/* nothing happened */
Multiple revisions 375993-375994 ........ r375993 | mmichelson | 2012-11-07 11:01:13 -0600 (Wed, 07 Nov 2012) | 30 lines Fix misuses of timeouts throughout the code. Prior to this change, a common method for determining if a timeout was reached was to call a function such as ast_waitfor_n() and inspect the out parameter that told how many milliseconds were left, then use that as the input to ast_waitfor_n() on the next go-around. The problem with this is that in some cases, submillisecond timeouts can occur, resulting in the out parameter not decreasing any. When this happens thousands of times, the result is that the timeout takes much longer than intended to be reached. As an example, I had a situation where a 3 second timeout took multiple days to finally end since most wakeups from ast_waitfor_n() were under a millisecond. This patch seeks to fix this pattern throughout the code. Now we log the time when an operation began and find the difference in wall clock time between now and when the event started. This means that sub-millisecond timeouts now cannot play havoc when trying to determine if something has timed out. Part of this fix also includes changing the function ast_waitfor() so that it is possible for it to return less than zero when a negative timeout is given to it. This makes it actually possible to detect errors in ast_waitfor() when there is no timeout. (closes issue ASTERISK-20414) reported by David M. Lee Review: https://reviewboard.asterisk.org/r/2135/ ........ r375994 | mmichelson | 2012-11-07 11:08:44 -0600 (Wed, 07 Nov 2012) | 3 lines Remove some debugging that accidentally made it in the last commit. ........ Merged revisions 375993-375994 from http://svn.asterisk.org/svn/asterisk/branches/1.8 ........ Merged revisions 375995 from http://svn.asterisk.org/svn/asterisk/branches/10 ........ Merged revisions 376014 from http://svn.asterisk.org/svn/asterisk/branches/11 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@376015 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-11-07 19:15:26 +00:00
remaining_time = ast_remaining_ms(start, timeout);
if (remaining_time <= 0) {
GENERIC_FAX_EXEC_ERROR(fax, chan, "TIMEOUT", "fax session timed-out");
break;
}
}
}
}
Multiple revisions 375993-375994 ........ r375993 | mmichelson | 2012-11-07 11:01:13 -0600 (Wed, 07 Nov 2012) | 30 lines Fix misuses of timeouts throughout the code. Prior to this change, a common method for determining if a timeout was reached was to call a function such as ast_waitfor_n() and inspect the out parameter that told how many milliseconds were left, then use that as the input to ast_waitfor_n() on the next go-around. The problem with this is that in some cases, submillisecond timeouts can occur, resulting in the out parameter not decreasing any. When this happens thousands of times, the result is that the timeout takes much longer than intended to be reached. As an example, I had a situation where a 3 second timeout took multiple days to finally end since most wakeups from ast_waitfor_n() were under a millisecond. This patch seeks to fix this pattern throughout the code. Now we log the time when an operation began and find the difference in wall clock time between now and when the event started. This means that sub-millisecond timeouts now cannot play havoc when trying to determine if something has timed out. Part of this fix also includes changing the function ast_waitfor() so that it is possible for it to return less than zero when a negative timeout is given to it. This makes it actually possible to detect errors in ast_waitfor() when there is no timeout. (closes issue ASTERISK-20414) reported by David M. Lee Review: https://reviewboard.asterisk.org/r/2135/ ........ r375994 | mmichelson | 2012-11-07 11:08:44 -0600 (Wed, 07 Nov 2012) | 3 lines Remove some debugging that accidentally made it in the last commit. ........ Merged revisions 375993-375994 from http://svn.asterisk.org/svn/asterisk/branches/1.8 ........ Merged revisions 375995 from http://svn.asterisk.org/svn/asterisk/branches/10 ........ Merged revisions 376014 from http://svn.asterisk.org/svn/asterisk/branches/11 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@376015 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-11-07 19:15:26 +00:00
ast_debug(3, "channel '%s' - event loop stopped { timeout: %d, remaining_time: %d }\n", ast_channel_name(chan), timeout, remaining_time);
set_channel_variables(chan, details);
ast_atomic_fetchadd_int(&faxregistry.fax_complete, 1);
if (!strcasecmp(details->result, "FAILED")) {
ast_atomic_fetchadd_int(&faxregistry.fax_failures, 1);
}
if (fax) {
ao2_unlink(faxregistry.container, fax);
ao2_ref(fax, -1);
}
/* if the channel is still alive, and we changed its read/write formats,
* restore them now
*/
if (chancount) {
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 (orig_read_format) {
ast_set_read_format(chan, orig_read_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
if (orig_write_format) {
ast_set_write_format(chan, orig_write_format);
}
}
/* return the chancount so the calling function can determine if the channel hungup during this FAX session or not */
return chancount;
}
static int receivefax_t38_init(struct ast_channel *chan, struct ast_fax_session_details *details)
{
Multiple revisions 375993-375994 ........ r375993 | mmichelson | 2012-11-07 11:01:13 -0600 (Wed, 07 Nov 2012) | 30 lines Fix misuses of timeouts throughout the code. Prior to this change, a common method for determining if a timeout was reached was to call a function such as ast_waitfor_n() and inspect the out parameter that told how many milliseconds were left, then use that as the input to ast_waitfor_n() on the next go-around. The problem with this is that in some cases, submillisecond timeouts can occur, resulting in the out parameter not decreasing any. When this happens thousands of times, the result is that the timeout takes much longer than intended to be reached. As an example, I had a situation where a 3 second timeout took multiple days to finally end since most wakeups from ast_waitfor_n() were under a millisecond. This patch seeks to fix this pattern throughout the code. Now we log the time when an operation began and find the difference in wall clock time between now and when the event started. This means that sub-millisecond timeouts now cannot play havoc when trying to determine if something has timed out. Part of this fix also includes changing the function ast_waitfor() so that it is possible for it to return less than zero when a negative timeout is given to it. This makes it actually possible to detect errors in ast_waitfor() when there is no timeout. (closes issue ASTERISK-20414) reported by David M. Lee Review: https://reviewboard.asterisk.org/r/2135/ ........ r375994 | mmichelson | 2012-11-07 11:08:44 -0600 (Wed, 07 Nov 2012) | 3 lines Remove some debugging that accidentally made it in the last commit. ........ Merged revisions 375993-375994 from http://svn.asterisk.org/svn/asterisk/branches/1.8 ........ Merged revisions 375995 from http://svn.asterisk.org/svn/asterisk/branches/10 ........ Merged revisions 376014 from http://svn.asterisk.org/svn/asterisk/branches/11 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@376015 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-11-07 19:15:26 +00:00
int timeout_ms;
struct ast_frame *frame = NULL;
struct ast_control_t38_parameters t38_parameters;
Multiple revisions 375993-375994 ........ r375993 | mmichelson | 2012-11-07 11:01:13 -0600 (Wed, 07 Nov 2012) | 30 lines Fix misuses of timeouts throughout the code. Prior to this change, a common method for determining if a timeout was reached was to call a function such as ast_waitfor_n() and inspect the out parameter that told how many milliseconds were left, then use that as the input to ast_waitfor_n() on the next go-around. The problem with this is that in some cases, submillisecond timeouts can occur, resulting in the out parameter not decreasing any. When this happens thousands of times, the result is that the timeout takes much longer than intended to be reached. As an example, I had a situation where a 3 second timeout took multiple days to finally end since most wakeups from ast_waitfor_n() were under a millisecond. This patch seeks to fix this pattern throughout the code. Now we log the time when an operation began and find the difference in wall clock time between now and when the event started. This means that sub-millisecond timeouts now cannot play havoc when trying to determine if something has timed out. Part of this fix also includes changing the function ast_waitfor() so that it is possible for it to return less than zero when a negative timeout is given to it. This makes it actually possible to detect errors in ast_waitfor() when there is no timeout. (closes issue ASTERISK-20414) reported by David M. Lee Review: https://reviewboard.asterisk.org/r/2135/ ........ r375994 | mmichelson | 2012-11-07 11:08:44 -0600 (Wed, 07 Nov 2012) | 3 lines Remove some debugging that accidentally made it in the last commit. ........ Merged revisions 375993-375994 from http://svn.asterisk.org/svn/asterisk/branches/1.8 ........ Merged revisions 375995 from http://svn.asterisk.org/svn/asterisk/branches/10 ........ Merged revisions 376014 from http://svn.asterisk.org/svn/asterisk/branches/11 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@376015 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-11-07 19:15:26 +00:00
struct timeval start;
int ms;
/* don't send any audio if we've already received a T.38 reinvite */
if (ast_channel_get_t38_state(chan) != T38_STATE_NEGOTIATING) {
/* generate 3 seconds of CED */
if (ast_playtones_start(chan, 1024, "!2100/3000", 1)) {
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_ERROR, "error generating CED tone on %s\n", ast_channel_name(chan));
return -1;
}
Multiple revisions 375993-375994 ........ r375993 | mmichelson | 2012-11-07 11:01:13 -0600 (Wed, 07 Nov 2012) | 30 lines Fix misuses of timeouts throughout the code. Prior to this change, a common method for determining if a timeout was reached was to call a function such as ast_waitfor_n() and inspect the out parameter that told how many milliseconds were left, then use that as the input to ast_waitfor_n() on the next go-around. The problem with this is that in some cases, submillisecond timeouts can occur, resulting in the out parameter not decreasing any. When this happens thousands of times, the result is that the timeout takes much longer than intended to be reached. As an example, I had a situation where a 3 second timeout took multiple days to finally end since most wakeups from ast_waitfor_n() were under a millisecond. This patch seeks to fix this pattern throughout the code. Now we log the time when an operation began and find the difference in wall clock time between now and when the event started. This means that sub-millisecond timeouts now cannot play havoc when trying to determine if something has timed out. Part of this fix also includes changing the function ast_waitfor() so that it is possible for it to return less than zero when a negative timeout is given to it. This makes it actually possible to detect errors in ast_waitfor() when there is no timeout. (closes issue ASTERISK-20414) reported by David M. Lee Review: https://reviewboard.asterisk.org/r/2135/ ........ r375994 | mmichelson | 2012-11-07 11:08:44 -0600 (Wed, 07 Nov 2012) | 3 lines Remove some debugging that accidentally made it in the last commit. ........ Merged revisions 375993-375994 from http://svn.asterisk.org/svn/asterisk/branches/1.8 ........ Merged revisions 375995 from http://svn.asterisk.org/svn/asterisk/branches/10 ........ Merged revisions 376014 from http://svn.asterisk.org/svn/asterisk/branches/11 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@376015 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-11-07 19:15:26 +00:00
timeout_ms = 3000;
start = ast_tvnow();
while ((ms = ast_remaining_ms(start, timeout_ms))) {
ms = ast_waitfor(chan, ms);
Multiple revisions 375993-375994 ........ r375993 | mmichelson | 2012-11-07 11:01:13 -0600 (Wed, 07 Nov 2012) | 30 lines Fix misuses of timeouts throughout the code. Prior to this change, a common method for determining if a timeout was reached was to call a function such as ast_waitfor_n() and inspect the out parameter that told how many milliseconds were left, then use that as the input to ast_waitfor_n() on the next go-around. The problem with this is that in some cases, submillisecond timeouts can occur, resulting in the out parameter not decreasing any. When this happens thousands of times, the result is that the timeout takes much longer than intended to be reached. As an example, I had a situation where a 3 second timeout took multiple days to finally end since most wakeups from ast_waitfor_n() were under a millisecond. This patch seeks to fix this pattern throughout the code. Now we log the time when an operation began and find the difference in wall clock time between now and when the event started. This means that sub-millisecond timeouts now cannot play havoc when trying to determine if something has timed out. Part of this fix also includes changing the function ast_waitfor() so that it is possible for it to return less than zero when a negative timeout is given to it. This makes it actually possible to detect errors in ast_waitfor() when there is no timeout. (closes issue ASTERISK-20414) reported by David M. Lee Review: https://reviewboard.asterisk.org/r/2135/ ........ r375994 | mmichelson | 2012-11-07 11:08:44 -0600 (Wed, 07 Nov 2012) | 3 lines Remove some debugging that accidentally made it in the last commit. ........ Merged revisions 375993-375994 from http://svn.asterisk.org/svn/asterisk/branches/1.8 ........ Merged revisions 375995 from http://svn.asterisk.org/svn/asterisk/branches/10 ........ Merged revisions 376014 from http://svn.asterisk.org/svn/asterisk/branches/11 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@376015 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-11-07 19:15:26 +00:00
if (ms < 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_ERROR, "error while generating CED tone on %s\n", ast_channel_name(chan));
ast_playtones_stop(chan);
return -1;
}
if (ms == 0) { /* all done, nothing happened */
break;
}
if (!(frame = ast_read(chan))) {
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_ERROR, "error reading frame while generating CED tone on %s\n", ast_channel_name(chan));
ast_playtones_stop(chan);
return -1;
}
if ((frame->frametype == AST_FRAME_CONTROL) &&
(frame->subclass.integer == AST_CONTROL_T38_PARAMETERS) &&
(frame->datalen == sizeof(t38_parameters))) {
struct ast_control_t38_parameters *parameters = frame->data.ptr;
switch (parameters->request_response) {
case AST_T38_REQUEST_NEGOTIATE:
/* the other end has requested a switch to T.38, so reply that we are willing, if we can
* do T.38 as well
*/
t38_parameters_fax_to_ast(&t38_parameters, &details->our_t38_parameters);
t38_parameters.request_response = (details->caps & AST_FAX_TECH_T38) ? AST_T38_NEGOTIATED : AST_T38_REFUSED;
ast_indicate_data(chan, AST_CONTROL_T38_PARAMETERS, &t38_parameters, sizeof(t38_parameters));
ast_playtones_stop(chan);
break;
case AST_T38_NEGOTIATED:
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_debug(1, "Negotiated T.38 for receive on %s\n", ast_channel_name(chan));
t38_parameters_ast_to_fax(&details->their_t38_parameters, parameters);
details->caps &= ~AST_FAX_TECH_AUDIO;
report_fax_status(chan, details, "T.38 Negotiated");
break;
default:
break;
}
}
ast_frfree(frame);
}
ast_playtones_stop(chan);
}
/* if T.38 was negotiated, we are done initializing */
if (ast_channel_get_t38_state(chan) == T38_STATE_NEGOTIATED) {
return 0;
}
/* request T.38 */
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_debug(1, "Negotiating T.38 for receive on %s\n", ast_channel_name(chan));
/* wait for negotiation to complete */
timeout_ms = details->t38timeout;
/* set parameters based on the session's parameters */
t38_parameters_fax_to_ast(&t38_parameters, &details->our_t38_parameters);
t38_parameters.request_response = AST_T38_REQUEST_NEGOTIATE;
if ((ast_indicate_data(chan, AST_CONTROL_T38_PARAMETERS, &t38_parameters, sizeof(t38_parameters)) != 0)) {
return -1;
}
Multiple revisions 375993-375994 ........ r375993 | mmichelson | 2012-11-07 11:01:13 -0600 (Wed, 07 Nov 2012) | 30 lines Fix misuses of timeouts throughout the code. Prior to this change, a common method for determining if a timeout was reached was to call a function such as ast_waitfor_n() and inspect the out parameter that told how many milliseconds were left, then use that as the input to ast_waitfor_n() on the next go-around. The problem with this is that in some cases, submillisecond timeouts can occur, resulting in the out parameter not decreasing any. When this happens thousands of times, the result is that the timeout takes much longer than intended to be reached. As an example, I had a situation where a 3 second timeout took multiple days to finally end since most wakeups from ast_waitfor_n() were under a millisecond. This patch seeks to fix this pattern throughout the code. Now we log the time when an operation began and find the difference in wall clock time between now and when the event started. This means that sub-millisecond timeouts now cannot play havoc when trying to determine if something has timed out. Part of this fix also includes changing the function ast_waitfor() so that it is possible for it to return less than zero when a negative timeout is given to it. This makes it actually possible to detect errors in ast_waitfor() when there is no timeout. (closes issue ASTERISK-20414) reported by David M. Lee Review: https://reviewboard.asterisk.org/r/2135/ ........ r375994 | mmichelson | 2012-11-07 11:08:44 -0600 (Wed, 07 Nov 2012) | 3 lines Remove some debugging that accidentally made it in the last commit. ........ Merged revisions 375993-375994 from http://svn.asterisk.org/svn/asterisk/branches/1.8 ........ Merged revisions 375995 from http://svn.asterisk.org/svn/asterisk/branches/10 ........ Merged revisions 376014 from http://svn.asterisk.org/svn/asterisk/branches/11 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@376015 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-11-07 19:15:26 +00:00
start = ast_tvnow();
while ((ms = ast_remaining_ms(start, timeout_ms))) {
int break_loop = 0;
ms = ast_waitfor(chan, ms);
if (ms < 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_${fieldname}' and then writing setters/getters and converting the existing code to using them. When all fields are done, we can move ast_channel to a C file from channel.h and lop off the '__do_not_use_'. This patch sets up main/channel_interal_api.c to be the only file that actually accesses the ast_channel's fields directly. The intent would be for any API functions in channel.c to use the accessor functions. No more monkeying around with channel internals. We should use our own APIs. The interesting changes in this patch are the addition of channel_internal_api.c, the moving of the AST_DATA stuff from channel.c to channel_internal_api.c (note: the AST_DATA stuff will have to be reworked to use accessor functions when ast_channel is really opaque), and some re-working of the way channel iterators/callbacks are handled so as to avoid creating fake ast_channels on the stack to pass in matching data by directly accessing fields (since "name" is a stringfield and the fake channel doesn't init the stringfields, you can't use the ast_channel_name_set() function). I went with ast_channel_name(chan) for a getter, and ast_channel_name_set(chan, name) for a setter. The majority of the grunt-work for this change was done by writing a semantic patch using Coccinelle ( http://coccinelle.lip6.fr/ ). Review: https://reviewboard.asterisk.org/r/1655/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@350223 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-01-09 22:15:50 +00:00
ast_log(LOG_WARNING, "error on '%s' while waiting for T.38 negotiation.\n", ast_channel_name(chan));
return -1;
}
if (ms == 0) { /* all done, nothing happened */
Replace direct access to channel name with accessor functions There are many benefits to making the ast_channel an opaque handle, from increasing maintainability to presenting ways to kill masquerades. This patch kicks things off by taking things a field at a time, renaming the field to '__do_not_use_${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' timed-out during the T.38 negotiation.\n", ast_channel_name(chan));
details->caps &= ~AST_FAX_TECH_T38;
break;
}
if (!(frame = ast_read(chan))) {
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, "error on '%s' while waiting for T.38 negotiation.\n", ast_channel_name(chan));
return -1;
}
if ((frame->frametype == AST_FRAME_CONTROL) &&
(frame->subclass.integer == AST_CONTROL_T38_PARAMETERS) &&
(frame->datalen == sizeof(t38_parameters))) {
struct ast_control_t38_parameters *parameters = frame->data.ptr;
switch (parameters->request_response) {
case AST_T38_REQUEST_NEGOTIATE:
t38_parameters_fax_to_ast(&t38_parameters, &details->our_t38_parameters);
t38_parameters.request_response = AST_T38_NEGOTIATED;
ast_indicate_data(chan, AST_CONTROL_T38_PARAMETERS, &t38_parameters, sizeof(t38_parameters));
break;
case AST_T38_NEGOTIATED:
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_debug(1, "Negotiated T.38 for receive on %s\n", ast_channel_name(chan));
t38_parameters_ast_to_fax(&details->their_t38_parameters, parameters);
details->caps &= ~AST_FAX_TECH_AUDIO;
report_fax_status(chan, details, "T.38 Negotiated");
