asterisk/apps/app_chanspy.c

1464 lines
46 KiB
C
Raw Normal View History

/*
* Asterisk -- An open source telephony toolkit.
*
* Copyright (C) 2005 Anthony Minessale II (anthmct@yahoo.com)
Merged revisions 108135 via svnmerge from https://origsvn.digium.com/svn/asterisk/branches/1.4 ........ r108135 | russell | 2008-03-12 14:57:42 -0500 (Wed, 12 Mar 2008) | 40 lines (closes issue #12187, reported by atis, fixed by me after some brainstorming on the issue with mmichelson) - Update copyright info on app_chanspy. - Fix a race condition that caused app_chanspy to crash. The issue was that the chanspy datastore magic that was used to ensure that spyee channels did not disappear out from under the code did not completely solve the problem. It was actually possible for chanspy to acquire a channel reference out of its datastore to a channel that was in the middle of being destroyed. That was because datastore destruction in ast_channel_free() was done near the end. So, this left the code in app_chanspy accessing a channel that was partially, or completely invalid because it was in the process of being free'd by another thread. The following sort of shows the code path where the race occurred: ============================================================================= Thread 1 (PBX thread for spyee chan) || Thread 2 (chanspy) --------------------------------------||------------------------------------- ast_channel_free() || - remove channel from channel list || - lock/unlock the channel to ensure || that no references retrieved from || the channel list exist. || --------------------------------------||------------------------------------- || channel_spy() - destroy some channel data || - Lock chanspy datastore || - Retrieve reference to channel || - lock channel || - Unlock chanspy datastore --------------------------------------||------------------------------------- - destroy channel datastores || - call chanspy datastore d'tor || which NULL's out the ds' || - Operate on the channel ... reference to the channel || || - free the channel || || || - unlock the channel --------------------------------------||------------------------------------- ============================================================================= ........ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@108137 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2008-03-12 19:59:05 +00:00
* Copyright (C) 2005 - 2008, Digium, Inc.
*
* A license has been granted to Digium (via disclaimer) for the use of
* this code.
*
* See http://www.asterisk.org for more information about
* the Asterisk project. Please do not directly contact
* any of the maintainers of this project for assistance;
* the project provides a web site, mailing lists and IRC
* channels for your use.
*
* This program is free software, distributed under the terms of
* the GNU General Public License Version 2. See the LICENSE file
* at the top of the source tree.
*/
/*! \file
*
* \brief ChanSpy: Listen in on any channel.
*
* \author Anthony Minessale II <anthmct@yahoo.com>
Merged revisions 108135 via svnmerge from https://origsvn.digium.com/svn/asterisk/branches/1.4 ........ r108135 | russell | 2008-03-12 14:57:42 -0500 (Wed, 12 Mar 2008) | 40 lines (closes issue #12187, reported by atis, fixed by me after some brainstorming on the issue with mmichelson) - Update copyright info on app_chanspy. - Fix a race condition that caused app_chanspy to crash. The issue was that the chanspy datastore magic that was used to ensure that spyee channels did not disappear out from under the code did not completely solve the problem. It was actually possible for chanspy to acquire a channel reference out of its datastore to a channel that was in the middle of being destroyed. That was because datastore destruction in ast_channel_free() was done near the end. So, this left the code in app_chanspy accessing a channel that was partially, or completely invalid because it was in the process of being free'd by another thread. The following sort of shows the code path where the race occurred: ============================================================================= Thread 1 (PBX thread for spyee chan) || Thread 2 (chanspy) --------------------------------------||------------------------------------- ast_channel_free() || - remove channel from channel list || - lock/unlock the channel to ensure || that no references retrieved from || the channel list exist. || --------------------------------------||------------------------------------- || channel_spy() - destroy some channel data || - Lock chanspy datastore || - Retrieve reference to channel || - lock channel || - Unlock chanspy datastore --------------------------------------||------------------------------------- - destroy channel datastores || - call chanspy datastore d'tor || which NULL's out the ds' || - Operate on the channel ... reference to the channel || || - free the channel || || || - unlock the channel --------------------------------------||------------------------------------- ============================================================================= ........ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@108137 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2008-03-12 19:59:05 +00:00
* \author Joshua Colp <jcolp@digium.com>
* \author Russell Bryant <russell@digium.com>
*
* \ingroup applications
*/
/*** MODULEINFO
<support_level>core</support_level>
***/
#include "asterisk.h"
ASTERISK_FILE_VERSION(__FILE__, "$Revision$")
#include <ctype.h>
#include <errno.h>
#include "asterisk/paths.h" /* use ast_config_AST_MONITOR_DIR */
#include "asterisk/file.h"
#include "asterisk/channel.h"
#include "asterisk/audiohook.h"
#include "asterisk/features.h"
#include "asterisk/app.h"
#include "asterisk/utils.h"
#include "asterisk/say.h"
#include "asterisk/pbx.h"
#include "asterisk/translate.h"
#include "asterisk/manager.h"
#include "asterisk/module.h"
#include "asterisk/lock.h"
#include "asterisk/options.h"
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
#include "asterisk/autochan.h"
#include "asterisk/stasis_channels.h"
#include "asterisk/json.h"
#define AST_NAME_STRLEN 256
#define NUM_SPYGROUPS 128
/*** DOCUMENTATION
<application name="ChanSpy" language="en_US">
<synopsis>
Listen to a channel, and optionally whisper into it.