Multiple revisions 375993-375994 ........ r375993 | mmichelson | 2012-11-07 11:01:13 -0600 (Wed, 07 Nov 2012) | 30 lines Fix misuses of timeouts throughout the code. Prior to this change, a common method for determining if a timeout was reached was to call a function such as ast_waitfor_n() and inspect the out parameter that told how many milliseconds were left, then use that as the input to ast_waitfor_n() on the next go-around. The problem with this is that in some cases, submillisecond timeouts can occur, resulting in the out parameter not decreasing any. When this happens thousands of times, the result is that the timeout takes much longer than intended to be reached. As an example, I had a situation where a 3 second timeout took multiple days to finally end since most wakeups from ast_waitfor_n() were under a millisecond. This patch seeks to fix this pattern throughout the code. Now we log the time when an operation began and find the difference in wall clock time between now and when the event started. This means that sub-millisecond timeouts now cannot play havoc when trying to determine if something has timed out. Part of this fix also includes changing the function ast_waitfor() so that it is possible for it to return less than zero when a negative timeout is given to it. This makes it actually possible to detect errors in ast_waitfor() when there is no timeout. (closes issue ASTERISK-20414) reported by David M. Lee Review: https://reviewboard.asterisk.org/r/2135/ ........ r375994 | mmichelson | 2012-11-07 11:08:44 -0600 (Wed, 07 Nov 2012) | 3 lines Remove some debugging that accidentally made it in the last commit. ........ Merged revisions 375993-375994 from http://svn.asterisk.org/svn/asterisk/branches/1.8 ........ Merged revisions 375995 from http://svn.asterisk.org/svn/asterisk/branches/10 ........ Merged revisions 376014 from http://svn.asterisk.org/svn/asterisk/branches/11 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@376015 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-11-07 19:15:26 +00:00
break_loop = 1;
break;
case AST_T38_REFUSED:
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' refused to negotiate T.38\n", ast_channel_name(chan));
details->caps &= ~AST_FAX_TECH_T38;
Multiple revisions 375993-375994 ........ r375993 | mmichelson | 2012-11-07 11:01:13 -0600 (Wed, 07 Nov 2012) | 30 lines Fix misuses of timeouts throughout the code. Prior to this change, a common method for determining if a timeout was reached was to call a function such as ast_waitfor_n() and inspect the out parameter that told how many milliseconds were left, then use that as the input to ast_waitfor_n() on the next go-around. The problem with this is that in some cases, submillisecond timeouts can occur, resulting in the out parameter not decreasing any. When this happens thousands of times, the result is that the timeout takes much longer than intended to be reached. As an example, I had a situation where a 3 second timeout took multiple days to finally end since most wakeups from ast_waitfor_n() were under a millisecond. This patch seeks to fix this pattern throughout the code. Now we log the time when an operation began and find the difference in wall clock time between now and when the event started. This means that sub-millisecond timeouts now cannot play havoc when trying to determine if something has timed out. Part of this fix also includes changing the function ast_waitfor() so that it is possible for it to return less than zero when a negative timeout is given to it. This makes it actually possible to detect errors in ast_waitfor() when there is no timeout. (closes issue ASTERISK-20414) reported by David M. Lee Review: https://reviewboard.asterisk.org/r/2135/ ........ r375994 | mmichelson | 2012-11-07 11:08:44 -0600 (Wed, 07 Nov 2012) | 3 lines Remove some debugging that accidentally made it in the last commit. ........ Merged revisions 375993-375994 from http://svn.asterisk.org/svn/asterisk/branches/1.8 ........ Merged revisions 375995 from http://svn.asterisk.org/svn/asterisk/branches/10 ........ Merged revisions 376014 from http://svn.asterisk.org/svn/asterisk/branches/11 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@376015 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-11-07 19:15:26 +00:00
break_loop = 1;
break;
default:
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_ERROR, "channel '%s' failed to negotiate T.38\n", ast_channel_name(chan));
details->caps &= ~AST_FAX_TECH_T38;
Multiple revisions 375993-375994 ........ r375993 | mmichelson | 2012-11-07 11:01:13 -0600 (Wed, 07 Nov 2012) | 30 lines Fix misuses of timeouts throughout the code. Prior to this change, a common method for determining if a timeout was reached was to call a function such as ast_waitfor_n() and inspect the out parameter that told how many milliseconds were left, then use that as the input to ast_waitfor_n() on the next go-around. The problem with this is that in some cases, submillisecond timeouts can occur, resulting in the out parameter not decreasing any. When this happens thousands of times, the result is that the timeout takes much longer than intended to be reached. As an example, I had a situation where a 3 second timeout took multiple days to finally end since most wakeups from ast_waitfor_n() were under a millisecond. This patch seeks to fix this pattern throughout the code. Now we log the time when an operation began and find the difference in wall clock time between now and when the event started. This means that sub-millisecond timeouts now cannot play havoc when trying to determine if something has timed out. Part of this fix also includes changing the function ast_waitfor() so that it is possible for it to return less than zero when a negative timeout is given to it. This makes it actually possible to detect errors in ast_waitfor() when there is no timeout. (closes issue ASTERISK-20414) reported by David M. Lee Review: https://reviewboard.asterisk.org/r/2135/ ........ r375994 | mmichelson | 2012-11-07 11:08:44 -0600 (Wed, 07 Nov 2012) | 3 lines Remove some debugging that accidentally made it in the last commit. ........ Merged revisions 375993-375994 from http://svn.asterisk.org/svn/asterisk/branches/1.8 ........ Merged revisions 375995 from http://svn.asterisk.org/svn/asterisk/branches/10 ........ Merged revisions 376014 from http://svn.asterisk.org/svn/asterisk/branches/11 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@376015 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-11-07 19:15:26 +00:00
break_loop = 1;
break;
}
}
ast_frfree(frame);
Multiple revisions 375993-375994 ........ r375993 | mmichelson | 2012-11-07 11:01:13 -0600 (Wed, 07 Nov 2012) | 30 lines Fix misuses of timeouts throughout the code. Prior to this change, a common method for determining if a timeout was reached was to call a function such as ast_waitfor_n() and inspect the out parameter that told how many milliseconds were left, then use that as the input to ast_waitfor_n() on the next go-around. The problem with this is that in some cases, submillisecond timeouts can occur, resulting in the out parameter not decreasing any. When this happens thousands of times, the result is that the timeout takes much longer than intended to be reached. As an example, I had a situation where a 3 second timeout took multiple days to finally end since most wakeups from ast_waitfor_n() were under a millisecond. This patch seeks to fix this pattern throughout the code. Now we log the time when an operation began and find the difference in wall clock time between now and when the event started. This means that sub-millisecond timeouts now cannot play havoc when trying to determine if something has timed out. Part of this fix also includes changing the function ast_waitfor() so that it is possible for it to return less than zero when a negative timeout is given to it. This makes it actually possible to detect errors in ast_waitfor() when there is no timeout. (closes issue ASTERISK-20414) reported by David M. Lee Review: https://reviewboard.asterisk.org/r/2135/ ........ r375994 | mmichelson | 2012-11-07 11:08:44 -0600 (Wed, 07 Nov 2012) | 3 lines Remove some debugging that accidentally made it in the last commit. ........ Merged revisions 375993-375994 from http://svn.asterisk.org/svn/asterisk/branches/1.8 ........ Merged revisions 375995 from http://svn.asterisk.org/svn/asterisk/branches/10 ........ Merged revisions 376014 from http://svn.asterisk.org/svn/asterisk/branches/11 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@376015 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-11-07 19:15:26 +00:00
if (break_loop) {
break;
}
}
/* if T.38 was negotiated, we are done initializing */
if (ast_channel_get_t38_state(chan) == T38_STATE_NEGOTIATED) {
return 0;
}
/* if we made it here, then T.38 failed, check the 'f' flag */
if (details->option.allow_audio != AST_FAX_OPTFLAG_TRUE) {
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, "Audio FAX not allowed on channel '%s' and T.38 negotiation failed; aborting.\n", ast_channel_name(chan));
return -1;
}
/* ok, audio fallback is allowed */
details->caps |= AST_FAX_TECH_AUDIO;
return 0;
}
/*! \brief Report on the final state of a receive fax operation
* \note This will lock the \ref ast_channel
*/
static int report_receive_fax_status(struct ast_channel *chan, const char *filename)
{
RAII_VAR(struct ast_json *, json_object, NULL, ast_json_unref);
RAII_VAR(struct stasis_message *, message, NULL, ao2_cleanup);
RAII_VAR(struct ast_json *, json_array, ast_json_array_create(), ast_json_unref);
struct ast_json *json_filename = ast_json_string_create(filename);
if (!json_array || !json_filename) {
ast_json_unref(json_filename);
return -1;
}
ast_json_array_append(json_array, json_filename);
{
const char *remote_station_id;
const char *local_station_id;
const char *fax_pages;
const char *fax_resolution;
const char *fax_bitrate;
SCOPED_CHANNELLOCK(lock, chan);
remote_station_id = AST_JSON_UTF8_VALIDATE(pbx_builtin_getvar_helper(chan, "REMOTESTATIONID"));
if (!ast_strlen_zero(remote_station_id)) {
remote_station_id = ast_strdupa(remote_station_id);
}
local_station_id = AST_JSON_UTF8_VALIDATE(pbx_builtin_getvar_helper(chan, "LOCALSTATIONID"));
if (!ast_strlen_zero(local_station_id)) {
local_station_id = ast_strdupa(local_station_id);
}
fax_pages = S_OR(pbx_builtin_getvar_helper(chan, "FAXPAGES"), "");
if (!ast_strlen_zero(fax_pages)) {
fax_pages = ast_strdupa(fax_pages);
}
fax_resolution = S_OR(pbx_builtin_getvar_helper(chan, "FAXRESOLUTION"), "");
if (!ast_strlen_zero(fax_resolution)) {
fax_resolution = ast_strdupa(fax_resolution);
}
fax_bitrate = S_OR(pbx_builtin_getvar_helper(chan, "FAXBITRATE"), "");
if (!ast_strlen_zero(fax_bitrate)) {
fax_bitrate = ast_strdupa(fax_bitrate);
}
json_object = ast_json_pack("{s: s, s: s, s: s, s: s, s: s, s: s, s: o}",
"type", "receive",
"remote_station_id", S_OR(remote_station_id, ""),
"local_station_id", S_OR(local_station_id, ""),
"fax_pages", S_OR(fax_pages, ""),
"fax_resolution", S_OR(fax_resolution, ""),
"fax_bitrate", S_OR(fax_bitrate, ""),
"filenames", ast_json_ref(json_array));
if (!json_object) {
return -1;
}
message = ast_channel_blob_create_from_cache(ast_channel_uniqueid(chan), ast_channel_fax_type(), json_object);
if (!message) {
return -1;
}
stasis_publish(ast_channel_topic(chan), message);
}
return 0;
}
/*! \brief initiate a receive FAX session */
static int receivefax_exec(struct ast_channel *chan, const char *data)
{
char *parse, modems[128] = "";
int channel_alive;
RAII_VAR(struct ast_fax_session *, s, NULL, ao2_cleanup);
RAII_VAR(struct ast_fax_session_details *, details, NULL, ao2_cleanup);
struct ast_fax_tech_token *token = NULL;
struct ast_fax_document *doc;
AST_DECLARE_APP_ARGS(args,
AST_APP_ARG(filename);
AST_APP_ARG(options);
);
struct ast_flags opts = { 0, };
enum ast_t38_state t38state;
/* initialize output channel variables */
pbx_builtin_setvar_helper(chan, "FAXSTATUS", "FAILED");
pbx_builtin_setvar_helper(chan, "REMOTESTATIONID", NULL);
pbx_builtin_setvar_helper(chan, "FAXPAGES", "0");
pbx_builtin_setvar_helper(chan, "FAXBITRATE", NULL);
pbx_builtin_setvar_helper(chan, "FAXRESOLUTION", NULL);
pbx_builtin_setvar_helper(chan, "FAXMODE", NULL);
/* Get a FAX session details structure from the channel's FAX datastore and create one if
* it does not already exist. */
if (!(details = find_or_create_details(chan))) {
pbx_builtin_setvar_helper(chan, "FAXERROR", "MEMORY_ERROR");
pbx_builtin_setvar_helper(chan, "FAXSTATUSSTRING", "error allocating memory");
ast_log(LOG_ERROR, "System cannot provide memory for session requirements.\n");
return -1;
}
ast_string_field_set(details, result, "FAILED");
ast_string_field_set(details, resultstr, "error starting fax session");
ast_string_field_set(details, error, "INIT_ERROR");
set_channel_variables(chan, details);
if (details->gateway_id > 0) {
ast_string_field_set(details, resultstr, "can't receive a fax on a channel with a T.38 gateway");
set_channel_variables(chan, details);
ast_log(LOG_ERROR, "executing ReceiveFAX on a channel with a T.38 Gateway is not supported\n");
return -1;
}
if (details->maxrate < details->minrate) {
ast_string_field_set(details, error, "INVALID_ARGUMENTS");
ast_string_field_set(details, resultstr, "maxrate is less than minrate");
set_channel_variables(chan, details);
ast_log(LOG_ERROR, "maxrate %u is less than minrate %u\n", details->maxrate, details->minrate);
return -1;
}
if (check_modem_rate(details->modems, details->minrate)) {
ast_fax_modem_to_str(details->modems, modems, sizeof(modems));
ast_log(LOG_ERROR, "'modems' setting '%s' is incompatible with 'minrate' setting %u\n", modems, details->minrate);
ast_string_field_set(details, error, "INVALID_ARGUMENTS");
ast_string_field_set(details, resultstr, "incompatible 'modems' and 'minrate' settings");
set_channel_variables(chan, details);
return -1;
}
if (check_modem_rate(details->modems, details->maxrate)) {
ast_fax_modem_to_str(details->modems, modems, sizeof(modems));
ast_log(LOG_ERROR, "'modems' setting '%s' is incompatible with 'maxrate' setting %u\n", modems, details->maxrate);
ast_string_field_set(details, error, "INVALID_ARGUMENTS");
ast_string_field_set(details, resultstr, "incompatible 'modems' and 'maxrate' settings");
set_channel_variables(chan, details);
return -1;
}
if (ast_strlen_zero(data)) {
ast_string_field_set(details, error, "INVALID_ARGUMENTS");
ast_string_field_set(details, resultstr, "invalid arguments");
set_channel_variables(chan, details);
ast_log(LOG_WARNING, "%s requires an argument (filename[,options])\n", app_receivefax);
return -1;
}
parse = ast_strdupa(data);
AST_STANDARD_APP_ARGS(args, parse);
if (!ast_strlen_zero(args.options) &&
ast_app_parse_options(fax_exec_options, &opts, NULL, args.options)) {
ast_string_field_set(details, error, "INVALID_ARGUMENTS");
ast_string_field_set(details, resultstr, "invalid arguments");
set_channel_variables(chan, details);
return -1;
}
if (ast_strlen_zero(args.filename)) {
ast_string_field_set(details, error, "INVALID_ARGUMENTS");
ast_string_field_set(details, resultstr, "invalid arguments");
set_channel_variables(chan, details);
ast_log(LOG_WARNING, "%s requires an argument (filename[,options])\n", app_receivefax);
return -1;
}
/* check for unsupported FAX application options */
if (ast_test_flag(&opts, OPT_CALLERMODE) || ast_test_flag(&opts, OPT_CALLEDMODE)) {
ast_string_field_set(details, error, "INVALID_ARGUMENTS");
ast_string_field_set(details, resultstr, "invalid arguments");
set_channel_variables(chan, details);
ast_log(LOG_WARNING, "%s does not support polling\n", app_receivefax);
return -1;
}
ast_atomic_fetchadd_int(&faxregistry.fax_rx_attempts, 1);
pbx_builtin_setvar_helper(chan, "FAXERROR", "Channel Problems");
pbx_builtin_setvar_helper(chan, "FAXSTATUSSTRING", "Error before FAX transmission started.");
if (!(doc = ast_calloc(1, sizeof(*doc) + strlen(args.filename) + 1))) {
ast_string_field_set(details, error, "MEMORY_ERROR");
ast_string_field_set(details, resultstr, "error allocating memory");
set_channel_variables(chan, details);
ast_log(LOG_ERROR, "System cannot provide memory for session requirements.\n");
return -1;
}
strcpy(doc->filename, args.filename);
AST_LIST_INSERT_TAIL(&details->documents, doc, next);
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_verb(3, "Channel '%s' receiving FAX '%s'\n", ast_channel_name(chan), args.filename);
details->caps = AST_FAX_TECH_RECEIVE;
details->option.send_ced = AST_FAX_OPTFLAG_TRUE;
/* check for debug */
if (ast_test_flag(&opts, OPT_DEBUG) || global_fax_debug) {
details->option.debug = AST_FAX_OPTFLAG_TRUE;
}
/* check for request for status events */
if (ast_test_flag(&opts, OPT_STATUS)) {
details->option.statusevents = AST_FAX_OPTFLAG_TRUE;
}
t38state = ast_channel_get_t38_state(chan);
if ((t38state == T38_STATE_UNAVAILABLE) || (t38state == T38_STATE_REJECTED) ||
ast_test_flag(&opts, OPT_ALLOWAUDIO) ||
ast_test_flag(&opts, OPT_FORCE_AUDIO)) {
details->option.allow_audio = AST_FAX_OPTFLAG_TRUE;
}
if (!(s = fax_session_reserve(details, &token))) {
ast_string_field_set(details, resultstr, "error reserving fax session");
set_channel_variables(chan, details);
ast_log(LOG_ERROR, "Unable to reserve FAX session.\n");
return -1;
}
/* make sure the channel is up */
if (ast_channel_state(chan) != AST_STATE_UP) {
if (ast_answer(chan)) {
ast_string_field_set(details, resultstr, "error answering channel");
set_channel_variables(chan, details);
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 answer attempt.\n", ast_channel_name(chan));
fax_session_release(s, token);
return -1;
}
}
if (!ast_test_flag(&opts, OPT_FORCE_AUDIO)) {
if (set_fax_t38_caps(chan, details)) {
ast_string_field_set(details, error, "T38_NEG_ERROR");
ast_string_field_set(details, resultstr, "error negotiating T.38");
set_channel_variables(chan, details);
fax_session_release(s, token);
return -1;
}
} else {
details->caps |= AST_FAX_TECH_AUDIO;
}
if (!ast_test_flag(&opts, OPT_FORCE_AUDIO) && (details->caps & AST_FAX_TECH_T38)) {
if (receivefax_t38_init(chan, details)) {
ast_string_field_set(details, error, "T38_NEG_ERROR");
ast_string_field_set(details, resultstr, "error negotiating T.38");
set_channel_variables(chan, details);
fax_session_release(s, token);
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_ERROR, "error initializing channel '%s' in T.38 mode\n", ast_channel_name(chan));
return -1;
}
}
if ((channel_alive = generic_fax_exec(chan, details, s, token)) < 0) {
ast_atomic_fetchadd_int(&faxregistry.fax_failures, 1);
}
if (ast_channel_get_t38_state(chan) == T38_STATE_NEGOTIATED) {
if (disable_t38(chan)) {
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_debug(1, "error disabling T.38 mode on %s\n", ast_channel_name(chan));
}
}
if (report_receive_fax_status(chan, args.filename)) {
ast_log(AST_LOG_ERROR, "Error publishing ReceiveFax status message\n");
}
/* If the channel hungup return -1; otherwise, return 0 to continue in the dialplan */
return (!channel_alive) ? -1 : 0;
}
static int sendfax_t38_init(struct ast_channel *chan, struct ast_fax_session_details *details)
{
Multiple revisions 375993-375994 ........ r375993 | mmichelson | 2012-11-07 11:01:13 -0600 (Wed, 07 Nov 2012) | 30 lines Fix misuses of timeouts throughout the code. Prior to this change, a common method for determining if a timeout was reached was to call a function such as ast_waitfor_n() and inspect the out parameter that told how many milliseconds were left, then use that as the input to ast_waitfor_n() on the next go-around. The problem with this is that in some cases, submillisecond timeouts can occur, resulting in the out parameter not decreasing any. When this happens thousands of times, the result is that the timeout takes much longer than intended to be reached. As an example, I had a situation where a 3 second timeout took multiple days to finally end since most wakeups from ast_waitfor_n() were under a millisecond. This patch seeks to fix this pattern throughout the code. Now we log the time when an operation began and find the difference in wall clock time between now and when the event started. This means that sub-millisecond timeouts now cannot play havoc when trying to determine if something has timed out. Part of this fix also includes changing the function ast_waitfor() so that it is possible for it to return less than zero when a negative timeout is given to it. This makes it actually possible to detect errors in ast_waitfor() when there is no timeout. (closes issue ASTERISK-20414) reported by David M. Lee Review: https://reviewboard.asterisk.org/r/2135/ ........ r375994 | mmichelson | 2012-11-07 11:08:44 -0600 (Wed, 07 Nov 2012) | 3 lines Remove some debugging that accidentally made it in the last commit. ........ Merged revisions 375993-375994 from http://svn.asterisk.org/svn/asterisk/branches/1.8 ........ Merged revisions 375995 from http://svn.asterisk.org/svn/asterisk/branches/10 ........ Merged revisions 376014 from http://svn.asterisk.org/svn/asterisk/branches/11 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@376015 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-11-07 19:15:26 +00:00
int timeout_ms;
struct ast_frame *frame = NULL;
struct ast_control_t38_parameters t38_parameters;
Multiple revisions 375993-375994 ........ r375993 | mmichelson | 2012-11-07 11:01:13 -0600 (Wed, 07 Nov 2012) | 30 lines Fix misuses of timeouts throughout the code. Prior to this change, a common method for determining if a timeout was reached was to call a function such as ast_waitfor_n() and inspect the out parameter that told how many milliseconds were left, then use that as the input to ast_waitfor_n() on the next go-around. The problem with this is that in some cases, submillisecond timeouts can occur, resulting in the out parameter not decreasing any. When this happens thousands of times, the result is that the timeout takes much longer than intended to be reached. As an example, I had a situation where a 3 second timeout took multiple days to finally end since most wakeups from ast_waitfor_n() were under a millisecond. This patch seeks to fix this pattern throughout the code. Now we log the time when an operation began and find the difference in wall clock time between now and when the event started. This means that sub-millisecond timeouts now cannot play havoc when trying to determine if something has timed out. Part of this fix also includes changing the function ast_waitfor() so that it is possible for it to return less than zero when a negative timeout is given to it. This makes it actually possible to detect errors in ast_waitfor() when there is no timeout. (closes issue ASTERISK-20414) reported by David M. Lee Review: https://reviewboard.asterisk.org/r/2135/ ........ r375994 | mmichelson | 2012-11-07 11:08:44 -0600 (Wed, 07 Nov 2012) | 3 lines Remove some debugging that accidentally made it in the last commit. ........ Merged revisions 375993-375994 from http://svn.asterisk.org/svn/asterisk/branches/1.8 ........ Merged revisions 375995 from http://svn.asterisk.org/svn/asterisk/branches/10 ........ Merged revisions 376014 from http://svn.asterisk.org/svn/asterisk/branches/11 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@376015 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-11-07 19:15:26 +00:00
struct timeval start;
int ms;
/* send CNG tone while listening for the receiver to initiate a switch
* to T.38 mode; if they do, stop sending the CNG tone and proceed with
* the switch.
*
* 10500 is enough time for 3 CNG tones
*/
Multiple revisions 375993-375994 ........ r375993 | mmichelson | 2012-11-07 11:01:13 -0600 (Wed, 07 Nov 2012) | 30 lines Fix misuses of timeouts throughout the code. Prior to this change, a common method for determining if a timeout was reached was to call a function such as ast_waitfor_n() and inspect the out parameter that told how many milliseconds were left, then use that as the input to ast_waitfor_n() on the next go-around. The problem with this is that in some cases, submillisecond timeouts can occur, resulting in the out parameter not decreasing any. When this happens thousands of times, the result is that the timeout takes much longer than intended to be reached. As an example, I had a situation where a 3 second timeout took multiple days to finally end since most wakeups from ast_waitfor_n() were under a millisecond. This patch seeks to fix this pattern throughout the code. Now we log the time when an operation began and find the difference in wall clock time between now and when the event started. This means that sub-millisecond timeouts now cannot play havoc when trying to determine if something has timed out. Part of this fix also includes changing the function ast_waitfor() so that it is possible for it to return less than zero when a negative timeout is given to it. This makes it actually possible to detect errors in ast_waitfor() when there is no timeout. (closes issue ASTERISK-20414) reported by David M. Lee Review: https://reviewboard.asterisk.org/r/2135/ ........ r375994 | mmichelson | 2012-11-07 11:08:44 -0600 (Wed, 07 Nov 2012) | 3 lines Remove some debugging that accidentally made it in the last commit. ........ Merged revisions 375993-375994 from http://svn.asterisk.org/svn/asterisk/branches/1.8 ........ Merged revisions 375995 from http://svn.asterisk.org/svn/asterisk/branches/10 ........ Merged revisions 376014 from http://svn.asterisk.org/svn/asterisk/branches/11 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@376015 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-11-07 19:15:26 +00:00
timeout_ms = 10500;
/* don't send any audio if we've already received a T.38 reinvite */
if (ast_channel_get_t38_state(chan) != T38_STATE_NEGOTIATING) {
if (ast_playtones_start(chan, 1024, "!1100/500,!0/3000,!1100/500,!0/3000,!1100/500,!0/3000", 1)) {
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_ERROR, "error generating CNG tone on %s\n", ast_channel_name(chan));
return -1;
}
}
Multiple revisions 375993-375994 ........ r375993 | mmichelson | 2012-11-07 11:01:13 -0600 (Wed, 07 Nov 2012) | 30 lines Fix misuses of timeouts throughout the code. Prior to this change, a common method for determining if a timeout was reached was to call a function such as ast_waitfor_n() and inspect the out parameter that told how many milliseconds were left, then use that as the input to ast_waitfor_n() on the next go-around. The problem with this is that in some cases, submillisecond timeouts can occur, resulting in the out parameter not decreasing any. When this happens thousands of times, the result is that the timeout takes much longer than intended to be reached. As an example, I had a situation where a 3 second timeout took multiple days to finally end since most wakeups from ast_waitfor_n() were under a millisecond. This patch seeks to fix this pattern throughout the code. Now we log the time when an operation began and find the difference in wall clock time between now and when the event started. This means that sub-millisecond timeouts now cannot play havoc when trying to determine if something has timed out. Part of this fix also includes changing the function ast_waitfor() so that it is possible for it to return less than zero when a negative timeout is given to it. This makes it actually possible to detect errors in ast_waitfor() when there is no timeout. (closes issue ASTERISK-20414) reported by David M. Lee Review: https://reviewboard.asterisk.org/r/2135/ ........ r375994 | mmichelson | 2012-11-07 11:08:44 -0600 (Wed, 07 Nov 2012) | 3 lines Remove some debugging that accidentally made it in the last commit. ........ Merged revisions 375993-375994 from http://svn.asterisk.org/svn/asterisk/branches/1.8 ........ Merged revisions 375995 from http://svn.asterisk.org/svn/asterisk/branches/10 ........ Merged revisions 376014 from http://svn.asterisk.org/svn/asterisk/branches/11 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@376015 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-11-07 19:15:26 +00:00
start = ast_tvnow();
while ((ms = ast_remaining_ms(start, timeout_ms))) {
int break_loop = 0;
ms = ast_waitfor(chan, ms);
Multiple revisions 375993-375994 ........ r375993 | mmichelson | 2012-11-07 11:01:13 -0600 (Wed, 07 Nov 2012) | 30 lines Fix misuses of timeouts throughout the code. Prior to this change, a common method for determining if a timeout was reached was to call a function such as ast_waitfor_n() and inspect the out parameter that told how many milliseconds were left, then use that as the input to ast_waitfor_n() on the next go-around. The problem with this is that in some cases, submillisecond timeouts can occur, resulting in the out parameter not decreasing any. When this happens thousands of times, the result is that the timeout takes much longer than intended to be reached. As an example, I had a situation where a 3 second timeout took multiple days to finally end since most wakeups from ast_waitfor_n() were under a millisecond. This patch seeks to fix this pattern throughout the code. Now we log the time when an operation began and find the difference in wall clock time between now and when the event started. This means that sub-millisecond timeouts now cannot play havoc when trying to determine if something has timed out. Part of this fix also includes changing the function ast_waitfor() so that it is possible for it to return less than zero when a negative timeout is given to it. This makes it actually possible to detect errors in ast_waitfor() when there is no timeout. (closes issue ASTERISK-20414) reported by David M. Lee Review: https://reviewboard.asterisk.org/r/2135/ ........ r375994 | mmichelson | 2012-11-07 11:08:44 -0600 (Wed, 07 Nov 2012) | 3 lines Remove some debugging that accidentally made it in the last commit. ........ Merged revisions 375993-375994 from http://svn.asterisk.org/svn/asterisk/branches/1.8 ........ Merged revisions 375995 from http://svn.asterisk.org/svn/asterisk/branches/10 ........ Merged revisions 376014 from http://svn.asterisk.org/svn/asterisk/branches/11 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@376015 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-11-07 19:15:26 +00:00
if (ms < 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_ERROR, "error while generating CNG tone on %s\n", ast_channel_name(chan));
ast_playtones_stop(chan);
return -1;
}
if (ms == 0) { /* all done, nothing happened */
break;
}
if (!(frame = ast_read(chan))) {
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_ERROR, "error reading frame while generating CNG tone on %s\n", ast_channel_name(chan));
ast_playtones_stop(chan);
return -1;
}
if ((frame->frametype == AST_FRAME_CONTROL) &&
(frame->subclass.integer == AST_CONTROL_T38_PARAMETERS) &&
(frame->datalen == sizeof(t38_parameters))) {
struct ast_control_t38_parameters *parameters = frame->data.ptr;
switch (parameters->request_response) {
case AST_T38_REQUEST_NEGOTIATE:
/* the other end has requested a switch to T.38, so reply that we are willing, if we can
* do T.38 as well
*/
t38_parameters_fax_to_ast(&t38_parameters, &details->our_t38_parameters);
t38_parameters.request_response = (details->caps & AST_FAX_TECH_T38) ? AST_T38_NEGOTIATED : AST_T38_REFUSED;
ast_indicate_data(chan, AST_CONTROL_T38_PARAMETERS, &t38_parameters, sizeof(t38_parameters));
ast_playtones_stop(chan);
break;
case AST_T38_NEGOTIATED:
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_debug(1, "Negotiated T.38 for send on %s\n", ast_channel_name(chan));
t38_parameters_ast_to_fax(&details->their_t38_parameters, parameters);
details->caps &= ~AST_FAX_TECH_AUDIO;
report_fax_status(chan, details, "T.38 Negotiated");
Multiple revisions 375993-375994 ........ r375993 | mmichelson | 2012-11-07 11:01:13 -0600 (Wed, 07 Nov 2012) | 30 lines Fix misuses of timeouts throughout the code. Prior to this change, a common method for determining if a timeout was reached was to call a function such as ast_waitfor_n() and inspect the out parameter that told how many milliseconds were left, then use that as the input to ast_waitfor_n() on the next go-around. The problem with this is that in some cases, submillisecond timeouts can occur, resulting in the out parameter not decreasing any. When this happens thousands of times, the result is that the timeout takes much longer than intended to be reached. As an example, I had a situation where a 3 second timeout took multiple days to finally end since most wakeups from ast_waitfor_n() were under a millisecond. This patch seeks to fix this pattern throughout the code. Now we log the time when an operation began and find the difference in wall clock time between now and when the event started. This means that sub-millisecond timeouts now cannot play havoc when trying to determine if something has timed out. Part of this fix also includes changing the function ast_waitfor() so that it is possible for it to return less than zero when a negative timeout is given to it. This makes it actually possible to detect errors in ast_waitfor() when there is no timeout. (closes issue ASTERISK-20414) reported by David M. Lee Review: https://reviewboard.asterisk.org/r/2135/ ........ r375994 | mmichelson | 2012-11-07 11:08:44 -0600 (Wed, 07 Nov 2012) | 3 lines Remove some debugging that accidentally made it in the last commit. ........ Merged revisions 375993-375994 from http://svn.asterisk.org/svn/asterisk/branches/1.8 ........ Merged revisions 375995 from http://svn.asterisk.org/svn/asterisk/branches/10 ........ Merged revisions 376014 from http://svn.asterisk.org/svn/asterisk/branches/11 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@376015 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-11-07 19:15:26 +00:00
break_loop = 1;
break;
default:
break;
}
}
ast_frfree(frame);
Multiple revisions 375993-375994 ........ r375993 | mmichelson | 2012-11-07 11:01:13 -0600 (Wed, 07 Nov 2012) | 30 lines Fix misuses of timeouts throughout the code. Prior to this change, a common method for determining if a timeout was reached was to call a function such as ast_waitfor_n() and inspect the out parameter that told how many milliseconds were left, then use that as the input to ast_waitfor_n() on the next go-around. The problem with this is that in some cases, submillisecond timeouts can occur, resulting in the out parameter not decreasing any. When this happens thousands of times, the result is that the timeout takes much longer than intended to be reached. As an example, I had a situation where a 3 second timeout took multiple days to finally end since most wakeups from ast_waitfor_n() were under a millisecond. This patch seeks to fix this pattern throughout the code. Now we log the time when an operation began and find the difference in wall clock time between now and when the event started. This means that sub-millisecond timeouts now cannot play havoc when trying to determine if something has timed out. Part of this fix also includes changing the function ast_waitfor() so that it is possible for it to return less than zero when a negative timeout is given to it. This makes it actually possible to detect errors in ast_waitfor() when there is no timeout. (closes issue ASTERISK-20414) reported by David M. Lee Review: https://reviewboard.asterisk.org/r/2135/ ........ r375994 | mmichelson | 2012-11-07 11:08:44 -0600 (Wed, 07 Nov 2012) | 3 lines Remove some debugging that accidentally made it in the last commit. ........ Merged revisions 375993-375994 from http://svn.asterisk.org/svn/asterisk/branches/1.8 ........ Merged revisions 375995 from http://svn.asterisk.org/svn/asterisk/branches/10 ........ Merged revisions 376014 from http://svn.asterisk.org/svn/asterisk/branches/11 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@376015 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-11-07 19:15:26 +00:00
if (break_loop) {
break;
}
}
ast_playtones_stop(chan);
if (ast_channel_get_t38_state(chan) == T38_STATE_NEGOTIATED) {
return 0;
}
/* T.38 negotiation did not happen, initiate a switch if requested */
if (details->option.request_t38 == AST_FAX_OPTFLAG_TRUE) {
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_debug(1, "Negotiating T.38 for send on %s\n", ast_channel_name(chan));
/* wait up to five seconds for negotiation to complete */
Multiple revisions 375993-375994 ........ r375993 | mmichelson | 2012-11-07 11:01:13 -0600 (Wed, 07 Nov 2012) | 30 lines Fix misuses of timeouts throughout the code. Prior to this change, a common method for determining if a timeout was reached was to call a function such as ast_waitfor_n() and inspect the out parameter that told how many milliseconds were left, then use that as the input to ast_waitfor_n() on the next go-around. The problem with this is that in some cases, submillisecond timeouts can occur, resulting in the out parameter not decreasing any. When this happens thousands of times, the result is that the timeout takes much longer than intended to be reached. As an example, I had a situation where a 3 second timeout took multiple days to finally end since most wakeups from ast_waitfor_n() were under a millisecond. This patch seeks to fix this pattern throughout the code. Now we log the time when an operation began and find the difference in wall clock time between now and when the event started. This means that sub-millisecond timeouts now cannot play havoc when trying to determine if something has timed out. Part of this fix also includes changing the function ast_waitfor() so that it is possible for it to return less than zero when a negative timeout is given to it. This makes it actually possible to detect errors in ast_waitfor() when there is no timeout. (closes issue ASTERISK-20414) reported by David M. Lee Review: https://reviewboard.asterisk.org/r/2135/ ........ r375994 | mmichelson | 2012-11-07 11:08:44 -0600 (Wed, 07 Nov 2012) | 3 lines Remove some debugging that accidentally made it in the last commit. ........ Merged revisions 375993-375994 from http://svn.asterisk.org/svn/asterisk/branches/1.8 ........ Merged revisions 375995 from http://svn.asterisk.org/svn/asterisk/branches/10 ........ Merged revisions 376014 from http://svn.asterisk.org/svn/asterisk/branches/11 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@376015 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-11-07 19:15:26 +00:00
timeout_ms = 5000;
/* set parameters based on the session's parameters */
t38_parameters_fax_to_ast(&t38_parameters, &details->our_t38_parameters);
t38_parameters.request_response = AST_T38_REQUEST_NEGOTIATE;
if ((ast_indicate_data(chan, AST_CONTROL_T38_PARAMETERS, &t38_parameters, sizeof(t38_parameters)) != 0)) {
return -1;
}
Multiple revisions 375993-375994 ........ r375993 | mmichelson | 2012-11-07 11:01:13 -0600 (Wed, 07 Nov 2012) | 30 lines Fix misuses of timeouts throughout the code. Prior to this change, a common method for determining if a timeout was reached was to call a function such as ast_waitfor_n() and inspect the out parameter that told how many milliseconds were left, then use that as the input to ast_waitfor_n() on the next go-around. The problem with this is that in some cases, submillisecond timeouts can occur, resulting in the out parameter not decreasing any. When this happens thousands of times, the result is that the timeout takes much longer than intended to be reached. As an example, I had a situation where a 3 second timeout took multiple days to finally end since most wakeups from ast_waitfor_n() were under a millisecond. This patch seeks to fix this pattern throughout the code. Now we log the time when an operation began and find the difference in wall clock time between now and when the event started. This means that sub-millisecond timeouts now cannot play havoc when trying to determine if something has timed out. Part of this fix also includes changing the function ast_waitfor() so that it is possible for it to return less than zero when a negative timeout is given to it. This makes it actually possible to detect errors in ast_waitfor() when there is no timeout. (closes issue ASTERISK-20414) reported by David M. Lee Review: https://reviewboard.asterisk.org/r/2135/ ........ r375994 | mmichelson | 2012-11-07 11:08:44 -0600 (Wed, 07 Nov 2012) | 3 lines Remove some debugging that accidentally made it in the last commit. ........ Merged revisions 375993-375994 from http://svn.asterisk.org/svn/asterisk/branches/1.8 ........ Merged revisions 375995 from http://svn.asterisk.org/svn/asterisk/branches/10 ........ Merged revisions 376014 from http://svn.asterisk.org/svn/asterisk/branches/11 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@376015 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-11-07 19:15:26 +00:00
start = ast_tvnow();
while ((ms = ast_remaining_ms(start, timeout_ms))) {
int break_loop = 0;
ms = ast_waitfor(chan, ms);
if (ms < 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_${fieldname}' and then writing setters/getters and converting the existing code to using them. When all fields are done, we can move ast_channel to a C file from channel.h and lop off the '__do_not_use_'. This patch sets up main/channel_interal_api.c to be the only file that actually accesses the ast_channel's fields directly. The intent would be for any API functions in channel.c to use the accessor functions. No more monkeying around with channel internals. We should use our own APIs. The interesting changes in this patch are the addition of channel_internal_api.c, the moving of the AST_DATA stuff from channel.c to channel_internal_api.c (note: the AST_DATA stuff will have to be reworked to use accessor functions when ast_channel is really opaque), and some re-working of the way channel iterators/callbacks are handled so as to avoid creating fake ast_channels on the stack to pass in matching data by directly accessing fields (since "name" is a stringfield and the fake channel doesn't init the stringfields, you can't use the ast_channel_name_set() function). I went with ast_channel_name(chan) for a getter, and ast_channel_name_set(chan, name) for a setter. The majority of the grunt-work for this change was done by writing a semantic patch using Coccinelle ( http://coccinelle.lip6.fr/ ). Review: https://reviewboard.asterisk.org/r/1655/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@350223 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-01-09 22:15:50 +00:00
ast_log(LOG_WARNING, "error on '%s' while waiting for T.38 negotiation.\n", ast_channel_name(chan));
return -1;
}
if (ms == 0) { /* all done, nothing happened */
Replace direct access to channel name with accessor functions There are many benefits to making the ast_channel an opaque handle, from increasing maintainability to presenting ways to kill masquerades. This patch kicks things off by taking things a field at a time, renaming the field to '__do_not_use_${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' timed-out during the T.38 negotiation.\n", ast_channel_name(chan));
details->caps &= ~AST_FAX_TECH_T38;
break;
}
if (!(frame = ast_read(chan))) {
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, "error on '%s' while waiting for T.38 negotiation.\n", ast_channel_name(chan));
return -1;
}
if ((frame->frametype == AST_FRAME_CONTROL) &&
(frame->subclass.integer == AST_CONTROL_T38_PARAMETERS) &&
(frame->datalen == sizeof(t38_parameters))) {
struct ast_control_t38_parameters *parameters = frame->data.ptr;
switch (parameters->request_response) {
case AST_T38_REQUEST_NEGOTIATE:
t38_parameters_fax_to_ast(&t38_parameters, &details->our_t38_parameters);
t38_parameters.request_response = AST_T38_NEGOTIATED;
ast_indicate_data(chan, AST_CONTROL_T38_PARAMETERS, &t38_parameters, sizeof(t38_parameters));
break;
case AST_T38_NEGOTIATED:
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_debug(1, "Negotiated T.38 for receive on %s\n", ast_channel_name(chan));
t38_parameters_ast_to_fax(&details->their_t38_parameters, parameters);
details->caps &= ~AST_FAX_TECH_AUDIO;
report_fax_status(chan, details, "T.38 Negotiated");
Multiple revisions 375993-375994 ........ r375993 | mmichelson | 2012-11-07 11:01:13 -0600 (Wed, 07 Nov 2012) | 30 lines Fix misuses of timeouts throughout the code. Prior to this change, a common method for determining if a timeout was reached was to call a function such as ast_waitfor_n() and inspect the out parameter that told how many milliseconds were left, then use that as the input to ast_waitfor_n() on the next go-around. The problem with this is that in some cases, submillisecond timeouts can occur, resulting in the out parameter not decreasing any. When this happens thousands of times, the result is that the timeout takes much longer than intended to be reached. As an example, I had a situation where a 3 second timeout took multiple days to finally end since most wakeups from ast_waitfor_n() were under a millisecond. This patch seeks to fix this pattern throughout the code. Now we log the time when an operation began and find the difference in wall clock time between now and when the event started. This means that sub-millisecond timeouts now cannot play havoc when trying to determine if something has timed out. Part of this fix also includes changing the function ast_waitfor() so that it is possible for it to return less than zero when a negative timeout is given to it. This makes it actually possible to detect errors in ast_waitfor() when there is no timeout. (closes issue ASTERISK-20414) reported by David M. Lee Review: https://reviewboard.asterisk.org/r/2135/ ........ r375994 | mmichelson | 2012-11-07 11:08:44 -0600 (Wed, 07 Nov 2012) | 3 lines Remove some debugging that accidentally made it in the last commit. ........ Merged revisions 375993-375994 from http://svn.asterisk.org/svn/asterisk/branches/1.8 ........ Merged revisions 375995 from http://svn.asterisk.org/svn/asterisk/branches/10 ........ Merged revisions 376014 from http://svn.asterisk.org/svn/asterisk/branches/11 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@376015 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-11-07 19:15:26 +00:00
break_loop = 1;
break;
case AST_T38_REFUSED:
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' refused to negotiate T.38\n", ast_channel_name(chan));
details->caps &= ~AST_FAX_TECH_T38;
Multiple revisions 375993-375994 ........ r375993 | mmichelson | 2012-11-07 11:01:13 -0600 (Wed, 07 Nov 2012) | 30 lines Fix misuses of timeouts throughout the code. Prior to this change, a common method for determining if a timeout was reached was to call a function such as ast_waitfor_n() and inspect the out parameter that told how many milliseconds were left, then use that as the input to ast_waitfor_n() on the next go-around. The problem with this is that in some cases, submillisecond timeouts can occur, resulting in the out parameter not decreasing any. When this happens thousands of times, the result is that the timeout takes much longer than intended to be reached. As an example, I had a situation where a 3 second timeout took multiple days to finally end since most wakeups from ast_waitfor_n() were under a millisecond. This patch seeks to fix this pattern throughout the code. Now we log the time when an operation began and find the difference in wall clock time between now and when the event started. This means that sub-millisecond timeouts now cannot play havoc when trying to determine if something has timed out. Part of this fix also includes changing the function ast_waitfor() so that it is possible for it to return less than zero when a negative timeout is given to it. This makes it actually possible to detect errors in ast_waitfor() when there is no timeout. (closes issue ASTERISK-20414) reported by David M. Lee Review: https://reviewboard.asterisk.org/r/2135/ ........ r375994 | mmichelson | 2012-11-07 11:08:44 -0600 (Wed, 07 Nov 2012) | 3 lines Remove some debugging that accidentally made it in the last commit. ........ Merged revisions 375993-375994 from http://svn.asterisk.org/svn/asterisk/branches/1.8 ........ Merged revisions 375995 from http://svn.asterisk.org/svn/asterisk/branches/10 ........ Merged revisions 376014 from http://svn.asterisk.org/svn/asterisk/branches/11 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@376015 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-11-07 19:15:26 +00:00
break_loop = 1;
break;
default:
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_ERROR, "channel '%s' failed to negotiate T.38\n", ast_channel_name(chan));