</synopsis>
<syntax>
<parameter name="chanprefix" />
<parameter name="options">
<optionlist>
<option name="b">
<para>Only spy on channels involved in a bridged call.</para>
</option>
<option name="B">
<para>Instead of whispering on a single channel barge in on both
channels involved in the call.</para>
</option>
<option name="c">
<argument name="digit" required="true">
<para>Specify a DTMF digit that can be used to spy on the next available channel.</para>
</argument>
</option>
<option name="d">
<para>Override the typical numeric DTMF functionality and instead
use DTMF to switch between spy modes.</para>
<enumlist>
<enum name="4">
<para>spy mode</para>
</enum>
<enum name="5">
<para>whisper mode</para>
</enum>
<enum name="6">
<para>barge mode</para>
</enum>
</enumlist>
</option>
<option name="e">
<argument name="ext" required="true" />
<para>Enable <emphasis>enforced</emphasis> mode, so the spying channel can
only monitor extensions whose name is in the <replaceable>ext</replaceable> : delimited
list.</para>
</option>
<option name="E">
<para>Exit when the spied-on channel hangs up.</para>
</option>
<option name="g">
<argument name="grp" required="true">
<para>Only spy on channels in which one or more of the groups
listed in <replaceable>grp</replaceable> matches one or more groups from the
<variable>SPYGROUP</variable> variable set on the channel to be spied upon.</para>
</argument>
<note><para>both <replaceable>grp</replaceable> and <variable>SPYGROUP</variable> can contain
either a single group or a colon-delimited list of groups, such
as <literal>sales:support:accounting</literal>.</para></note>
</option>
<option name="n" argsep="@">
<para>Say the name of the person being spied on if that person has recorded
his/her name. If a context is specified, then that voicemail context will
be searched when retrieving the name, otherwise the <literal>default</literal> context
be used when searching for the name (i.e. if SIP/1000 is the channel being
spied on and no mailbox is specified, then <literal>1000</literal> will be used when searching
for the name).</para>
<argument name="mailbox" />
<argument name="context" />
</option>
<option name="o">
<para>Only listen to audio coming from this channel.</para>
</option>
<option name="q">
<para>Don't play a beep when beginning to spy on a channel, or speak the
selected channel name.</para>
</option>
<option name="r">
<para>Record the session to the monitor spool directory. An optional base for the filename
may be specified. The default is <literal>chanspy</literal>.</para>
<argument name="basename" />
</option>
<option name="s">
<para>Skip the playback of the channel type (i.e. SIP, IAX, etc) when
speaking the selected channel name.</para>
</option>
<option name="S">
<para>Stop when no more channels are left to spy on.</para>
</option>
<option name="u">
<para>The <literal>chanprefix</literal> parameter is a channel uniqueid
or fully specified channel name.</para>
</option>
<option name="v">
<argument name="value" />
<para>Adjust the initial volume in the range from <literal>-4</literal>
to <literal>4</literal>. A negative value refers to a quieter setting.</para>
</option>
<option name="w">
<para>Enable <literal>whisper</literal> mode, so the spying channel can talk to
the spied-on channel.</para>
</option>
<option name="W">
<para>Enable <literal>private whisper</literal> mode, so the spying channel can
talk to the spied-on channel but cannot listen to that channel.</para>
</option>
<option name="x">
<argument name="digit" required="true">
<para>Specify a DTMF digit that can be used to exit the application while actively
spying on a channel. If there is no channel being spied on, the DTMF digit will be