details->caps &= ~AST_FAX_TECH_T38;
Multiple revisions 375993-375994 ........ r375993 | mmichelson | 2012-11-07 11:01:13 -0600 (Wed, 07 Nov 2012) | 30 lines Fix misuses of timeouts throughout the code. Prior to this change, a common method for determining if a timeout was reached was to call a function such as ast_waitfor_n() and inspect the out parameter that told how many milliseconds were left, then use that as the input to ast_waitfor_n() on the next go-around. The problem with this is that in some cases, submillisecond timeouts can occur, resulting in the out parameter not decreasing any. When this happens thousands of times, the result is that the timeout takes much longer than intended to be reached. As an example, I had a situation where a 3 second timeout took multiple days to finally end since most wakeups from ast_waitfor_n() were under a millisecond. This patch seeks to fix this pattern throughout the code. Now we log the time when an operation began and find the difference in wall clock time between now and when the event started. This means that sub-millisecond timeouts now cannot play havoc when trying to determine if something has timed out. Part of this fix also includes changing the function ast_waitfor() so that it is possible for it to return less than zero when a negative timeout is given to it. This makes it actually possible to detect errors in ast_waitfor() when there is no timeout. (closes issue ASTERISK-20414) reported by David M. Lee Review: https://reviewboard.asterisk.org/r/2135/ ........ r375994 | mmichelson | 2012-11-07 11:08:44 -0600 (Wed, 07 Nov 2012) | 3 lines Remove some debugging that accidentally made it in the last commit. ........ Merged revisions 375993-375994 from http://svn.asterisk.org/svn/asterisk/branches/1.8 ........ Merged revisions 375995 from http://svn.asterisk.org/svn/asterisk/branches/10 ........ Merged revisions 376014 from http://svn.asterisk.org/svn/asterisk/branches/11 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@376015 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-11-07 19:15:26 +00:00
break_loop = 1;
break;
}
}
ast_frfree(frame);
Multiple revisions 375993-375994 ........ r375993 | mmichelson | 2012-11-07 11:01:13 -0600 (Wed, 07 Nov 2012) | 30 lines Fix misuses of timeouts throughout the code. Prior to this change, a common method for determining if a timeout was reached was to call a function such as ast_waitfor_n() and inspect the out parameter that told how many milliseconds were left, then use that as the input to ast_waitfor_n() on the next go-around. The problem with this is that in some cases, submillisecond timeouts can occur, resulting in the out parameter not decreasing any. When this happens thousands of times, the result is that the timeout takes much longer than intended to be reached. As an example, I had a situation where a 3 second timeout took multiple days to finally end since most wakeups from ast_waitfor_n() were under a millisecond. This patch seeks to fix this pattern throughout the code. Now we log the time when an operation began and find the difference in wall clock time between now and when the event started. This means that sub-millisecond timeouts now cannot play havoc when trying to determine if something has timed out. Part of this fix also includes changing the function ast_waitfor() so that it is possible for it to return less than zero when a negative timeout is given to it. This makes it actually possible to detect errors in ast_waitfor() when there is no timeout. (closes issue ASTERISK-20414) reported by David M. Lee Review: https://reviewboard.asterisk.org/r/2135/ ........ r375994 | mmichelson | 2012-11-07 11:08:44 -0600 (Wed, 07 Nov 2012) | 3 lines Remove some debugging that accidentally made it in the last commit. ........ Merged revisions 375993-375994 from http://svn.asterisk.org/svn/asterisk/branches/1.8 ........ Merged revisions 375995 from http://svn.asterisk.org/svn/asterisk/branches/10 ........ Merged revisions 376014 from http://svn.asterisk.org/svn/asterisk/branches/11 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@376015 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-11-07 19:15:26 +00:00
if (break_loop) {
break;
}
}
/* if T.38 was negotiated, we are done initializing */
if (ast_channel_get_t38_state(chan) == T38_STATE_NEGOTIATED) {
return 0;
}
/* send one more CNG tone to get audio going again for some
* carriers if we are going to fall back to audio mode */
if (details->option.allow_audio == AST_FAX_OPTFLAG_TRUE) {
if (ast_playtones_start(chan, 1024, "!1100/500,!0/3000", 1)) {
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_ERROR, "error generating second CNG tone on %s\n", ast_channel_name(chan));
return -1;
}
Multiple revisions 375993-375994 ........ r375993 | mmichelson | 2012-11-07 11:01:13 -0600 (Wed, 07 Nov 2012) | 30 lines Fix misuses of timeouts throughout the code. Prior to this change, a common method for determining if a timeout was reached was to call a function such as ast_waitfor_n() and inspect the out parameter that told how many milliseconds were left, then use that as the input to ast_waitfor_n() on the next go-around. The problem with this is that in some cases, submillisecond timeouts can occur, resulting in the out parameter not decreasing any. When this happens thousands of times, the result is that the timeout takes much longer than intended to be reached. As an example, I had a situation where a 3 second timeout took multiple days to finally end since most wakeups from ast_waitfor_n() were under a millisecond. This patch seeks to fix this pattern throughout the code. Now we log the time when an operation began and find the difference in wall clock time between now and when the event started. This means that sub-millisecond timeouts now cannot play havoc when trying to determine if something has timed out. Part of this fix also includes changing the function ast_waitfor() so that it is possible for it to return less than zero when a negative timeout is given to it. This makes it actually possible to detect errors in ast_waitfor() when there is no timeout. (closes issue ASTERISK-20414) reported by David M. Lee Review: https://reviewboard.asterisk.org/r/2135/ ........ r375994 | mmichelson | 2012-11-07 11:08:44 -0600 (Wed, 07 Nov 2012) | 3 lines Remove some debugging that accidentally made it in the last commit. ........ Merged revisions 375993-375994 from http://svn.asterisk.org/svn/asterisk/branches/1.8 ........ Merged revisions 375995 from http://svn.asterisk.org/svn/asterisk/branches/10 ........ Merged revisions 376014 from http://svn.asterisk.org/svn/asterisk/branches/11 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@376015 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-11-07 19:15:26 +00:00
timeout_ms = 3500;
start = ast_tvnow();
while ((ms = ast_remaining_ms(start, timeout_ms))) {
int break_loop = 0;
ms = ast_waitfor(chan, ms);
if (ms < 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_ERROR, "error while generating second CNG tone on %s\n", ast_channel_name(chan));
ast_playtones_stop(chan);
return -1;
}
if (ms == 0) { /* all done, nothing happened */
break;
}
if (!(frame = ast_read(chan))) {
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_ERROR, "error reading frame while generating second CNG tone on %s\n", ast_channel_name(chan));
ast_playtones_stop(chan);
return -1;
}
if ((frame->frametype == AST_FRAME_CONTROL) &&
(frame->subclass.integer == AST_CONTROL_T38_PARAMETERS) &&
(frame->datalen == sizeof(t38_parameters))) {
struct ast_control_t38_parameters *parameters = frame->data.ptr;
switch (parameters->request_response) {
case AST_T38_REQUEST_NEGOTIATE:
/* the other end has requested a switch to T.38, so reply that we are willing, if we can
* do T.38 as well
*/
t38_parameters_fax_to_ast(&t38_parameters, &details->our_t38_parameters);
t38_parameters.request_response = (details->caps & AST_FAX_TECH_T38) ? AST_T38_NEGOTIATED : AST_T38_REFUSED;
ast_indicate_data(chan, AST_CONTROL_T38_PARAMETERS, &t38_parameters, sizeof(t38_parameters));
ast_playtones_stop(chan);
break;
case AST_T38_NEGOTIATED:
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_debug(1, "Negotiated T.38 for send on %s\n", ast_channel_name(chan));
t38_parameters_ast_to_fax(&details->their_t38_parameters, parameters);
details->caps &= ~AST_FAX_TECH_AUDIO;
report_fax_status(chan, details, "T.38 Negotiated");
Multiple revisions 375993-375994 ........ r375993 | mmichelson | 2012-11-07 11:01:13 -0600 (Wed, 07 Nov 2012) | 30 lines Fix misuses of timeouts throughout the code. Prior to this change, a common method for determining if a timeout was reached was to call a function such as ast_waitfor_n() and inspect the out parameter that told how many milliseconds were left, then use that as the input to ast_waitfor_n() on the next go-around. The problem with this is that in some cases, submillisecond timeouts can occur, resulting in the out parameter not decreasing any. When this happens thousands of times, the result is that the timeout takes much longer than intended to be reached. As an example, I had a situation where a 3 second timeout took multiple days to finally end since most wakeups from ast_waitfor_n() were under a millisecond. This patch seeks to fix this pattern throughout the code. Now we log the time when an operation began and find the difference in wall clock time between now and when the event started. This means that sub-millisecond timeouts now cannot play havoc when trying to determine if something has timed out. Part of this fix also includes changing the function ast_waitfor() so that it is possible for it to return less than zero when a negative timeout is given to it. This makes it actually possible to detect errors in ast_waitfor() when there is no timeout. (closes issue ASTERISK-20414) reported by David M. Lee Review: https://reviewboard.asterisk.org/r/2135/ ........ r375994 | mmichelson | 2012-11-07 11:08:44 -0600 (Wed, 07 Nov 2012) | 3 lines Remove some debugging that accidentally made it in the last commit. ........ Merged revisions 375993-375994 from http://svn.asterisk.org/svn/asterisk/branches/1.8 ........ Merged revisions 375995 from http://svn.asterisk.org/svn/asterisk/branches/10 ........ Merged revisions 376014 from http://svn.asterisk.org/svn/asterisk/branches/11 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@376015 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-11-07 19:15:26 +00:00
break_loop = 1;
break;
default:
break;
}
}
ast_frfree(frame);
Multiple revisions 375993-375994 ........ r375993 | mmichelson | 2012-11-07 11:01:13 -0600 (Wed, 07 Nov 2012) | 30 lines Fix misuses of timeouts throughout the code. Prior to this change, a common method for determining if a timeout was reached was to call a function such as ast_waitfor_n() and inspect the out parameter that told how many milliseconds were left, then use that as the input to ast_waitfor_n() on the next go-around. The problem with this is that in some cases, submillisecond timeouts can occur, resulting in the out parameter not decreasing any. When this happens thousands of times, the result is that the timeout takes much longer than intended to be reached. As an example, I had a situation where a 3 second timeout took multiple days to finally end since most wakeups from ast_waitfor_n() were under a millisecond. This patch seeks to fix this pattern throughout the code. Now we log the time when an operation began and find the difference in wall clock time between now and when the event started. This means that sub-millisecond timeouts now cannot play havoc when trying to determine if something has timed out. Part of this fix also includes changing the function ast_waitfor() so that it is possible for it to return less than zero when a negative timeout is given to it. This makes it actually possible to detect errors in ast_waitfor() when there is no timeout. (closes issue ASTERISK-20414) reported by David M. Lee Review: https://reviewboard.asterisk.org/r/2135/ ........ r375994 | mmichelson | 2012-11-07 11:08:44 -0600 (Wed, 07 Nov 2012) | 3 lines Remove some debugging that accidentally made it in the last commit. ........ Merged revisions 375993-375994 from http://svn.asterisk.org/svn/asterisk/branches/1.8 ........ Merged revisions 375995 from http://svn.asterisk.org/svn/asterisk/branches/10 ........ Merged revisions 376014 from http://svn.asterisk.org/svn/asterisk/branches/11 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@376015 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-11-07 19:15:26 +00:00
if (break_loop) {
break;
}
}
ast_playtones_stop(chan);
/* if T.38 was negotiated, we are done initializing */
if (ast_channel_get_t38_state(chan) == T38_STATE_NEGOTIATED) {
return 0;
}
}
}
/* if we made it here, then T.38 failed, check the 'f' flag */
if (details->option.allow_audio == AST_FAX_OPTFLAG_FALSE) {
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, "Audio FAX not allowed on channel '%s' and T.38 negotiation failed; aborting.\n", ast_channel_name(chan));
return -1;
}
/* ok, audio fallback is allowed */
details->caps |= AST_FAX_TECH_AUDIO;
return 0;
}
/*!
* \brief Report on the status of a completed fax send attempt
* \note This will lock the \ref ast_channel
*/
static int report_send_fax_status(struct ast_channel *chan, struct ast_fax_session_details *details)
{
RAII_VAR(struct ast_json *, json_obj, NULL, ast_json_unref);
RAII_VAR(struct stasis_message *, message, NULL, ao2_cleanup);
struct ast_json *json_filenames;
json_filenames = generate_filenames_json(details);
if (!json_filenames) {
return -1;
}
{
const char *remote_station_id;
const char *local_station_id;
const char *fax_pages;
const char *fax_resolution;
const char *fax_bitrate;
SCOPED_CHANNELLOCK(lock, chan);
remote_station_id = AST_JSON_UTF8_VALIDATE(pbx_builtin_getvar_helper(chan, "REMOTESTATIONID"));
if (!ast_strlen_zero(remote_station_id)) {
remote_station_id = ast_strdupa(remote_station_id);
}
local_station_id = AST_JSON_UTF8_VALIDATE(pbx_builtin_getvar_helper(chan, "LOCALSTATIONID"));
if (!ast_strlen_zero(local_station_id)) {
local_station_id = ast_strdupa(local_station_id);
}
fax_pages = S_OR(pbx_builtin_getvar_helper(chan, "FAXPAGES"), "");
if (!ast_strlen_zero(fax_pages)) {
fax_pages = ast_strdupa(fax_pages);
}
fax_resolution = S_OR(pbx_builtin_getvar_helper(chan, "FAXRESOLUTION"), "");
if (!ast_strlen_zero(fax_resolution)) {
fax_resolution = ast_strdupa(fax_resolution);
}
fax_bitrate = S_OR(pbx_builtin_getvar_helper(chan, "FAXBITRATE"), "");
if (!ast_strlen_zero(fax_bitrate)) {
fax_bitrate = ast_strdupa(fax_bitrate);
}
json_obj = ast_json_pack("{s: s, s: s, s: s, s: s, s: s, s: s, s: o}",
"type", "send",
"remote_station_id", S_OR(remote_station_id, ""),
"local_station_id", S_OR(local_station_id, ""),
"fax_pages", S_OR(fax_pages, ""),
"fax_resolution", S_OR(fax_resolution, ""),
"fax_bitrate", S_OR(fax_bitrate, ""),
"filenames", json_filenames);
if (!json_obj) {
return -1;
}
message = ast_channel_blob_create_from_cache(ast_channel_uniqueid(chan), ast_channel_fax_type(), json_obj);
if (!message) {
return -1;
}
stasis_publish(ast_channel_topic(chan), message);
}
return 0;
}
/*! \brief initiate a send FAX session */
static int sendfax_exec(struct ast_channel *chan, const char *data)
{
char *parse, *filenames, *c, modems[128] = "";
int channel_alive, file_count;
RAII_VAR(struct ast_fax_session_details *, details, NULL, ao2_cleanup);
RAII_VAR(struct ast_fax_session *, s, NULL, ao2_cleanup);
struct ast_fax_tech_token *token = NULL;
struct ast_fax_document *doc;
AST_DECLARE_APP_ARGS(args,
AST_APP_ARG(filenames);
AST_APP_ARG(options);
);
struct ast_flags opts = { 0, };
enum ast_t38_state t38state;
/* initialize output channel variables */
pbx_builtin_setvar_helper(chan, "FAXSTATUS", "FAILED");
pbx_builtin_setvar_helper(chan, "REMOTESTATIONID", NULL);
pbx_builtin_setvar_helper(chan, "FAXPAGES", "0");
pbx_builtin_setvar_helper(chan, "FAXBITRATE", NULL);
pbx_builtin_setvar_helper(chan, "FAXRESOLUTION", NULL);
pbx_builtin_setvar_helper(chan, "FAXMODE", NULL);
/* Get a requirement structure and set it. This structure is used
* to tell the FAX technology module about the higher level FAX session */
if (!(details = find_or_create_details(chan))) {
pbx_builtin_setvar_helper(chan, "FAXERROR", "MEMORY_ERROR");
pbx_builtin_setvar_helper(chan, "FAXSTATUSSTRING", "error allocating memory");
ast_log(LOG_ERROR, "System cannot provide memory for session requirements.\n");
return -1;
}
ast_string_field_set(details, result, "FAILED");
ast_string_field_set(details, resultstr, "error starting fax session");
ast_string_field_set(details, error, "INIT_ERROR");
set_channel_variables(chan, details);
if (details->gateway_id > 0) {
ast_string_field_set(details, resultstr, "can't send a fax on a channel with a T.38 gateway");
set_channel_variables(chan, details);
ast_log(LOG_ERROR, "executing SendFAX on a channel with a T.38 Gateway is not supported\n");
return -1;
}
if (details->maxrate < details->minrate) {
ast_string_field_set(details, error, "INVALID_ARGUMENTS");
ast_string_field_set(details, resultstr, "maxrate is less than minrate");
set_channel_variables(chan, details);
ast_log(LOG_ERROR, "maxrate %u is less than minrate %u\n", details->maxrate, details->minrate);
return -1;
}
if (check_modem_rate(details->modems, details->minrate)) {
ast_fax_modem_to_str(details->modems, modems, sizeof(modems));
ast_log(LOG_ERROR, "'modems' setting '%s' is incompatible with 'minrate' setting %u\n", modems, details->minrate);
ast_string_field_set(details, error, "INVALID_ARGUMENTS");
ast_string_field_set(details, resultstr, "incompatible 'modems' and 'minrate' settings");
set_channel_variables(chan, details);
return -1;
}
if (check_modem_rate(details->modems, details->maxrate)) {
ast_fax_modem_to_str(details->modems, modems, sizeof(modems));
ast_log(LOG_ERROR, "'modems' setting '%s' is incompatible with 'maxrate' setting %u\n", modems, details->maxrate);
ast_string_field_set(details, error, "INVALID_ARGUMENTS");
ast_string_field_set(details, resultstr, "incompatible 'modems' and 'maxrate' settings");
set_channel_variables(chan, details);
return -1;
}
if (ast_strlen_zero(data)) {
ast_string_field_set(details, error, "INVALID_ARGUMENTS");
ast_string_field_set(details, resultstr, "invalid arguments");
set_channel_variables(chan, details);
ast_log(LOG_WARNING, "%s requires an argument (filename[&filename[&filename]][,options])\n", app_sendfax);
return -1;
}
parse = ast_strdupa(data);
AST_STANDARD_APP_ARGS(args, parse);
if (!ast_strlen_zero(args.options) &&
ast_app_parse_options(fax_exec_options, &opts, NULL, args.options)) {
ast_string_field_set(details, error, "INVALID_ARGUMENTS");
ast_string_field_set(details, resultstr, "invalid arguments");
set_channel_variables(chan, details);
return -1;
}
if (ast_strlen_zero(args.filenames)) {
ast_string_field_set(details, error, "INVALID_ARGUMENTS");
ast_string_field_set(details, resultstr, "invalid arguments");
set_channel_variables(chan, details);
ast_log(LOG_WARNING, "%s requires an argument (filename[&filename[&filename]],options])\n", app_sendfax);
return -1;
}
/* check for unsupported FAX application options */
if (ast_test_flag(&opts, OPT_CALLERMODE) || ast_test_flag(&opts, OPT_CALLEDMODE)) {
ast_string_field_set(details, error, "INVALID_ARGUMENTS");
ast_string_field_set(details, resultstr, "invalid arguments");
set_channel_variables(chan, details);
ast_log(LOG_WARNING, "%s does not support polling\n", app_sendfax);
return -1;
}
ast_atomic_fetchadd_int(&faxregistry.fax_tx_attempts, 1);
file_count = 0;
filenames = args.filenames;
while ((c = strsep(&filenames, "&"))) {
if (access(c, (F_OK | R_OK)) < 0) {
ast_string_field_set(details, error, "FILE_ERROR");
ast_string_field_set(details, resultstr, "error reading file");
set_channel_variables(chan, details);
ast_log(LOG_ERROR, "access failure. Verify '%s' exists and check permissions.\n", args.filenames);
return -1;
}
if (!(doc = ast_calloc(1, sizeof(*doc) + strlen(c) + 1))) {
ast_string_field_set(details, error, "MEMORY_ERROR");
ast_string_field_set(details, resultstr, "error allocating memory");
set_channel_variables(chan, details);
ast_log(LOG_ERROR, "System cannot provide memory for session requirements.\n");
return -1;
}
strcpy(doc->filename, c);
AST_LIST_INSERT_TAIL(&details->documents, doc, next);
file_count++;
}
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_verb(3, "Channel '%s' sending FAX:\n", ast_channel_name(chan));
AST_LIST_TRAVERSE(&details->documents, doc, next) {
ast_verb(3, " %s\n", doc->filename);
}
details->caps = AST_FAX_TECH_SEND;
if (file_count > 1) {
details->caps |= AST_FAX_TECH_MULTI_DOC;
}
/* check for debug */
if (ast_test_flag(&opts, OPT_DEBUG) || global_fax_debug) {
details->option.debug = AST_FAX_OPTFLAG_TRUE;
}
/* check for request for status events */
if (ast_test_flag(&opts, OPT_STATUS)) {
details->option.statusevents = AST_FAX_OPTFLAG_TRUE;
}
t38state = ast_channel_get_t38_state(chan);
if ((t38state == T38_STATE_UNAVAILABLE) || (t38state == T38_STATE_REJECTED) ||
ast_test_flag(&opts, OPT_ALLOWAUDIO) ||
ast_test_flag(&opts, OPT_FORCE_AUDIO)) {
details->option.allow_audio = AST_FAX_OPTFLAG_TRUE;
}
if (ast_test_flag(&opts, OPT_REQUEST_T38)) {
details->option.request_t38 = AST_FAX_OPTFLAG_TRUE;
}
if (!(s = fax_session_reserve(details, &token))) {
ast_string_field_set(details, resultstr, "error reserving fax session");
set_channel_variables(chan, details);
ast_log(LOG_ERROR, "Unable to reserve FAX session.\n");
return -1;
}
/* make sure the channel is up */
if (ast_channel_state(chan) != AST_STATE_UP) {
if (ast_answer(chan)) {
ast_string_field_set(details, resultstr, "error answering channel");
set_channel_variables(chan, details);
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 answer attempt.\n", ast_channel_name(chan));
fax_session_release(s, token);
return -1;
}
}
if (!ast_test_flag(&opts, OPT_FORCE_AUDIO)) {
if (set_fax_t38_caps(chan, details)) {
ast_string_field_set(details, error, "T38_NEG_ERROR");
ast_string_field_set(details, resultstr, "error negotiating T.38");
set_channel_variables(chan, details);
fax_session_release(s, token);
return -1;
}
} else {
details->caps |= AST_FAX_TECH_AUDIO;
}
if (!ast_test_flag(&opts, OPT_FORCE_AUDIO) && (details->caps & AST_FAX_TECH_T38)) {
if (sendfax_t38_init(chan, details)) {
ast_string_field_set(details, error, "T38_NEG_ERROR");
ast_string_field_set(details, resultstr, "error negotiating T.38");
set_channel_variables(chan, details);
fax_session_release(s, token);
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_ERROR, "error initializing channel '%s' in T.38 mode\n", ast_channel_name(chan));
return -1;
}
} else {
details->option.send_cng = 1;
}
if ((channel_alive = generic_fax_exec(chan, details, s, token)) < 0) {
ast_atomic_fetchadd_int(&faxregistry.fax_failures, 1);
}
if (ast_channel_get_t38_state(chan) == T38_STATE_NEGOTIATED) {
if (disable_t38(chan)) {
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_debug(1, "error disabling T.38 mode on %s\n", ast_channel_name(chan));
}
}
if (!(filenames = generate_filenames_string(details, "FileName: ", "\r\n"))) {
ast_log(LOG_ERROR, "Error generating SendFAX manager event\n");
return (!channel_alive) ? -1 : 0;
}
/* send out the AMI completion event */
if (report_send_fax_status(chan, details)) {
ast_log(AST_LOG_ERROR, "Error publishing SendFAX status message\n");
}
/* If the channel hungup return -1; otherwise, return 0 to continue in the dialplan */
return (!channel_alive) ? -1 : 0;
}
/*! \brief destroy the v21 detection parts of a fax gateway session */
static void destroy_v21_sessions(struct fax_gateway *gateway)
{
if (gateway->chan_v21_session) {
ao2_unlink(faxregistry.container, gateway->chan_v21_session);
ao2_ref(gateway->chan_v21_session, -1);
gateway->chan_v21_session = NULL;
}
if (gateway->peer_v21_session) {
ao2_unlink(faxregistry.container, gateway->peer_v21_session);
ao2_ref(gateway->peer_v21_session, -1);
gateway->peer_v21_session = NULL;
}
}
/*! \brief destroy a FAX gateway session structure */
static void destroy_gateway(void *data)
{
struct fax_gateway *gateway = data;
destroy_v21_sessions(gateway);
if (gateway->s) {
fax_session_release(gateway->s, gateway->token);
gateway->token = NULL;
ao2_unlink(faxregistry.container, gateway->s);
ao2_ref(gateway->s, -1);
gateway->s = 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
ao2_cleanup(gateway->chan_read_format);
ao2_cleanup(gateway->chan_write_format);
ao2_cleanup(gateway->peer_read_format);
ao2_cleanup(gateway->peer_write_format);
}
static struct ast_fax_session *fax_v21_session_new (struct ast_channel *chan) {
struct ast_fax_session_details *v21_details;
struct ast_fax_session *v21_session;
if (!chan || !(v21_details = session_details_new())) {
return NULL;
}
v21_details->caps = AST_FAX_TECH_V21_DETECT;
v21_session = fax_session_new(v21_details, chan, NULL, NULL);
ao2_ref(v21_details, -1);
return v21_session;
}
/*! \brief Create a new fax gateway object.