ignored.</para>
</argument>
</option>
<option name="X">
<para>Allow the user to exit ChanSpy to a valid single digit
numeric extension in the current context or the context
specified by the <variable>SPY_EXIT_CONTEXT</variable> channel variable. The
name of the last channel that was spied on will be stored
in the <variable>SPY_CHANNEL</variable> variable.</para>
</option>
</optionlist>
</parameter>
</syntax>
<description>
<para>This application is used to listen to the audio from an Asterisk channel. This includes the audio
coming in and out of the channel being spied on. If the <literal>chanprefix</literal> parameter is specified,
only channels beginning with this string will be spied upon.</para>
<para>While spying, the following actions may be performed:</para>
<para> - Dialing <literal>#</literal> cycles the volume level.</para>
<para> - Dialing <literal>*</literal> will stop spying and look for another channel to spy on.</para>
<para> - Dialing a series of digits followed by <literal>#</literal> builds a channel name to append
to <literal>chanprefix</literal>. For example, executing ChanSpy(Agent) and then dialing the digits '1234#'
while spying will begin spying on the channel 'Agent/1234'. Note that this feature will be overridden
if the 'd' or 'u' options are used.</para>
<note><para>The <replaceable>X</replaceable> option supersedes the three features above in that if a valid
single digit extension exists in the correct context ChanSpy will exit to it.
This also disables choosing a channel based on <literal>chanprefix</literal> and a digit sequence.</para></note>
</description>
<see-also>
<ref type="application">ExtenSpy</ref>
<ref type="managerEvent">ChanSpyStart</ref>
<ref type="managerEvent">ChanSpyStop</ref>
</see-also>
</application>
<application name="ExtenSpy" language="en_US">
<synopsis>
Listen to a channel, and optionally whisper into it.
</synopsis>
<syntax>
<parameter name="exten" required="true" argsep="@">
<argument name="exten" required="true">
<para>Specify extension.</para>
</argument>
<argument name="context">
<para>Optionally specify a context, defaults to <literal>default</literal>.</para>
</argument>
</parameter>
<parameter name="options">
<optionlist>
<option name="b">
<para>Only spy on channels involved in a bridged call.</para>
</option>
<option name="B">
<para>Instead of whispering on a single channel barge in on both
channels involved in the call.</para>
</option>
<option name="c">
<argument name="digit" required="true">
<para>Specify a DTMF digit that can be used to spy on the next available channel.</para>
</argument>
</option>
<option name="d">
<para>Override the typical numeric DTMF functionality and instead
use DTMF to switch between spy modes.</para>
<enumlist>
<enum name="4">
<para>spy mode</para>
</enum>
<enum name="5">
<para>whisper mode</para>
</enum>
<enum name="6">
<para>barge mode</para>
</enum>
</enumlist>
</option>
<option name="e">
<argument name="ext" required="true" />
<para>Enable <emphasis>enforced</emphasis> mode, so the spying channel can
only monitor extensions whose name is in the <replaceable>ext</replaceable> : delimited
list.</para>
</option>
<option name="E">
<para>Exit when the spied-on channel hangs up.</para>
</option>
<option name="g">
<argument name="grp" required="true">
<para>Only spy on channels in which one or more of the groups
listed in <replaceable>grp</replaceable> matches one or more groups from the
<variable>SPYGROUP</variable> variable set on the channel to be spied upon.</para>
</argument>
<note><para>both <replaceable>grp</replaceable> and <variable>SPYGROUP</variable> can contain
either a single group or a colon-delimited list of groups, such
as <literal>sales:support:accounting</literal>.</para></note>
</option>
<option name="n" argsep="@">
<para>Say the name of the person being spied on if that person has recorded
his/her name. If a context is specified, then that voicemail context will
be searched when retrieving the name, otherwise the <literal>default</literal> context
be used when searching for the name (i.e. if SIP/1000 is the channel being
spied on and no mailbox is specified, then <literal>1000</literal> will be used when searching
for the name).</para>
<argument name="mailbox" />
<argument name="context" />
</option>
<option name="o">
<para>Only listen to audio coming from this channel.</para>
</option>
<option name="q">
<para>Don't play a beep when beginning to spy on a channel, or speak the
selected channel name.</para>
</option>
<option name="r">
<para>Record the session to the monitor spool directory. An optional base for the filename
may be specified. The default is <literal>chanspy</literal>.</para>
<argument name="basename" />
</option>
<option name="s">
<para>Skip the playback of the channel type (i.e. SIP, IAX, etc) when
speaking the selected channel name.</para>
</option>
<option name="S">
<para>Stop when there are no more extensions left to spy on.</para>
</option>
<option name="v">
<argument name="value" />
<para>Adjust the initial volume in the range from <literal>-4</literal>
to <literal>4</literal>. A negative value refers to a quieter setting.</para>
</option>
<option name="w">
<para>Enable <literal>whisper</literal> mode, so the spying channel can talk to
the spied-on channel.</para>
</option>
<option name="W">
<para>Enable <literal>private whisper</literal> mode, so the spying channel can
talk to the spied-on channel but cannot listen to that channel.</para>
</option>
<option name="x">
<argument name="digit" required="true">
<para>Specify a DTMF digit that can be used to exit the application while actively
spying on a channel. If there is no channel being spied on, the DTMF digit will be
ignored.</para>
</argument>
</option>
<option name="X">
<para>Allow the user to exit ChanSpy to a valid single digit
numeric extension in the current context or the context
specified by the <variable>SPY_EXIT_CONTEXT</variable> channel variable. The
name of the last channel that was spied on will be stored
in the <variable>SPY_CHANNEL</variable> variable.</para>
</option>
</optionlist>
</parameter>
</syntax>
<description>
<para>This application is used to listen to the audio from an Asterisk channel. This includes
the audio coming in and out of the channel being spied on. Only channels created by outgoing calls for the
specified extension will be selected for spying. If the optional context is not supplied,
the current channel's context will be used.</para>
<para>While spying, the following actions may be performed:</para>
<para> - Dialing <literal>#</literal> cycles the volume level.</para>
<para> - Dialing <literal>*</literal> will stop spying and look for another channel to spy on.</para>
<note><para>The <replaceable>X</replaceable> option supersedes the three features above in that if a valid
single digit extension exists in the correct context ChanSpy will exit to it.
This also disables choosing a channel based on <literal>chanprefix</literal> and a digit sequence.</para></note>
</description>
<see-also>
<ref type="application">ChanSpy</ref>
<ref type="managerEvent">ChanSpyStart</ref>
<ref type="managerEvent">ChanSpyStop</ref>
</see-also>
</application>
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
<application name="DAHDIScan" language="en_US">
<synopsis>
Scan DAHDI channels to monitor calls.