* \param chan the channel the gateway object will be attached to
* \param details the fax session details
* \return NULL or a fax gateway object
*/
static struct fax_gateway *fax_gateway_new(struct ast_channel *chan, struct ast_fax_session_details *details)
{
struct fax_gateway *gateway = ao2_alloc(sizeof(*gateway), destroy_gateway);
if (!gateway) {
return NULL;
}
if (!(gateway->chan_v21_session = fax_v21_session_new(chan))) {
ast_log(LOG_ERROR, "Can't create V21 session on chan %s for T.38 gateway session\n", ast_channel_name(chan));
ao2_ref(gateway, -1);
return NULL;
}
gateway->framehook = -1;
details->caps = AST_FAX_TECH_GATEWAY;
if (details->gateway_timeout && !(gateway->s = fax_session_reserve(details, &gateway->token))) {
details->caps &= ~AST_FAX_TECH_GATEWAY;
ast_log(LOG_ERROR, "Can't reserve a FAX session, gateway attempt failed.\n");
ao2_ref(gateway, -1);
return NULL;
}
return gateway;
}
/*!
* \brief Create a fax session and start T.30<->T.38 gateway mode
*
* \param gateway a fax gateway object
* \param details fax session details
* \param chan active channel
*
* \pre chan is locked on entry
*
* \return 0 on error 1 on success
*/
static int fax_gateway_start(struct fax_gateway *gateway, struct ast_fax_session_details *details, struct ast_channel *chan)
{
struct ast_fax_session *s;
int start_res;
/* if the fax gateway is already started then do nothing */
if (gateway->s &&
gateway->s->state != AST_FAX_STATE_RESERVED && gateway->s->state != AST_FAX_STATE_INACTIVE) {
return 0;
}
/* if we start gateway we don't need v21 detection sessions any more */
destroy_v21_sessions(gateway);
/* create the FAX session */
if (!(s = fax_session_new(details, chan, gateway->s, gateway->token))) {
gateway->token = NULL;
ast_string_field_set(details, result, "FAILED");
ast_string_field_set(details, resultstr, "error starting gateway session");
ast_string_field_set(details, error, "INIT_ERROR");
details->is_t38_negotiated = 0;
set_channel_variables(chan, details);
report_fax_status(chan, details, "No Available Resource");
ast_log(LOG_ERROR, "Can't create a FAX session, gateway attempt failed.\n");
return -1;
}
/* release the reference for the reserved session and replace it with
* the real session */
if (gateway->s) {
ao2_ref(gateway->s, -1);
}
gateway->s = s;
gateway->token = NULL;
ast_channel_unlock(chan);
start_res = gateway->s->tech->start_session(gateway->s);
ast_channel_lock(chan);
if (start_res < 0) {
ast_string_field_set(details, result, "FAILED");
ast_string_field_set(details, resultstr, "error starting gateway session");
ast_string_field_set(details, error, "INIT_ERROR");
details->is_t38_negotiated = 0;
set_channel_variables(chan, details);
return -1;
}
gateway->timeout_start.tv_sec = 0;
gateway->timeout_start.tv_usec = 0;
report_fax_status(chan, details, "FAX Transmission In Progress");
return 0;
}
/*! \pre chan is locked on entry */
static struct ast_frame *fax_gateway_request_t38(struct fax_gateway *gateway, struct ast_channel *chan)
{
struct ast_frame *fp;
struct ast_control_t38_parameters t38_parameters = {
.request_response = AST_T38_REQUEST_NEGOTIATE,
};
struct ast_frame control_frame = {
.src = "res_fax",
.frametype = AST_FRAME_CONTROL,
.datalen = sizeof(t38_parameters),
.subclass.integer = AST_CONTROL_T38_PARAMETERS,
.data.ptr = &t38_parameters,
};
struct ast_fax_session_details *details = find_details(chan);
if (!details) {
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_ERROR, "no FAX session details found on chan %s for T.38 gateway session, odd\n", ast_channel_name(chan));
ast_framehook_detach(chan, gateway->framehook);
return NULL;
}
t38_parameters_fax_to_ast(&t38_parameters, &details->our_t38_parameters);
ao2_ref(details, -1);
if (!(fp = ast_frisolate(&control_frame))) {
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_ERROR, "error generating T.38 request control frame on chan %s for T.38 gateway session\n", ast_channel_name(chan));
return NULL;
}
gateway->t38_state = T38_STATE_NEGOTIATING;
gateway->timeout_start = ast_tvnow();
details->is_t38_negotiated = 0;
details->gateway_timeout = FAX_GATEWAY_TIMEOUT;
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_debug(1, "requesting T.38 for gateway session for %s\n", ast_channel_name(chan));
return fp;
}
/*! \pre chan is locked on entry */
static struct ast_frame *fax_gateway_detect_v21(struct fax_gateway *gateway, struct ast_channel *chan, struct ast_channel *peer, struct ast_channel *active, struct ast_frame *f)
{
struct ast_channel *other = (active == chan) ? peer : chan;
struct ast_fax_session *active_v21_session = (active == chan) ? gateway->chan_v21_session : gateway->peer_v21_session;
if (!active_v21_session || gateway->detected_v21) {
return f;
}
if (active_v21_session->tech->write(active_v21_session, f) == 0 &&
active_v21_session->details->option.v21_detected) {
gateway->detected_v21 = 1;
}
if (gateway->detected_v21) {
enum ast_t38_state state_other;
enum ast_t38_state state_active;
struct ast_frame *fp;
struct ast_fax_session_details *details;
int negotiate_both = 0;
/*
* The default behavior is to wait for the active endpoint to initiate negotiation.
* Find out if this has been overridden. If so, instead of waiting have Asterisk
* initiate the negotiation requests out to both endpoints.
*/
details = find_or_create_details(active);
if (details) {
negotiate_both = details->negotiate_both;
ao2_ref(details, -1);
} else {
ast_log(LOG_WARNING, "Detect v21 - no session details for channel '%s'\n",
ast_channel_name(chan));
}
destroy_v21_sessions(gateway);
ast_channel_unlock(chan);
state_active = ast_channel_get_t38_state(active);
state_other = ast_channel_get_t38_state(other);
ast_channel_lock(chan);
ast_debug(1, "detected v21 preamble from %s\n", ast_channel_name(active));
if (state_active == T38_STATE_UNKNOWN || state_other == T38_STATE_UNKNOWN) {
if (!(fp = fax_gateway_request_t38(gateway, chan))) {
return f;
}
/* May be called endpoint is improperly configured to rely on the calling endpoint
* to initiate T.38 re-INVITEs, send T.38 negotiation request to called endpoint */
if (negotiate_both && state_active == T38_STATE_UNKNOWN) {
ast_debug(1, "sending T.38 negotiation request to %s\n", ast_channel_name(active));
if (active == chan) {
ast_channel_unlock(chan);
}
ast_write(active, fp);
if (active == chan) {
ast_channel_lock(chan);
}
}
if (state_other == T38_STATE_UNKNOWN) {
ast_debug(1, "sending T.38 negotiation request to %s\n", ast_channel_name(other));
return fp;
}
} else {
ast_debug(1, "neither %s nor %s support T.38 for T.38 gateway session\n", ast_channel_name(active), ast_channel_name(other));
}
}
return f;
}
/*! \pre chan is locked on entry */
static void fax_gateway_indicate_t38(struct ast_channel *chan, struct ast_channel *active, struct ast_control_t38_parameters *control_params)
{
if (active == chan) {
ast_channel_unlock(chan);
ast_indicate_data(chan, AST_CONTROL_T38_PARAMETERS, control_params, sizeof(*control_params));
ast_channel_lock(chan);
} else {
ast_queue_control_data(chan, AST_CONTROL_T38_PARAMETERS, control_params, sizeof(*control_params));
}
}
/*!
* \brief T38 Gateway Negotiate t38 parameters
*
* \param gateway gateway object
* \param chan channel running the gateway
* \param peer channel im bridged too
* \param active channel the frame originated on
* \param f the control frame to process
*
* \pre chan is locked on entry
*
* \return processed control frame or null frame
*/
static struct ast_frame *fax_gateway_detect_t38(struct fax_gateway *gateway, struct ast_channel *chan, struct ast_channel *peer, struct ast_channel *active, struct ast_frame *f)
{
struct ast_control_t38_parameters *control_params = f->data.ptr;
struct ast_channel *other = (active == chan) ? peer : chan;
struct ast_fax_session_details *details;
enum ast_t38_state state_other;
if (f->datalen != sizeof(struct ast_control_t38_parameters)) {
/* invalaid AST_CONTROL_T38_PARAMETERS frame, we can't
* do anything with it, pass it on */
return f;
}
/* ignore frames from ourselves */
if ((gateway->t38_state == T38_STATE_NEGOTIATED && control_params->request_response == AST_T38_NEGOTIATED)
|| (gateway->t38_state == T38_STATE_REJECTED && control_params->request_response == AST_T38_REFUSED)
|| (gateway->t38_state == T38_STATE_NEGOTIATING && control_params->request_response == AST_T38_REQUEST_TERMINATE)) {
return f;
}
if (!(details = find_details(chan))) {
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_ERROR, "no FAX session details found on chan %s for T.38 gateway session, odd\n", ast_channel_name(chan));
ast_framehook_detach(chan, gateway->framehook);
return f;
}
if (control_params->request_response == AST_T38_REQUEST_NEGOTIATE) {
ast_channel_unlock(chan);
state_other = ast_channel_get_t38_state(other);
ast_channel_lock(chan);
if (state_other == T38_STATE_UNKNOWN) {
/* we detected a request to negotiate T.38 and the
* other channel appears to support T.38, we'll pass
* the request through and only step in if the other
* channel rejects the request */
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_debug(1, "%s is attempting to negotiate T.38 with %s, we'll see what happens\n", ast_channel_name(active), ast_channel_name(other));
t38_parameters_ast_to_fax(&details->their_t38_parameters, control_params);
gateway->t38_state = T38_STATE_UNKNOWN;
gateway->timeout_start = ast_tvnow();
details->is_t38_negotiated = 0;
details->gateway_timeout = FAX_GATEWAY_TIMEOUT;
ao2_ref(details, -1);
return f;
} else if (state_other == T38_STATE_UNAVAILABLE || state_other == T38_STATE_REJECTED) {
/* the other channel does not support T.38, we need to
* step in here */
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_debug(1, "%s is attempting to negotiate T.38 but %s does not support it\n", ast_channel_name(active), ast_channel_name(other));
ast_debug(1, "starting T.38 gateway for T.38 channel %s and G.711 channel %s\n", ast_channel_name(active), ast_channel_name(other));
t38_parameters_ast_to_fax(&details->their_t38_parameters, control_params);
t38_parameters_fax_to_ast(control_params, &details->our_t38_parameters);
if (fax_gateway_start(gateway, details, chan)) {
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_ERROR, "error starting T.38 gateway for T.38 channel %s and G.711 channel %s\n", ast_channel_name(active), ast_channel_name(other));
gateway->t38_state = T38_STATE_REJECTED;
details->is_t38_negotiated = 0;
control_params->request_response = AST_T38_REFUSED;
ast_framehook_detach(chan, details->gateway_id);
details->gateway_id = -1;
} else {
gateway->t38_state = T38_STATE_NEGOTIATED;
details->is_t38_negotiated = chan == active;
control_params->request_response = AST_T38_NEGOTIATED;
report_fax_status(chan, details, "T.38 Negotiated");
}
fax_gateway_indicate_t38(chan, active, control_params);
ao2_ref(details, -1);
return &ast_null_frame;
} else if (gateway->t38_state == T38_STATE_NEGOTIATING) {
/* we got a request to negotiate T.38 after we already
* sent one to the other party based on v21 preamble
* detection. We'll just pretend we passed this request
* through in the first place. */
t38_parameters_ast_to_fax(&details->their_t38_parameters, control_params);
gateway->t38_state = T38_STATE_UNKNOWN;
gateway->timeout_start = ast_tvnow();
details->is_t38_negotiated = 0;
details->gateway_timeout = FAX_GATEWAY_TIMEOUT;
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_debug(1, "%s is attempting to negotiate T.38 after we already sent a negotiation request based on v21 preamble detection\n", ast_channel_name(active));
ao2_ref(details, -1);
return &ast_null_frame;
} else if (gateway->t38_state == T38_STATE_NEGOTIATED) {
/* we got a request to negotiate T.38 after we already
* sent one to the other party based on v21 preamble
* detection and received a response. We need to
* respond to this and shut down the gateway. */
t38_parameters_fax_to_ast(control_params, &details->their_t38_parameters);
ast_framehook_detach(chan, details->gateway_id);
details->gateway_id = -1;
control_params->request_response = AST_T38_NEGOTIATED;
fax_gateway_indicate_t38(chan, active, control_params);
ast_string_field_set(details, result, "SUCCESS");
ast_string_field_set(details, resultstr, "no gateway necessary");
ast_string_field_set(details, error, "NATIVE_T38");
details->is_t38_negotiated = 1;
set_channel_variables(chan, details);
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_debug(1, "%s is attempting to negotiate T.38 after we already negotiated T.38 with %s, disabling the gateway\n", ast_channel_name(active), ast_channel_name(other));
ao2_ref(details, -1);
return &ast_null_frame;
} else {
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, "%s is attempting to negotiate T.38 while %s is in an unsupported state\n", ast_channel_name(active), ast_channel_name(other));
ao2_ref(details, -1);
return f;
}
} else if (gateway->t38_state == T38_STATE_NEGOTIATING
&& control_params->request_response == AST_T38_REFUSED) {
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_debug(1, "unable to negotiate T.38 on %s for fax gateway\n", ast_channel_name(active));
details->is_t38_negotiated = 0;
/* our request to negotiate T.38 was refused, if the other
* channel supports T.38, they might still reinvite and save
* the day. Otherwise disable the gateway. */
ast_channel_unlock(chan);
state_other = ast_channel_get_t38_state(other);
ast_channel_lock(chan);
if (state_other == T38_STATE_UNKNOWN) {
gateway->t38_state = T38_STATE_UNAVAILABLE;
} else if (state_other != T38_STATE_NEGOTIATING) {
ast_framehook_detach(chan, details->gateway_id);
details->gateway_id = -1;
ast_string_field_set(details, result, "FAILED");
ast_string_field_set(details, resultstr, "unable to negotiate T.38");
ast_string_field_set(details, error, "T38_NEG_ERROR");
set_channel_variables(chan, details);
}
ao2_ref(details, -1);
return &ast_null_frame;
} else if (gateway->t38_state == T38_STATE_NEGOTIATING
&& control_params->request_response == AST_T38_NEGOTIATED) {
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_debug(1, "starting T.38 gateway for T.38 channel %s and G.711 channel %s\n", ast_channel_name(active), ast_channel_name(other));
t38_parameters_ast_to_fax(&details->their_t38_parameters, control_params);
if (fax_gateway_start(gateway, details, chan)) {
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_ERROR, "error starting T.38 gateway for T.38 channel %s and G.711 channel %s\n", ast_channel_name(active), ast_channel_name(other));
gateway->t38_state = T38_STATE_NEGOTIATING;
details->is_t38_negotiated = 0;
control_params->request_response = AST_T38_REQUEST_TERMINATE;
fax_gateway_indicate_t38(chan, active, control_params);
} else {
gateway->t38_state = T38_STATE_NEGOTIATED;
details->is_t38_negotiated = chan == active;
report_fax_status(chan, details, "T.38 Negotiated");
}
ao2_ref(details, -1);
return &ast_null_frame;
} else if (control_params->request_response == AST_T38_REFUSED) {
/* the other channel refused the request to negotiate T.38,
* we'll step in here and pretend the request was accepted */
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_debug(1, "%s attempted to negotiate T.38 but %s refused the request\n", ast_channel_name(other), ast_channel_name(active));
ast_debug(1, "starting T.38 gateway for T.38 channel %s and G.711 channel %s\n", ast_channel_name(other), ast_channel_name(active));
t38_parameters_fax_to_ast(control_params, &details->our_t38_parameters);
if (fax_gateway_start(gateway, details, chan)) {
ast_log(LOG_ERROR, "error starting T.38 gateway for T.38 channel %s and G.711 channel %s\n", ast_channel_name(other), ast_channel_name(active));
gateway->t38_state = T38_STATE_REJECTED;
details->is_t38_negotiated = 0;
control_params->request_response = AST_T38_REFUSED;
ast_framehook_detach(chan, details->gateway_id);
details->gateway_id = -1;
} else {
gateway->t38_state = T38_STATE_NEGOTIATED;
details->is_t38_negotiated = chan == other;
control_params->request_response = AST_T38_NEGOTIATED;
}
ao2_ref(details, -1);
return f;
} else if (control_params->request_response == AST_T38_REQUEST_TERMINATE) {
/* the channel wishes to end our short relationship, we shall
* oblige */
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_debug(1, "T.38 channel %s is requesting a shutdown of T.38, disabling the gateway\n", ast_channel_name(active));
ast_framehook_detach(chan, details->gateway_id);
details->gateway_id = -1;
gateway->t38_state = T38_STATE_REJECTED;
details->is_t38_negotiated = 0;
control_params->request_response = AST_T38_TERMINATED;
fax_gateway_indicate_t38(chan, active, control_params);
ao2_ref(details, -1);
return &ast_null_frame;
} else if (control_params->request_response == AST_T38_NEGOTIATED) {
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_debug(1, "T.38 successfully negotiated between %s and %s, no gateway necessary\n", ast_channel_name(active), ast_channel_name(other));
ast_framehook_detach(chan, details->gateway_id);
details->gateway_id = -1;
ast_string_field_set(details, result, "SUCCESS");
ast_string_field_set(details, resultstr, "no gateway necessary");
ast_string_field_set(details, error, "NATIVE_T38");
details->is_t38_negotiated = 1;
set_channel_variables(chan, details);
ao2_ref(details, -1);
return f;
} else if (control_params->request_response == AST_T38_TERMINATED) {
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_debug(1, "T.38 disabled on channel %s\n", ast_channel_name(active));
ast_framehook_detach(chan, details->gateway_id);
details->gateway_id = -1;
ao2_ref(details, -1);
return &ast_null_frame;
}
ao2_ref(details, -1);
return f;
}
/*! \brief Destroy the gateway data structure when the framehook is detached
* \param data framehook data (gateway data)*/
static void fax_gateway_framehook_destroy(void *data)
{
struct fax_gateway *gateway = data;
if (gateway->s) {
switch (gateway->s->state) {
case AST_FAX_STATE_INITIALIZED:
case AST_FAX_STATE_OPEN:
case AST_FAX_STATE_ACTIVE:
case AST_FAX_STATE_COMPLETE:
if (gateway->s->tech->cancel_session) {
gateway->s->tech->cancel_session(gateway->s);
}
/* fall through */
default:
break;
}
}
ao2_ref(gateway, -1);
}
/*!
* \brief T.30<->T.38 gateway framehook.
*
* Intercept packets on bridged channels and determine if a T.38 gateway is
* required. If a gateway is required, start a gateway and handle T.38
* negotiation if necessary.
*
* \param chan channel running the gateway
* \param f frame to handle may be NULL
* \param event framehook event
* \param data framehook data (struct fax_gateway *)
*
* \pre chan is locked on entry
*
* \return processed frame or NULL when f is NULL or a null frame
*/
static struct ast_frame *fax_gateway_framehook(struct ast_channel *chan, struct ast_frame *f, enum ast_framehook_event event, void *data)
{
struct fax_gateway *gateway = data;
struct ast_channel *active;
RAII_VAR(struct ast_fax_session_details *, details, NULL, ao2_cleanup);
RAII_VAR(struct ast_channel *, peer, NULL, ao2_cleanup);
RAII_VAR(struct ast_channel *, chan_ref, chan, ao2_cleanup);
/* Ref bump channel for when we have to unlock it */
ao2_ref(chan_ref, 1);
if (gateway->s) {
details = gateway->s->details;
ao2_ref(details, 1);
} else {
if (!(details = find_details(chan))) {
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_ERROR, "no FAX session details found on chan %s for T.38 gateway session, odd\n", ast_channel_name(chan));
ast_framehook_detach(chan, gateway->framehook);
return f;
}
}
/* restore audio formats when we are detached */
if (event == AST_FRAMEHOOK_EVENT_DETACHED) {
set_channel_variables(chan, details);
if (gateway->bridged) {
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_set_read_format(chan, gateway->chan_read_format);
ast_set_write_format(chan, gateway->chan_write_format);
ast_channel_unlock(chan);
peer = ast_channel_bridge_peer(chan);
if (peer) {
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_set_read_format(peer, gateway->peer_read_format);
ast_set_write_format(peer, gateway->peer_write_format);
ast_channel_make_compatible(chan, peer);
}
ast_channel_lock(chan);
}
return NULL;
}
if (!f || (event == AST_FRAMEHOOK_EVENT_ATTACHED)) {
return NULL;
};
/* this frame was generated by the fax gateway, pass it on */
if (ast_test_flag(f, AST_FAX_FRFLAG_GATEWAY)) {
return f;
}
/* If we aren't bridged or we don't have a peer, don't do anything */
ast_channel_unlock(chan);
peer = ast_channel_bridge_peer(chan);
ast_channel_lock(chan);
if (!peer) {
return f;
}
if (!gateway->bridged) {
enum ast_t38_state state_chan;
enum ast_t38_state state_peer;
int chan_is_hungup;
int peer_is_hungup;
chan_is_hungup = ast_check_hangup(chan);
peer_is_hungup = ast_check_hangup(peer);
/* Don't start a gateway if either channel is hung up */
if (chan_is_hungup || peer_is_hungup) {
return f;
}
ast_channel_unlock(chan);
state_chan = ast_channel_get_t38_state(chan);
state_peer = ast_channel_get_t38_state(peer);
ast_channel_lock(chan);
/* don't start a gateway if neither channel can handle T.38 */
if (state_chan == T38_STATE_UNAVAILABLE && state_peer == T38_STATE_UNAVAILABLE) {
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_debug(1, "not starting gateway for %s and %s; neither channel supports T.38\n", ast_channel_name(chan), ast_channel_name(peer));
ast_framehook_detach(chan, gateway->framehook);
details->gateway_id = -1;
ast_string_field_set(details, result, "FAILED");
ast_string_field_set(details, resultstr, "neither channel supports T.38");
ast_string_field_set(details, error, "T38_NEG_ERROR");
details->is_t38_negotiated = 0;
set_channel_variables(chan, details);
return f;
}
if (details->gateway_timeout) {
gateway->timeout_start = ast_tvnow();
}
ast_channel_unlock(chan);
ast_channel_lock_both(chan, peer);
/* we are bridged, change r/w formats to SLIN for v21 preamble
* detection and T.30 */
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_replace(gateway->chan_read_format, ast_channel_readformat(chan));
ao2_replace(gateway->chan_write_format, ast_channel_writeformat(chan));
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_replace(gateway->peer_read_format, ast_channel_readformat(peer));
ao2_replace(gateway->peer_write_format, ast_channel_writeformat(peer));
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_set_read_format(chan, ast_format_slin);
ast_set_write_format(chan, ast_format_slin);
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_set_read_format(peer, ast_format_slin);
ast_set_write_format(peer, ast_format_slin);
ast_channel_unlock(peer);
gateway->bridged = 1;
if (!(gateway->peer_v21_session = fax_v21_session_new(peer))) {
ast_log(LOG_ERROR, "Can't create V21 session on chan %s for T.38 gateway session\n", ast_channel_name(peer));
ast_framehook_detach(chan, gateway->framehook);
return f;
}
}
if (gateway->bridged && !ast_tvzero(gateway->timeout_start)) {
if (ast_tvdiff_ms(ast_tvnow(), gateway->timeout_start) > details->gateway_timeout) {
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_debug(1, "no fax activity between %s and %s after %d ms, disabling gateway\n", ast_channel_name(chan), ast_channel_name(peer), details->gateway_timeout);
ast_framehook_detach(chan, gateway->framehook);
details->gateway_id = -1;
ast_string_field_set(details, result, "FAILED");
ast_string_field_build(details, resultstr, "no fax activity after %d ms", details->gateway_timeout);
ast_string_field_set(details, error, "TIMEOUT");
details->is_t38_negotiated = 0;
set_channel_variables(chan, details);
return f;
}
}
/* only handle VOICE, MODEM, and CONTROL frames*/
switch (f->frametype) {
case AST_FRAME_VOICE:
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(f->subclass.format, ast_format_slin) != AST_FORMAT_CMP_EQUAL) &&
(ast_format_cmp(f->subclass.format, ast_format_alaw) != AST_FORMAT_CMP_EQUAL) &&
(ast_format_cmp(f->subclass.format, ast_format_ulaw) != AST_FORMAT_CMP_EQUAL)) {
return f;
}
break;
case AST_FRAME_MODEM:
if (f->subclass.integer == AST_MODEM_T38) {
break;
}
return f;
case AST_FRAME_CONTROL:
if (f->subclass.integer == AST_CONTROL_T38_PARAMETERS) {
break;
}
return f;
default:
return f;
}
/* detect the active channel */
switch (event) {
case AST_FRAMEHOOK_EVENT_WRITE:
active = peer;
break;
case AST_FRAMEHOOK_EVENT_READ:
active = chan;
break;
default:
ast_log(LOG_WARNING, "unhandled framehook event %u\n", event);
return f;
}
/* handle control frames */
if (f->frametype == AST_FRAME_CONTROL && f->subclass.integer == AST_CONTROL_T38_PARAMETERS) {
return fax_gateway_detect_t38(gateway, chan, peer, active, f);
}
if (!gateway->detected_v21 && gateway->t38_state == T38_STATE_UNAVAILABLE && f->frametype == AST_FRAME_VOICE) {
/* not in gateway mode and have not detected v21 yet, listen
* for v21 */
return fax_gateway_detect_v21(gateway, chan, peer, active, f);
}
/* in gateway mode, gateway some packets */
if (gateway->t38_state == T38_STATE_NEGOTIATED) {
struct ast_trans_pvt *readtrans;
if (!gateway->s || !gateway->s->tech_pvt) {
ast_log(LOG_ERROR, "no FAX session on chan %s for T.38 gateway session, odd", ast_channel_name(chan));
return f;
}
/* framehooks are called in __ast_read() before frame format
* translation is done, so we need to translate here */
if ((f->frametype == AST_FRAME_VOICE) && (ast_format_cmp(f->subclass.format, ast_format_slin) != AST_FORMAT_CMP_EQUAL)
&& (readtrans = ast_channel_readtrans(active))) {
if ((f = ast_translate(readtrans, f, event == AST_FRAMEHOOK_EVENT_WRITE ? 0 : 1)) == NULL) {
f = &ast_null_frame;
return f;
}
/* XXX we ignore the return value here, perhaps we should
* disable the gateway if a write fails. I am not sure how a
* write would fail, or even if a failure would be fatal so for
* now we'll just ignore the return value. */
gateway->s->tech->write(gateway->s, f);
ast_frfree(f);
} else {
gateway->s->tech->write(gateway->s, f);
}
f = &ast_null_frame;
return f;
}
/* force silence on the line if T.38 negotiation might be taking place */
if (gateway->t38_state != T38_STATE_UNAVAILABLE && gateway->t38_state != T38_STATE_REJECTED) {
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 (f->frametype == AST_FRAME_VOICE &&
(ast_format_cmp(f->subclass.format, ast_format_slin) == AST_FORMAT_CMP_EQUAL)) {
short silence_buf[f->samples];
struct ast_frame silence_frame = {
.frametype = AST_FRAME_VOICE,
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
.subclass.format = ast_format_slin,
.data.ptr = silence_buf,
.samples = f->samples,
.datalen = sizeof(silence_buf),
};
memset(silence_buf, 0, sizeof(silence_buf));
return ast_frisolate(&silence_frame);
} else {
return &ast_null_frame;
}
}
return f;
}
/*! \brief Attach a gateway framehook object to a channel.
* \param chan the channel to attach to
* \param details fax session details
* \return the framehook id of the attached framehook or -1 on error
* \retval -1 error
*/
static int fax_gateway_attach(struct ast_channel *chan, struct ast_fax_session_details *details)
{
struct fax_gateway *gateway;
struct ast_framehook_interface fr_hook = {
.version = AST_FRAMEHOOK_INTERFACE_VERSION,
.event_cb = fax_gateway_framehook,
.destroy_cb = fax_gateway_framehook_destroy,
.disable_inheritance = 1, /* Masquerade inheritance is handled through the datastore fixup */
};
if (global_fax_debug) {
details->option.debug = AST_FAX_OPTFLAG_TRUE;
}
ast_string_field_set(details, result, "SUCCESS");
ast_string_field_set(details, resultstr, "gateway operation started successfully");
ast_string_field_set(details, error, "NO_ERROR");
set_channel_variables(chan, details);
/* set up the frame hook*/
gateway = fax_gateway_new(chan, details);
if (!gateway) {
ast_string_field_set(details, result, "FAILED");
ast_string_field_set(details, resultstr, "error initializing gateway session");
ast_string_field_set(details, error, "INIT_ERROR");
details->is_t38_negotiated = 0;
set_channel_variables(chan, details);
report_fax_status(chan, details, "No Available Resource");
return -1;
}
fr_hook.data = gateway;
ast_channel_lock(chan);
gateway->framehook = ast_framehook_attach(chan, &fr_hook);
ast_channel_unlock(chan);
if (gateway->framehook < 0) {
ao2_ref(gateway, -1);
ast_string_field_set(details, result, "FAILED");
ast_string_field_set(details, resultstr, "error attaching gateway to channel");
ast_string_field_set(details, error, "INIT_ERROR");
details->is_t38_negotiated = 0;
set_channel_variables(chan, details);
return -1;
}
return gateway->framehook;
}
/*! \brief destroy a FAX detect structure */
static void destroy_faxdetect(void *data)
{
struct fax_detect *faxdetect = data;
if (faxdetect->dsp) {
ast_dsp_free(faxdetect->dsp);
faxdetect->dsp = NULL;
}
ao2_cleanup(faxdetect->details);
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(faxdetect->orig_format);
}
/*! \brief Create a new fax detect object.
* \param chan the channel attaching to
* \param timeout in ms to remove framehook in this time if not zero
* \param flags required options
* \return NULL or a fax gateway object
*/
static struct fax_detect *fax_detect_new(struct ast_channel *chan, int timeout, int flags)
{
struct fax_detect *faxdetect = ao2_alloc(sizeof(*faxdetect), destroy_faxdetect);
if (!faxdetect) {
return NULL;
}
faxdetect->flags = flags;
if (timeout) {
faxdetect->timeout_start = ast_tvnow();
} else {
faxdetect->timeout_start.tv_sec = 0;
faxdetect->timeout_start.tv_usec = 0;
}
if (faxdetect->flags & FAX_DETECT_MODE_CNG) {
faxdetect->dsp = ast_dsp_new();
if (!faxdetect->dsp) {
ao2_ref(faxdetect, -1);
return NULL;
}
ast_dsp_set_features(faxdetect->dsp, DSP_FEATURE_FAX_DETECT);
ast_dsp_set_faxmode(faxdetect->dsp, DSP_FAXMODE_DETECT_CNG | DSP_FAXMODE_DETECT_SQUELCH);
} else {
faxdetect->dsp = NULL;
}
return faxdetect;
}
/*! \brief Deref the faxdetect data structure when the faxdetect framehook is detached
* \param data framehook data (faxdetect data)*/
static void fax_detect_framehook_destroy(void *data)
{
struct fax_detect *faxdetect = data;
ao2_ref(faxdetect, -1);
}
/*! \brief Fax Detect Framehook
*
* Listen for fax tones in audio path and enable jumping to a extension when detected.
*
* \param chan channel
* \param f frame to handle may be NULL
* \param event framehook event
* \param data framehook data (struct fax_detect *)
*
* \return processed frame or NULL when f is NULL or a null frame
*/
static struct ast_frame *fax_detect_framehook(struct ast_channel *chan, struct ast_frame *f, enum ast_framehook_event event, void *data)
{
struct fax_detect *faxdetect = data;
struct ast_fax_session_details *details;
struct ast_control_t38_parameters *control_params;
RAII_VAR(struct ast_channel *, peer, NULL, ao2_cleanup);
RAII_VAR(struct ast_channel *, chan_ref, chan, ao2_cleanup);
int result = 0;
/* Ref bump the channel for when we have to unlock it */
ao2_ref(chan, 1);
details = faxdetect->details;
switch (event) {
case AST_FRAMEHOOK_EVENT_ATTACHED:
/* Setup format for DSP on ATTACH*/
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_replace(faxdetect->orig_format, ast_channel_readformat(chan));
if ((ast_format_cmp(ast_channel_readformat(chan), ast_format_slin) != AST_FORMAT_CMP_EQUAL) &&
(ast_format_cmp(ast_channel_readformat(chan), ast_format_alaw) != AST_FORMAT_CMP_EQUAL) &&
(ast_format_cmp(ast_channel_readformat(chan), ast_format_ulaw) != AST_FORMAT_CMP_EQUAL)) {
if (ast_set_read_format(chan, ast_format_slin)) {
ast_framehook_detach(chan, details->faxdetect_id);
details->faxdetect_id = -1;
return f;
}
}
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;
case AST_FRAMEHOOK_EVENT_DETACHED:
/* restore audio formats when we are detached */
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_set_read_format(chan, faxdetect->orig_format);
ast_channel_unlock(chan);
peer = ast_channel_bridge_peer(chan);
if (peer) {
ast_channel_make_compatible(chan, peer);
}
ast_channel_lock(chan);
return NULL;
case AST_FRAMEHOOK_EVENT_READ:
if (f) {
break;
}
default:
return f;
};
if (details->faxdetect_id < 0) {
return f;
}
if (!ast_tvzero(faxdetect->timeout_start)
&& ast_tvdiff_ms(ast_tvnow(), faxdetect->timeout_start) > details->faxdetect_timeout) {
ast_debug(1, "FAXOPT(faxdetect) timeout on %s\n", ast_channel_name(chan));
ast_framehook_detach(chan, details->faxdetect_id);
details->faxdetect_id = -1;
return f;
}
/* only handle VOICE and CONTROL frames*/
switch (f->frametype) {
case AST_FRAME_VOICE:
/* we have no DSP this means we not detecting CNG */
if (!faxdetect->dsp) {
return f;
}
/* We can only process some formats*/
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(f->subclass.format, ast_format_slin) != AST_FORMAT_CMP_EQUAL) &&
(ast_format_cmp(f->subclass.format, ast_format_alaw) != AST_FORMAT_CMP_EQUAL) &&
(ast_format_cmp(f->subclass.format, ast_format_ulaw) != AST_FORMAT_CMP_EQUAL)) {
return f;
}
break;
case AST_FRAME_CONTROL:
if ((f->subclass.integer == AST_CONTROL_T38_PARAMETERS) &&
(faxdetect->flags & FAX_DETECT_MODE_T38)) {
break;
}
return f;
default:
return f;
}
if (f->frametype == AST_FRAME_VOICE) {
f = ast_dsp_process(chan, faxdetect->dsp, f);
if (f->frametype == AST_FRAME_DTMF) {
result = f->subclass.integer;
}
} else if ((f->frametype == AST_FRAME_CONTROL) && (f->datalen == sizeof(struct ast_control_t38_parameters))) {
control_params = f->data.ptr;
switch (control_params->request_response) {
case AST_T38_NEGOTIATED:
case AST_T38_REQUEST_NEGOTIATE:
result = 't';
break;
default:
break;
}
}
if (result) {
const char *target_context;
switch (result) {
case 'f':
case 't':
target_context = S_OR(ast_channel_macrocontext(chan), ast_channel_context(chan));
ast_channel_unlock(chan);
ast_frfree(f);
f = &ast_null_frame;
if (ast_exists_extension(chan, target_context, "fax", 1,
S_COR(ast_channel_caller(chan)->id.number.valid, ast_channel_caller(chan)->id.number.str, NULL))) {
ast_verb(2, "Redirecting '%s' to fax extension due to %s detection\n",
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_channel_name(chan), (result == 'f') ? "CNG" : "T38");
pbx_builtin_setvar_helper(chan, "FAXEXTEN", ast_channel_exten(chan));
if (ast_async_goto(chan, target_context, "fax", 1)) {
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_NOTICE, "Failed to async goto '%s' into fax of '%s'\n", ast_channel_name(chan), target_context);
}
} else {
ast_log(LOG_NOTICE, "FAX %s detected but no fax extension in context (%s)\n",
(result == 'f') ? "CNG" : "T38", target_context);
}
ast_channel_lock(chan);
ast_framehook_detach(chan, details->faxdetect_id);
details->faxdetect_id = -1;
break;
default:
break;
}
}
return f;
}
/*! \brief Attach a faxdetect framehook object to a channel.
* \param chan the channel to attach to
* \param timeout in ms to remove framehook in this time if not zero
* \return the faxdetect structure or NULL on error
* \param flags required options
* \retval -1 error
*/
static int fax_detect_attach(struct ast_channel *chan, int timeout, int flags)
{
struct fax_detect *faxdetect;
struct ast_fax_session_details *details;
struct ast_framehook_interface fr_hook = {
.version = AST_FRAMEHOOK_INTERFACE_VERSION,
.event_cb = fax_detect_framehook,
.destroy_cb = fax_detect_framehook_destroy,
};
if (!(details = find_or_create_details(chan))) {
ast_log(LOG_ERROR, "System cannot provide memory for session requirements.\n");
return -1;
}
/* set up the frame hook*/
faxdetect = fax_detect_new(chan, timeout, flags);
if (!faxdetect) {
ao2_ref(details, -1);
return -1;
}
fr_hook.data = faxdetect;
faxdetect->details = details;
ast_channel_lock(chan);
details->faxdetect_id = ast_framehook_attach(chan, &fr_hook);
details->faxdetect_timeout = timeout;
details->faxdetect_flags = flags;
ast_channel_unlock(chan);
if (details->faxdetect_id < 0) {
ao2_ref(faxdetect, -1);
}
return details->faxdetect_id;
}
/*! \brief hash callback for ao2 */
static int session_hash_cb(const void *obj, const int flags)
{
const struct ast_fax_session *s = obj;
return s->id;
}
/*! \brief compare callback for ao2 */
static int session_cmp_cb(void *obj, void *arg, int flags)
{
struct ast_fax_session *lhs = obj, *rhs = arg;
return (lhs->id == rhs->id) ? CMP_MATCH | CMP_STOP : 0;
}
/*! \brief fax session tab completion */
static char *fax_session_tab_complete(struct ast_cli_args *a)
{
int tklen;
int wordnum = 0;
char *name = NULL;
struct ao2_iterator i;
struct ast_fax_session *s;
char tbuf[5];
if (a->pos != 3) {
return NULL;
}
tklen = strlen(a->word);
i = ao2_iterator_init(faxregistry.container, 0);
while ((s = ao2_iterator_next(&i))) {
snprintf(tbuf, sizeof(tbuf), "%u", s->id);
if (!strncasecmp(a->word, tbuf, tklen) && ++wordnum > a->n) {
name = ast_strdup(tbuf);
ao2_ref(s, -1);
break;
}
ao2_ref(s, -1);
}
ao2_iterator_destroy(&i);
return name;
}
static char *cli_fax_show_version(struct ast_cli_entry *e, int cmd, struct ast_cli_args *a)
{
struct fax_module *fax;
switch(cmd) {
case CLI_INIT:
e->command = "fax show version";
e->usage =
"Usage: fax show version\n"
" Show versions of FAX For Asterisk components.\n";
return NULL;
case CLI_GENERATE:
return NULL;
}
if (a->argc != 3) {
return CLI_SHOWUSAGE;
}
ast_cli(a->fd, "FAX For Asterisk Components:\n");
ast_cli(a->fd, "\tApplications: %s\n", ast_get_version());
AST_RWLIST_RDLOCK(&faxmodules);
AST_RWLIST_TRAVERSE(&faxmodules, fax, list) {
ast_cli(a->fd, "\t%s: %s\n", fax->tech->description, fax->tech->version);
}
AST_RWLIST_UNLOCK(&faxmodules);
ast_cli(a->fd, "\n");
return CLI_SUCCESS;
}
/*! \brief enable FAX debugging */
static char *cli_fax_set_debug(struct ast_cli_entry *e, int cmd, struct ast_cli_args *a)
{
int flag;
const char *what;
switch (cmd) {
case CLI_INIT:
e->command = "fax set debug {on|off}";
e->usage =
"Usage: fax set debug { on | off }\n"
" Enable/Disable FAX debugging on new FAX sessions. The basic FAX debugging will result in\n"
" additional events sent to manager sessions with 'call' class permissions. When\n"
" verbosity is greater than '5' events will be displayed to the console and audio versus\n"
" energy analysis will be performed and displayed to the console.\n";
return NULL;
case CLI_GENERATE:
return NULL;
}
what = a->argv[e->args-1]; /* guaranteed to exist */
if (!strcasecmp(what, "on")) {
flag = 1;
} else if (!strcasecmp(what, "off")) {
flag = 0;
} else {
return CLI_SHOWUSAGE;
}
global_fax_debug = flag;
ast_cli(a->fd, "\n\nFAX Debug %s\n\n", (flag) ? "Enabled" : "Disabled");
return CLI_SUCCESS;
}
/*! \brief display registered FAX capabilities */
static char *cli_fax_show_capabilities(struct ast_cli_entry *e, int cmd, struct ast_cli_args *a)
{
struct fax_module *fax;
unsigned int num_modules = 0;
switch (cmd) {
case CLI_INIT:
e->command = "fax show capabilities";
e->usage =
"Usage: fax show capabilities\n"
" Shows the capabilities of the registered FAX technology modules\n";
return NULL;
case CLI_GENERATE:
return NULL;
}
ast_cli(a->fd, "\n\nRegistered FAX Technology Modules:\n\n");
AST_RWLIST_RDLOCK(&faxmodules);
AST_RWLIST_TRAVERSE(&faxmodules, fax, list) {
ast_cli(a->fd, "%-15s : %s\n%-15s : %s\n%-15s : ", "Type", fax->tech->type, "Description", fax->tech->description, "Capabilities");
fax->tech->cli_show_capabilities(a->fd);
num_modules++;
}
AST_RWLIST_UNLOCK(&faxmodules);
ast_cli(a->fd, "%u registered modules\n\n", num_modules);
return CLI_SUCCESS;
}
/*! \brief display global defaults and settings */
static char *cli_fax_show_settings(struct ast_cli_entry *e, int cmd, struct ast_cli_args *a)
{
struct fax_module *fax;
char modems[128] = "";
struct fax_options options;
switch (cmd) {
case CLI_INIT:
e->command = "fax show settings";
e->usage =
"Usage: fax show settings\n"
" Show the global settings and defaults of both the FAX core and technology modules\n";
return NULL;
case CLI_GENERATE:
return NULL;
}
get_general_options(&options);
ast_cli(a->fd, "FAX For Asterisk Settings:\n");
ast_cli(a->fd, "\tECM: %s\n", options.ecm ? "Enabled" : "Disabled");
ast_cli(a->fd, "\tStatus Events: %s\n", options.statusevents ? "On" : "Off");
ast_cli(a->fd, "\tMinimum Bit Rate: %u\n", options.minrate);
ast_cli(a->fd, "\tMaximum Bit Rate: %u\n", options.maxrate);
ast_fax_modem_to_str(options.modems, modems, sizeof(modems));
ast_cli(a->fd, "\tModem Modulations Allowed: %s\n", modems);
ast_cli(a->fd, "\tT.38 Negotiation Timeout: %u\n", options.t38timeout);
ast_cli(a->fd, "\n\nFAX Technology Modules:\n\n");
AST_RWLIST_RDLOCK(&faxmodules);
AST_RWLIST_TRAVERSE(&faxmodules, fax, list) {
ast_cli(a->fd, "%s (%s) Settings:\n", fax->tech->type, fax->tech->description);
fax->tech->cli_show_settings(a->fd);
}
AST_RWLIST_UNLOCK(&faxmodules);
return CLI_SUCCESS;
}
/*! \brief display details of a specified fax session */
static char *cli_fax_show_session(struct ast_cli_entry *e, int cmd, struct ast_cli_args *a)
{
struct ast_fax_session *s, tmp;
switch (cmd) {
case CLI_INIT:
e->command = "fax show session";
e->usage =
"Usage: fax show session <session number>\n"
" Shows status of the named FAX session\n";
return NULL;
case CLI_GENERATE:
return fax_session_tab_complete(a);
}
if (a->argc != 4) {
return CLI_SHOWUSAGE;
}
if (sscanf(a->argv[3], "%u", &tmp.id) != 1) {
ast_log(LOG_ERROR, "invalid session id: '%s'\n", a->argv[3]);
return RESULT_SUCCESS;
}
ast_cli(a->fd, "\nFAX Session Details:\n--------------------\n\n");
s = ao2_find(faxregistry.container, &tmp, OBJ_POINTER);
if (s) {
ast_cli(a->fd, "%-22s : %s\n", "channel", s->channame);
s->tech->cli_show_session(s, a->fd);
ao2_ref(s, -1);
}
ast_cli(a->fd, "\n\n");
return CLI_SUCCESS;
}
static int manager_fax_session(struct mansession *s, const struct message *m)
{
const char *action_id = astman_get_header(m, "ActionID");
const char *session_number = astman_get_header(m, "SessionNumber");
char id_text[256] = "";
struct ast_fax_session *session;
struct ast_fax_session find_session;
if (sscanf(session_number, "%30u", &find_session.id) != 1) {
astman_send_error(s, m, "Invalid session ID");
return 0;
}
session = ao2_find(faxregistry.container, &find_session, OBJ_POINTER);
if (!session) {
astman_send_error(s, m, "Session not found");
return 0;
}
if (!session->tech->manager_fax_session) {
astman_send_error(s, m, "Fax technology doesn't provide a handler for FAXSession");
ao2_ref(session, -1);
return 0;
}
if (!ast_strlen_zero(action_id)) {
snprintf(id_text, sizeof(id_text), "ActionID: %s\r\n", action_id);
}
astman_send_ack(s, m, "FAXSession event will follow");
session->tech->manager_fax_session(s, id_text, session);
ao2_ref(session, -1);
return 0;
}
/*! \brief display fax stats */
static char *cli_fax_show_stats(struct ast_cli_entry *e, int cmd, struct ast_cli_args *a)
{
struct fax_module *fax;
switch (cmd) {
case CLI_INIT:
e->command = "fax show stats";
e->usage =
"Usage: fax show stats\n"
" Shows a statistical summary of FAX transmissions\n";
return NULL;
case CLI_GENERATE:
return NULL;
}
ast_cli(a->fd, "\nFAX Statistics:\n---------------\n\n");
ast_cli(a->fd, "%-20.20s : %d\n", "Current Sessions", faxregistry.active_sessions);
ast_cli(a->fd, "%-20.20s : %d\n", "Reserved Sessions", faxregistry.reserved_sessions);
ast_cli(a->fd, "%-20.20s : %d\n", "Transmit Attempts", faxregistry.fax_tx_attempts);
ast_cli(a->fd, "%-20.20s : %d\n", "Receive Attempts", faxregistry.fax_rx_attempts);
ast_cli(a->fd, "%-20.20s : %d\n", "Completed FAXes", faxregistry.fax_complete);
ast_cli(a->fd, "%-20.20s : %d\n", "Failed FAXes", faxregistry.fax_failures);
AST_RWLIST_RDLOCK(&faxmodules);
AST_RWLIST_TRAVERSE(&faxmodules, fax, list) {
fax->tech->cli_show_stats(a->fd);
}
AST_RWLIST_UNLOCK(&faxmodules);
ast_cli(a->fd, "\n\n");
return CLI_SUCCESS;
}
static int manager_fax_stats(struct mansession *s, const struct message *m)
{
const char *action_id = astman_get_header(m, "ActionID");
char id_text[256] = "";
astman_send_ack(s, m, "FAXStats event will follow");
if (!ast_strlen_zero(action_id)) {
snprintf(id_text, sizeof(id_text), "ActionID: %s\r\n", action_id);
}
astman_append(s, "Event: FAXStats\r\n"
"%s"
"CurrentSessions: %d\r\n"
"ReservedSessions: %d\r\n"
"TransmitAttempts: %d\r\n"
"ReceiveAttempts: %d\r\n"
"CompletedFAXes: %d\r\n"
"FailedFAXes: %d\r\n"
"\r\n",
id_text,
faxregistry.active_sessions, faxregistry.reserved_sessions,
faxregistry.fax_tx_attempts, faxregistry.fax_rx_attempts,
faxregistry.fax_complete, faxregistry.fax_failures);
return 0;
}
static const char *fax_session_type(struct ast_fax_session *s)
{
if (s->details->caps & AST_FAX_TECH_AUDIO) {
return "G.711";
}
if (s->details->caps & AST_FAX_TECH_T38) {
return "T.38";
}
return "none";
}
const char *ast_fax_session_operation_str(struct ast_fax_session *s)
{
if (s->details->caps & AST_FAX_TECH_GATEWAY) {
return "gateway";
}
if (s->details->caps & AST_FAX_TECH_SEND) {
return "send";
}
if (s->details->caps & AST_FAX_TECH_RECEIVE) {
return "receive";
}
if (s->details->caps & AST_FAX_TECH_V21_DETECT) {
return "V.21";
}
return "none";
}
/*! \brief display fax sessions */
static char *cli_fax_show_sessions(struct ast_cli_entry *e, int cmd, struct ast_cli_args *a)
{
struct ast_fax_session *s;
struct ao2_iterator i;
int session_count;
char *filenames;
switch (cmd) {
case CLI_INIT:
e->command = "fax show sessions";
e->usage =
"Usage: fax show sessions\n"
" Shows the current FAX sessions\n";
return NULL;
case CLI_GENERATE:
return NULL;
}
ast_cli(a->fd, "\nCurrent FAX Sessions:\n\n");
ast_cli(a->fd, "%-30.30s %-10.10s %-10.10s %-5.5s %-10.10s %-15.15s %-30.30s\n",
"Channel", "Tech", "FAXID", "Type", "Operation", "State", "File(s)");
i = ao2_iterator_init(faxregistry.container, 0);
while ((s = ao2_iterator_next(&i))) {
ao2_lock(s);
filenames = generate_filenames_string(s->details, "", ", ");
ast_cli(a->fd, "%-30.30s %-10.10s %-10u %-5.5s %-10.10s %-15.15s %-30s\n",
s->channame, s->tech->type, s->id,
fax_session_type(s),
ast_fax_session_operation_str(s),
ast_fax_state_to_str(s->state), S_OR(filenames, ""));
ast_free(filenames);
ao2_unlock(s);
ao2_ref(s, -1);
}
ao2_iterator_destroy(&i);
session_count = ao2_container_count(faxregistry.container);
ast_cli(a->fd, "\n%d FAX sessions\n\n", session_count);
return CLI_SUCCESS;
}
static int manager_fax_sessions_entry(struct mansession *s,
struct ast_fax_session *session, const char *id_text)
{
char *filenames;
ao2_lock(session);
filenames = generate_filenames_string(session->details, "", ",");
if (!filenames) {
ast_log(LOG_ERROR, "Error generating Files string");
ao2_unlock(session);
return -1;
}
astman_append(s, "Event: FAXSessionsEntry\r\n"
"%s" /* ActionID if present */
"Channel: %s\r\n" /* Channel name */
"Technology: %s\r\n" /* Fax session technology */
"SessionNumber: %u\r\n" /* Session ID */
"SessionType: %s\r\n" /* G711 or T38 */
"Operation: %s\r\n"
"State: %s\r\n"
"Files: %s\r\n"
"\r\n",
id_text, session->channame, session->tech->type, session->id,
fax_session_type(session), ast_fax_session_operation_str(session),
ast_fax_state_to_str(session->state), S_OR(filenames, ""));
ast_free(filenames);
ao2_unlock(session);
return 0;
}
static int manager_fax_sessions(struct mansession *s, const struct message *m)
{
const char *action_id = astman_get_header(m, "ActionID");
AMI: Make AMI actions that generate event lists consistent. * Made the following AMI actions use list API calls for consistency: Agents BridgeInfo BridgeList BridgeTechnologyList ConfbridgeLIst ConfbridgeLIstRooms CoreShowChannels DAHDIShowChannels DBGet DeviceStateList ExtensionStateList FAXSessions Hangup IAXpeerlist IAXpeers IAXregistry MeetmeList MeetmeListRooms MWIGet ParkedCalls Parkinglots PJSIPShowEndpoint PJSIPShowEndpoints PJSIPShowRegistrationsInbound PJSIPShowRegistrationsOutbound PJSIPShowResourceLists PJSIPShowSubscriptionsInbound PJSIPShowSubscriptionsOutbound PresenceStateList PRIShowSpans QueueStatus QueueSummary ShowDialPlan SIPpeers SIPpeerstatus SIPshowregistry SKINNYdevices SKINNYlines Status VoicemailUsersList * Incremented the AMI version to 2.7.0. * Changed astman_send_listack() to not use the listflag parameter and always set the value to "Start" so the start capitalization is consistent. i.e., The FAXSessions used "Start" while the rest of the system used "start". The corresponding complete event always used "Complete". * Fixed ami_show_resource_lists() "PJSIPShowResourceLists" to output the AMI ActionID for all of its list events. * Fixed off-nominal AMI protocol error in manager_bridge_info(), manager_parking_status_single_lot(), and manager_parking_status_all_lots(). Use of astman_send_error() after responding to the original AMI action request violates the action response pattern by sending two responses. * Fixed minor protocol error in action_getconfig() when no requested categories are found. Each line needs to be formatted as "Header: text". * Fixed off-nominal memory leak in manager_build_parked_call_string(). * Eliminated unnecessary use of RAII_VAR() in ami_subscription_detail(). ASTERISK-24049 #close Reported by: Jonathan Rose Review: https://reviewboard.asterisk.org/r/4315/ ........ Merged revisions 430434 from http://svn.asterisk.org/svn/asterisk/branches/13 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@430435 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2015-01-09 18:16:54 +00:00
char id_text[256];
struct ast_fax_session *session;
struct ao2_iterator iter;
int session_count = 0;
AMI: Make AMI actions that generate event lists consistent. * Made the following AMI actions use list API calls for consistency: Agents BridgeInfo BridgeList BridgeTechnologyList ConfbridgeLIst ConfbridgeLIstRooms CoreShowChannels DAHDIShowChannels DBGet DeviceStateList ExtensionStateList FAXSessions Hangup IAXpeerlist IAXpeers IAXregistry MeetmeList MeetmeListRooms MWIGet ParkedCalls Parkinglots PJSIPShowEndpoint PJSIPShowEndpoints PJSIPShowRegistrationsInbound PJSIPShowRegistrationsOutbound PJSIPShowResourceLists PJSIPShowSubscriptionsInbound PJSIPShowSubscriptionsOutbound PresenceStateList PRIShowSpans QueueStatus QueueSummary ShowDialPlan SIPpeers SIPpeerstatus SIPshowregistry SKINNYdevices SKINNYlines Status VoicemailUsersList * Incremented the AMI version to 2.7.0. * Changed astman_send_listack() to not use the listflag parameter and always set the value to "Start" so the start capitalization is consistent. i.e., The FAXSessions used "Start" while the rest of the system used "start". The corresponding complete event always used "Complete". * Fixed ami_show_resource_lists() "PJSIPShowResourceLists" to output the AMI ActionID for all of its list events. * Fixed off-nominal AMI protocol error in manager_bridge_info(), manager_parking_status_single_lot(), and manager_parking_status_all_lots(). Use of astman_send_error() after responding to the original AMI action request violates the action response pattern by sending two responses. * Fixed minor protocol error in action_getconfig() when no requested categories are found. Each line needs to be formatted as "Header: text". * Fixed off-nominal memory leak in manager_build_parked_call_string(). * Eliminated unnecessary use of RAII_VAR() in ami_subscription_detail(). ASTERISK-24049 #close Reported by: Jonathan Rose Review: https://reviewboard.asterisk.org/r/4315/ ........ Merged revisions 430434 from http://svn.asterisk.org/svn/asterisk/branches/13 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@430435 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2015-01-09 18:16:54 +00:00
id_text[0] = '\0';
if (!ast_strlen_zero(action_id)) {
snprintf(id_text, sizeof(id_text), "ActionID: %s\r\n", action_id);
}
astman_send_listack(s, m, "FAXSessionsEntry event list will follow", "Start");
AMI: Make AMI actions that generate event lists consistent. * Made the following AMI actions use list API calls for consistency: Agents BridgeInfo BridgeList BridgeTechnologyList ConfbridgeLIst ConfbridgeLIstRooms CoreShowChannels DAHDIShowChannels DBGet DeviceStateList ExtensionStateList FAXSessions Hangup IAXpeerlist IAXpeers IAXregistry MeetmeList MeetmeListRooms MWIGet ParkedCalls Parkinglots PJSIPShowEndpoint PJSIPShowEndpoints PJSIPShowRegistrationsInbound PJSIPShowRegistrationsOutbound PJSIPShowResourceLists PJSIPShowSubscriptionsInbound PJSIPShowSubscriptionsOutbound PresenceStateList PRIShowSpans QueueStatus QueueSummary ShowDialPlan SIPpeers SIPpeerstatus SIPshowregistry SKINNYdevices SKINNYlines Status VoicemailUsersList * Incremented the AMI version to 2.7.0. * Changed astman_send_listack() to not use the listflag parameter and always set the value to "Start" so the start capitalization is consistent. i.e., The FAXSessions used "Start" while the rest of the system used "start". The corresponding complete event always used "Complete". * Fixed ami_show_resource_lists() "PJSIPShowResourceLists" to output the AMI ActionID for all of its list events. * Fixed off-nominal AMI protocol error in manager_bridge_info(), manager_parking_status_single_lot(), and manager_parking_status_all_lots(). Use of astman_send_error() after responding to the original AMI action request violates the action response pattern by sending two responses. * Fixed minor protocol error in action_getconfig() when no requested categories are found. Each line needs to be formatted as "Header: text". * Fixed off-nominal memory leak in manager_build_parked_call_string(). * Eliminated unnecessary use of RAII_VAR() in ami_subscription_detail(). ASTERISK-24049 #close Reported by: Jonathan Rose Review: https://reviewboard.asterisk.org/r/4315/ ........ Merged revisions 430434 from http://svn.asterisk.org/svn/asterisk/branches/13 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@430435 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2015-01-09 18:16:54 +00:00
iter = ao2_iterator_init(faxregistry.container, 0);
while ((session = ao2_iterator_next(&iter))) {
if (!manager_fax_sessions_entry(s, session, id_text)) {
session_count++;
}
ao2_ref(session, -1);
}
ao2_iterator_destroy(&iter);
AMI: Make AMI actions that generate event lists consistent. * Made the following AMI actions use list API calls for consistency: Agents BridgeInfo BridgeList BridgeTechnologyList ConfbridgeLIst ConfbridgeLIstRooms CoreShowChannels DAHDIShowChannels DBGet DeviceStateList ExtensionStateList FAXSessions Hangup IAXpeerlist IAXpeers IAXregistry MeetmeList MeetmeListRooms MWIGet ParkedCalls Parkinglots PJSIPShowEndpoint PJSIPShowEndpoints PJSIPShowRegistrationsInbound PJSIPShowRegistrationsOutbound PJSIPShowResourceLists PJSIPShowSubscriptionsInbound PJSIPShowSubscriptionsOutbound PresenceStateList PRIShowSpans QueueStatus QueueSummary ShowDialPlan SIPpeers SIPpeerstatus SIPshowregistry SKINNYdevices SKINNYlines Status VoicemailUsersList * Incremented the AMI version to 2.7.0. * Changed astman_send_listack() to not use the listflag parameter and always set the value to "Start" so the start capitalization is consistent. i.e., The FAXSessions used "Start" while the rest of the system used "start". The corresponding complete event always used "Complete". * Fixed ami_show_resource_lists() "PJSIPShowResourceLists" to output the AMI ActionID for all of its list events. * Fixed off-nominal AMI protocol error in manager_bridge_info(), manager_parking_status_single_lot(), and manager_parking_status_all_lots(). Use of astman_send_error() after responding to the original AMI action request violates the action response pattern by sending two responses. * Fixed minor protocol error in action_getconfig() when no requested categories are found. Each line needs to be formatted as "Header: text". * Fixed off-nominal memory leak in manager_build_parked_call_string(). * Eliminated unnecessary use of RAII_VAR() in ami_subscription_detail(). ASTERISK-24049 #close Reported by: Jonathan Rose Review: https://reviewboard.asterisk.org/r/4315/ ........ Merged revisions 430434 from http://svn.asterisk.org/svn/asterisk/branches/13 git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@430435 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2015-01-09 18:16:54 +00:00
astman_send_list_complete_start(s, m, "FAXSessionsComplete", session_count);
astman_append(s, "Total: %d\r\n", session_count);
astman_send_list_complete_end(s);
return 0;
}
static struct ast_cli_entry fax_cli[] = {
AST_CLI_DEFINE(cli_fax_show_version, "Show versions of FAX For Asterisk components"),
AST_CLI_DEFINE(cli_fax_set_debug, "Enable/Disable FAX debugging on new FAX sessions"),
AST_CLI_DEFINE(cli_fax_show_capabilities, "Show the capabilities of the registered FAX technology modules"),
AST_CLI_DEFINE(cli_fax_show_settings, "Show the global settings and defaults of both the FAX core and technology modules"),
AST_CLI_DEFINE(cli_fax_show_session, "Show the status of the named FAX sessions"),
AST_CLI_DEFINE(cli_fax_show_sessions, "Show the current FAX sessions"),
AST_CLI_DEFINE(cli_fax_show_stats, "Summarize FAX session history"),
};
static void set_general_options(const struct fax_options *options)
{
ast_rwlock_wrlock(&options_lock);
general_options = *options;
ast_rwlock_unlock(&options_lock);
}
static void get_general_options(struct fax_options *options)
{
ast_rwlock_rdlock(&options_lock);
*options = general_options;
ast_rwlock_unlock(&options_lock);
}
static int set_t38timeout(const char *value, unsigned int *t38timeout)
{
unsigned int timeout;
if (sscanf(value, "%u", &timeout) != 1) {
ast_log(LOG_ERROR, "Unable to get timeout from '%s'\n", value);
return -1;
} else if (timeout) {
*t38timeout = timeout;
} else {
ast_log(LOG_ERROR, "T.38 negotiation timeout must be non-zero\n");
return -1;
}
return 0;
}
/*! \brief configure res_fax */
static int set_config(int reload)
{
struct ast_config *cfg;
struct ast_variable *v;
struct ast_flags config_flags = { reload ? CONFIG_FLAG_FILEUNCHANGED : 0 };
char modems[128] = "";
struct fax_options options;
int res = 0;
options = default_options;
/* When we're not reloading, we have to be certain to set the general options
* to the defaults in case config loading goes wrong at some point. On a reload,
* the general options need to stay the same as what they were prior to the
* reload rather than being reset to the defaults.
*/
if (!reload) {
set_general_options(&options);
}
/* read configuration */
if (!(cfg = ast_config_load2(config, "res_fax", config_flags))) {
ast_log(LOG_NOTICE, "Configuration file '%s' not found, %s options.\n",
config, reload ? "not changing" : "using default");
return 0;
}
if (cfg == CONFIG_STATUS_FILEINVALID) {
ast_log(LOG_NOTICE, "Configuration file '%s' is invalid, %s options.\n",
config, reload ? "not changing" : "using default");
return 0;
}
if (cfg == CONFIG_STATUS_FILEUNCHANGED) {
return 0;
}
if (reload) {
options = default_options;
}
/* create configuration */
for (v = ast_variable_browse(cfg, "general"); v; v = v->next) {
int rate;
if (!strcasecmp(v->name, "minrate")) {
ast_debug(3, "reading minrate '%s' from configuration file\n", v->value);
if ((rate = fax_rate_str_to_int(v->value)) == 0) {
res = -1;
goto end;
}
options.minrate = rate;
} else if (!strcasecmp(v->name, "maxrate")) {
ast_debug(3, "reading maxrate '%s' from configuration file\n", v->value);
if ((rate = fax_rate_str_to_int(v->value)) == 0) {
res = -1;
goto end;
}
options.maxrate = rate;
} else if (!strcasecmp(v->name, "statusevents")) {
ast_debug(3, "reading statusevents '%s' from configuration file\n", v->value);
options.statusevents = ast_true(v->value);
} else if (!strcasecmp(v->name, "ecm")) {
ast_debug(3, "reading ecm '%s' from configuration file\n", v->value);
options.ecm = ast_true(v->value);
} else if ((!strcasecmp(v->name, "modem")) || (!strcasecmp(v->name, "modems"))) {
options.modems = 0;
update_modem_bits(&options.modems, v->value);
} else if (!strcasecmp(v->name, "t38timeout")) {
if (set_t38timeout(v->value, &options.t38timeout)) {
res = -1;
goto end;
}
}
}
if (options.maxrate < options.minrate) {
ast_log(LOG_ERROR, "maxrate %u is less than minrate %u\n", options.maxrate, options.minrate);
res = -1;
goto end;
}
if (check_modem_rate(options.modems, options.minrate)) {
ast_fax_modem_to_str(options.modems, modems, sizeof(modems));
ast_log(LOG_ERROR, "'modems' setting '%s' is incompatible with 'minrate' setting %u\n", modems, options.minrate);
res = -1;
goto end;
}
if (check_modem_rate(options.modems, options.maxrate)) {
ast_fax_modem_to_str(options.modems, modems, sizeof(modems));
ast_log(LOG_ERROR, "'modems' setting '%s' is incompatible with 'maxrate' setting %u\n", modems, options.maxrate);
res = -1;
goto end;
}
set_general_options(&options);
end:
ast_config_destroy(cfg);
return res;
}
/*! \brief FAXOPT read function returns the contents of a FAX option */
static int acf_faxopt_read(struct ast_channel *chan, const char *cmd, char *data, char *buf, size_t len)
{
struct ast_fax_session_details *details = find_details(chan);
int res = 0;
char *filenames;
if (!details) {
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_ERROR, "channel '%s' can't read FAXOPT(%s) because it has never been written.\n", ast_channel_name(chan), data);
return -1;
}
if (!strcasecmp(data, "ecm")) {
ast_copy_string(buf, details->option.ecm ? "yes" : "no", len);
} else if (!strcasecmp(data, "t38gateway") || !strcasecmp(data, "gateway") ||
!strcasecmp(data, "t38_gateway") || !strcasecmp(data, "faxgateway")) {
ast_copy_string(buf, details->gateway_id != -1 ? "yes" : "no", len);
} else if (!strcasecmp(data, "faxdetect")) {
ast_copy_string(buf, details->faxdetect_id != -1 ? "yes" : "no", len);
} else if (!strcasecmp(data, "error")) {
ast_copy_string(buf, details->error, len);
} else if (!strcasecmp(data, "filename")) {
if (AST_LIST_EMPTY(&details->documents)) {
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_ERROR, "channel '%s' can't read FAXOPT(%s) because it has never been written.\n", ast_channel_name(chan), data);
res = -1;
} else {
ast_copy_string(buf, AST_LIST_FIRST(&details->documents)->filename, len);
}
} else if (!strcasecmp(data, "filenames")) {
if (AST_LIST_EMPTY(&details->documents)) {
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_ERROR, "channel '%s' can't read FAXOPT(%s) because it has never been written.\n", ast_channel_name(chan), data);
res = -1;
} else if ((filenames = generate_filenames_string(details, "", ","))) {
ast_copy_string(buf, filenames, len);
ast_free(filenames);
} else {
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_ERROR, "channel '%s' can't read FAXOPT(%s), there was an error generating the filenames list.\n", ast_channel_name(chan), data);
res = -1;
}
} else if (!strcasecmp(data, "headerinfo")) {
ast_copy_string(buf, details->headerinfo, len);
} else if (!strcasecmp(data, "localstationid")) {
ast_copy_string(buf, details->localstationid, len);
} else if (!strcasecmp(data, "maxrate")) {
snprintf(buf, len, "%u", details->maxrate);
} else if (!strcasecmp(data, "minrate")) {
snprintf(buf, len, "%u", details->minrate);
} else if (!strcasecmp(data, "pages")) {
snprintf(buf, len, "%u", details->pages_transferred);
} else if (!strcasecmp(data, "rate")) {
ast_copy_string(buf, details->transfer_rate, len);
} else if (!strcasecmp(data, "remotestationid")) {
ast_copy_string(buf, details->remotestationid, len);
} else if (!strcasecmp(data, "resolution")) {
ast_copy_string(buf, details->resolution, len);
} else if (!strcasecmp(data, "sessionid")) {
snprintf(buf, len, "%u", details->id);
} else if (!strcasecmp(data, "status")) {
ast_copy_string(buf, details->result, len);
} else if (!strcasecmp(data, "statusstr")) {
ast_copy_string(buf, details->resultstr, len);
} else if ((!strcasecmp(data, "modem")) || (!strcasecmp(data, "modems"))) {
ast_fax_modem_to_str(details->modems, buf, len);
} else if (!strcasecmp(data, "t38timeout")) {
snprintf(buf, len, "%u", details->t38timeout);
} else if (!strcasecmp(data, "negotiate_both")) {
ast_copy_string(buf, details->negotiate_both != -1 ? "yes" : "no", len);
} else {
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' can't read FAXOPT(%s) because it is unhandled!\n", ast_channel_name(chan), data);
res = -1;
}
ao2_ref(details, -1);
return res;
}
/*! \brief FAXOPT write function modifies the contents of a FAX option */
static int acf_faxopt_write(struct ast_channel *chan, const char *cmd, char *data, const char *value)
{
int res = 0;
struct ast_fax_session_details *details;
if (!(details = find_or_create_details(chan))) {
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_${fieldname}' and then writing setters/getters and converting the existing code to using them. When all fields are done, we can move ast_channel to a C file from channel.h and lop off the '__do_not_use_'. This patch sets up main/channel_interal_api.c to be the only file that actually accesses the ast_channel's fields directly. The intent would be for any API functions in channel.c to use the accessor functions. No more monkeying around with channel internals. We should use our own APIs. The interesting changes in this patch are the addition of channel_internal_api.c, the moving of the AST_DATA stuff from channel.c to channel_internal_api.c (note: the AST_DATA stuff will have to be reworked to use accessor functions when ast_channel is really opaque), and some re-working of the way channel iterators/callbacks are handled so as to avoid creating fake ast_channels on the stack to pass in matching data by directly accessing fields (since "name" is a stringfield and the fake channel doesn't init the stringfields, you can't use the ast_channel_name_set() function). I went with ast_channel_name(chan) for a getter, and ast_channel_name_set(chan, name) for a setter. The majority of the grunt-work for this change was done by writing a semantic patch using Coccinelle ( http://coccinelle.lip6.fr/ ). Review: https://reviewboard.asterisk.org/r/1655/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@350223 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-01-09 22:15:50 +00:00
ast_log(LOG_WARNING, "channel '%s' can't set FAXOPT(%s) to '%s' because it failed to create a datastore.\n", ast_channel_name(chan), data, value);
return -1;
}
Replace direct access to channel name with accessor functions There are many benefits to making the ast_channel an opaque handle, from increasing maintainability to presenting ways to kill masquerades. This patch kicks things off by taking things a field at a time, renaming the field to '__do_not_use_${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_debug(3, "channel '%s' setting FAXOPT(%s) to '%s'\n", ast_channel_name(chan), data, value);
if (!strcasecmp(data, "ecm")) {
const char *val = ast_skip_blanks(value);
if (ast_true(val)) {
details->option.ecm = AST_FAX_OPTFLAG_TRUE;
} else if (ast_false(val)) {
details->option.ecm = AST_FAX_OPTFLAG_FALSE;
} else {
ast_log(LOG_WARNING, "Unsupported value '%s' passed to FAXOPT(ecm).\n", value);
}
} else if (!strcasecmp(data, "t38gateway") || !strcasecmp(data, "gateway") ||
!strcasecmp(data, "t38_gateway") || !strcasecmp(data, "faxgateway")) {
const char *val = ast_skip_blanks(value);
char *timeout = strchr(val, ',');
if (timeout) {
*timeout++ = '\0';
}
if (ast_true(val)) {
if (details->gateway_id < 0) {
details->gateway_timeout = 0;
if (timeout) {
unsigned int gwtimeout;
if (sscanf(timeout, "%30u", &gwtimeout) == 1) {
details->gateway_timeout = gwtimeout * 1000;
} else {
ast_log(LOG_WARNING, "Unsupported timeout '%s' passed to FAXOPT(%s).\n", timeout, data);
}
}
details->gateway_id = fax_gateway_attach(chan, details);
if (details->gateway_id < 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_${fieldname}' and then writing setters/getters and converting the existing code to using them. When all fields are done, we can move ast_channel to a C file from channel.h and lop off the '__do_not_use_'. This patch sets up main/channel_interal_api.c to be the only file that actually accesses the ast_channel's fields directly. The intent would be for any API functions in channel.c to use the accessor functions. No more monkeying around with channel internals. We should use our own APIs. The interesting changes in this patch are the addition of channel_internal_api.c, the moving of the AST_DATA stuff from channel.c to channel_internal_api.c (note: the AST_DATA stuff will have to be reworked to use accessor functions when ast_channel is really opaque), and some re-working of the way channel iterators/callbacks are handled so as to avoid creating fake ast_channels on the stack to pass in matching data by directly accessing fields (since "name" is a stringfield and the fake channel doesn't init the stringfields, you can't use the ast_channel_name_set() function). I went with ast_channel_name(chan) for a getter, and ast_channel_name_set(chan, name) for a setter. The majority of the grunt-work for this change was done by writing a semantic patch using Coccinelle ( http://coccinelle.lip6.fr/ ). Review: https://reviewboard.asterisk.org/r/1655/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@350223 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-01-09 22:15:50 +00:00
ast_log(LOG_ERROR, "Error attaching T.38 gateway to channel %s.\n", ast_channel_name(chan));
res = -1;
} else {
Replace direct access to channel name with accessor functions There are many benefits to making the ast_channel an opaque handle, from increasing maintainability to presenting ways to kill masquerades. This patch kicks things off by taking things a field at a time, renaming the field to '__do_not_use_${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_debug(1, "Attached T.38 gateway to channel %s.\n", ast_channel_name(chan));
}
} else {
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, "Attempt to attach a T.38 gateway on channel (%s) with gateway already running.\n", ast_channel_name(chan));
}
} else if (ast_false(val)) {
ast_channel_lock(chan);
ast_framehook_detach(chan, details->gateway_id);
ast_channel_unlock(chan);
details->gateway_id = -1;
} else {
ast_log(LOG_WARNING, "Unsupported value '%s' passed to FAXOPT(%s).\n", value, data);
}
} else if (!strcasecmp(data, "faxdetect")) {
const char *val = ast_skip_blanks(value);
char *timeout = strchr(val, ',');
unsigned int fdtimeout = 0;
int flags;
int faxdetect;
if (timeout) {
*timeout++ = '\0';
}
if (ast_true(val) || !strcasecmp(val, "t38") || !strcasecmp(val, "cng")) {
if (details->faxdetect_id < 0) {
if (timeout) {
if (sscanf(timeout, "%30u", &fdtimeout) == 1) {
fdtimeout *= 1000;
} else {
ast_log(LOG_WARNING, "Unsupported timeout '%s' passed to FAXOPT(%s).\n",
timeout, data);
}
}
if (!strcasecmp(val, "t38")) {
flags = FAX_DETECT_MODE_T38;
} else if (!strcasecmp(val, "cng")) {
flags = FAX_DETECT_MODE_CNG;
} else {
flags = FAX_DETECT_MODE_BOTH;
}
faxdetect = fax_detect_attach(chan, fdtimeout, flags);
if (faxdetect < 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_ERROR, "Error attaching FAX detect to channel %s.\n", ast_channel_name(chan));
res = -1;
} else {
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_${fieldname}' and then writing setters/getters and converting the existing code to using them. When all fields are done, we can move ast_channel to a C file from channel.h and lop off the '__do_not_use_'. This patch sets up main/channel_interal_api.c to be the only file that actually accesses the ast_channel's fields directly. The intent would be for any API functions in channel.c to use the accessor functions. No more monkeying around with channel internals. We should use our own APIs. The interesting changes in this patch are the addition of channel_internal_api.c, the moving of the AST_DATA stuff from channel.c to channel_internal_api.c (note: the AST_DATA stuff will have to be reworked to use accessor functions when ast_channel is really opaque), and some re-working of the way channel iterators/callbacks are handled so as to avoid creating fake ast_channels on the stack to pass in matching data by directly accessing fields (since "name" is a stringfield and the fake channel doesn't init the stringfields, you can't use the ast_channel_name_set() function). I went with ast_channel_name(chan) for a getter, and ast_channel_name_set(chan, name) for a setter. The majority of the grunt-work for this change was done by writing a semantic patch using Coccinelle ( http://coccinelle.lip6.fr/ ). Review: https://reviewboard.asterisk.org/r/1655/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@350223 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2012-01-09 22:15:50 +00:00
ast_debug(1, "Attached FAX detect to channel %s.\n", ast_channel_name(chan));
}
} else {
Replace direct access to channel name with accessor functions There are many benefits to making the ast_channel an opaque handle, from increasing maintainability to presenting ways to kill masquerades. This patch kicks things off by taking things a field at a time, renaming the field to '__do_not_use_${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, "Attempt to attach a FAX detect on channel (%s) with FAX detect already running.\n", ast_channel_name(chan));
}
} else if (ast_false(val)) {
ast_channel_lock(chan);
ast_framehook_detach(chan, details->faxdetect_id);
ast_channel_unlock(chan);
details->faxdetect_id = -1;
} else {
ast_log(LOG_WARNING, "Unsupported value '%s' passed to FAXOPT(%s).\n", value, data);
}
} else if (!strcasecmp(data, "headerinfo")) {
ast_string_field_set(details, headerinfo, value);
} else if (!strcasecmp(data, "localstationid")) {
ast_string_field_set(details, localstationid, value);
} else if (!strcasecmp(data, "maxrate")) {
details->maxrate = fax_rate_str_to_int(value);
if (!details->maxrate) {
details->maxrate = ast_fax_maxrate();
}
} else if (!strcasecmp(data, "minrate")) {
details->minrate = fax_rate_str_to_int(value);
if (!details->minrate) {
details->minrate = ast_fax_minrate();
}
} else if (!strcasecmp(data, "t38timeout")) {
if (set_t38timeout(value, &details->t38timeout)) {
res = -1;
}
} else if ((!strcasecmp(data, "modem")) || (!strcasecmp(data, "modems"))) {
update_modem_bits(&details->modems, value);
} else if (!strcasecmp(data, "negotiate_both")) {
details->negotiate_both = ast_true(ast_skip_blanks(value));
} else {
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' set FAXOPT(%s) to '%s' is unhandled!\n", ast_channel_name(chan), data, value);
res = -1;
}
ao2_ref(details, -1);
return res;
}
/*! \brief FAXOPT dialplan function */
static struct ast_custom_function acf_faxopt = {
.name = "FAXOPT",
.read = acf_faxopt_read,
.write = acf_faxopt_write,
};
/*! \brief unload res_fax */
static int unload_module(void)
{
ast_cli_unregister_multiple(fax_cli, ARRAY_LEN(fax_cli));
if (ast_custom_function_unregister(&acf_faxopt) < 0) {
ast_log(LOG_WARNING, "failed to unregister function '%s'\n", acf_faxopt.name);
}
if (ast_unregister_application(app_sendfax) < 0) {
ast_log(LOG_WARNING, "failed to unregister '%s'\n", app_sendfax);
}
if (ast_unregister_application(app_receivefax) < 0) {
ast_log(LOG_WARNING, "failed to unregister '%s'\n", app_receivefax);
}
ast_manager_unregister("FAXSessions");
ast_manager_unregister("FAXSession");
ast_manager_unregister("FAXStats");
if (fax_logger_level != -1) {
ast_logger_unregister_level("FAX");
}
ao2_ref(faxregistry.container, -1);
return 0;
}
/*!
* \brief Load the module
*
* Module loading including tests for configuration or dependencies.
* This function can return AST_MODULE_LOAD_FAILURE, AST_MODULE_LOAD_DECLINE,
* or AST_MODULE_LOAD_SUCCESS. If a dependency or environment variable fails
* tests return AST_MODULE_LOAD_FAILURE. If the module can not load the
* configuration file or other non-critical problem return
* AST_MODULE_LOAD_DECLINE. On success return AST_MODULE_LOAD_SUCCESS.
*/
static int load_module(void)
{
int res;
/* initialize the registry */
faxregistry.active_sessions = 0;
faxregistry.reserved_sessions = 0;
faxregistry.container = ao2_container_alloc_hash(AO2_ALLOC_OPT_LOCK_MUTEX, 0,
FAX_MAXBUCKETS, session_hash_cb, NULL, session_cmp_cb);
if (!faxregistry.container) {
return AST_MODULE_LOAD_DECLINE;
}
if (set_config(0) < 0) {
ast_log(LOG_ERROR, "failed to load configuration file '%s'\n", config);
ao2_ref(faxregistry.container, -1);
return AST_MODULE_LOAD_DECLINE;
}
/* register CLI operations and applications */
if (ast_register_application_xml(app_sendfax, sendfax_exec) < 0) {
ast_log(LOG_WARNING, "failed to register '%s'.\n", app_sendfax);
ao2_ref(faxregistry.container, -1);
return AST_MODULE_LOAD_DECLINE;
}
if (ast_register_application_xml(app_receivefax, receivefax_exec) < 0) {
ast_log(LOG_WARNING, "failed to register '%s'.\n", app_receivefax);
ast_unregister_application(app_sendfax);
ao2_ref(faxregistry.container, -1);
return AST_MODULE_LOAD_DECLINE;
}
if (ast_manager_register_xml("FAXSessions", EVENT_FLAG_CALL, manager_fax_sessions)) {
ast_log(LOG_WARNING, "failed to register 'FAXSessions' AMI command.\n");
ast_unregister_application(app_receivefax);
ast_unregister_application(app_sendfax);
ao2_ref(faxregistry.container, -1);
return AST_MODULE_LOAD_DECLINE;
}
if (ast_manager_register_xml("FAXSession", EVENT_FLAG_CALL, manager_fax_session)) {
ast_log(LOG_WARNING, "failed to register 'FAXSession' AMI command.\n");
ast_manager_unregister("FAXSession");
ast_unregister_application(app_receivefax);
ast_unregister_application(app_sendfax);
ao2_ref(faxregistry.container, -1);
return AST_MODULE_LOAD_DECLINE;
}
if (ast_manager_register_xml("FAXStats", EVENT_FLAG_REPORTING, manager_fax_stats)) {
ast_log(LOG_WARNING, "failed to register 'FAXStats' AMI command.\n");
ast_manager_unregister("FAXSession");
ast_manager_unregister("FAXSessions");
ast_unregister_application(app_receivefax);
ast_unregister_application(app_sendfax);
ao2_ref(faxregistry.container, -1);
return AST_MODULE_LOAD_DECLINE;
}
ast_cli_register_multiple(fax_cli, ARRAY_LEN(fax_cli));
res = ast_custom_function_register(&acf_faxopt);
fax_logger_level = ast_logger_register_level("FAX");
return res;
}
static int reload_module(void)
{
set_config(1);
return 0;
}
AST_MODULE_INFO(ASTERISK_GPL_KEY, AST_MODFLAG_GLOBAL_SYMBOLS | AST_MODFLAG_LOAD_ORDER, "Generic FAX Applications",
.support_level = AST_MODULE_SUPPORT_CORE,
.load = load_module,
.unload = unload_module,
.reload = reload_module,
.load_pri = AST_MODPRI_APP_DEPEND,
);