</synopsis>
<syntax>
<parameter name="group">
<para>Limit scanning to a channel <replaceable>group</replaceable> by setting this option.</para>
</parameter>
</syntax>
<description>
<para>Allows a call center manager to monitor DAHDI channels in a
convenient way. Use <literal>#</literal> to select the next channel and use <literal>*</literal> to exit.</para>
</description>
<see-also>
<ref type="managerEvent">ChanSpyStart</ref>
<ref type="managerEvent">ChanSpyStop</ref>
</see-also>
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
</application>
***/
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
static const char app_chan[] = "ChanSpy";
static const char app_ext[] = "ExtenSpy";
static const char app_dahdiscan[] = "DAHDIScan";
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
enum {
OPTION_QUIET = (1 << 0), /* Quiet, no announcement */
OPTION_BRIDGED = (1 << 1), /* Only look at bridged calls */
OPTION_VOLUME = (1 << 2), /* Specify initial volume */
OPTION_GROUP = (1 << 3), /* Only look at channels in group */
OPTION_RECORD = (1 << 4),
OPTION_WHISPER = (1 << 5),
OPTION_PRIVATE = (1 << 6), /* Private Whisper mode */
OPTION_READONLY = (1 << 7), /* Don't mix the two channels */
OPTION_EXIT = (1 << 8), /* Exit to a valid single digit extension */
OPTION_ENFORCED = (1 << 9), /* Enforced mode */
OPTION_NOTECH = (1 << 10), /* Skip technology name playback */
OPTION_BARGE = (1 << 11), /* Barge mode (whisper to both channels) */
OPTION_NAME = (1 << 12), /* Say the name of the person on whom we will spy */
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
OPTION_DTMF_SWITCH_MODES = (1 << 13), /* Allow numeric DTMF to switch between chanspy modes */
OPTION_DTMF_EXIT = (1 << 14), /* Set DTMF to exit, added for DAHDIScan integration */
OPTION_DTMF_CYCLE = (1 << 15), /* Custom DTMF for cycling next available channel, (default is '*') */
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
OPTION_DAHDI_SCAN = (1 << 16), /* Scan groups in DAHDIScan mode */
OPTION_STOP = (1 << 17),
OPTION_EXITONHANGUP = (1 << 18), /* Hang up when the spied-on channel hangs up. */
OPTION_UNIQUEID = (1 << 19), /* The chanprefix is a channel uniqueid or fully specified channel name. */
};
enum {
OPT_ARG_VOLUME = 0,
OPT_ARG_GROUP,
OPT_ARG_RECORD,
OPT_ARG_ENFORCED,
OPT_ARG_NAME,
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
OPT_ARG_EXIT,
OPT_ARG_CYCLE,
OPT_ARG_ARRAY_SIZE,
};
AST_APP_OPTIONS(spy_opts, {
AST_APP_OPTION('b', OPTION_BRIDGED),
AST_APP_OPTION('B', OPTION_BARGE),
AST_APP_OPTION_ARG('c', OPTION_DTMF_CYCLE, OPT_ARG_CYCLE),
AST_APP_OPTION('d', OPTION_DTMF_SWITCH_MODES),
AST_APP_OPTION_ARG('e', OPTION_ENFORCED, OPT_ARG_ENFORCED),
AST_APP_OPTION('E', OPTION_EXITONHANGUP),
AST_APP_OPTION_ARG('g', OPTION_GROUP, OPT_ARG_GROUP),
AST_APP_OPTION_ARG('n', OPTION_NAME, OPT_ARG_NAME),
AST_APP_OPTION('o', OPTION_READONLY),
AST_APP_OPTION('q', OPTION_QUIET),
AST_APP_OPTION_ARG('r', OPTION_RECORD, OPT_ARG_RECORD),
AST_APP_OPTION('s', OPTION_NOTECH),
AST_APP_OPTION('S', OPTION_STOP),
AST_APP_OPTION('u', OPTION_UNIQUEID),
AST_APP_OPTION_ARG('v', OPTION_VOLUME, OPT_ARG_VOLUME),
AST_APP_OPTION('w', OPTION_WHISPER),
AST_APP_OPTION('W', OPTION_PRIVATE),
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
AST_APP_OPTION_ARG('x', OPTION_DTMF_EXIT, OPT_ARG_EXIT),
AST_APP_OPTION('X', OPTION_EXIT),
});
struct chanspy_translation_helper {
/* spy data */
struct ast_audiohook spy_audiohook;
struct ast_audiohook whisper_audiohook;
struct ast_audiohook bridge_whisper_audiohook;
int fd;
int volfactor;
struct ast_flags flags;
};
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
struct spy_dtmf_options {
char exit;
char cycle;
char volume;
};
static void *spy_alloc(struct ast_channel *chan, void *data)
{
/* just store the data pointer in the channel structure */
return data;
}
static void spy_release(struct ast_channel *chan, void *data)
{
/* nothing to do */
}
static int spy_generate(struct ast_channel *chan, void *data, int len, int samples)
{
struct chanspy_translation_helper *csth = data;
struct ast_frame *f, *cur;
struct ast_format format_slin;
ast_format_set(&format_slin, AST_FORMAT_SLINEAR, 0);
ast_audiohook_lock(&csth->spy_audiohook);
if (csth->spy_audiohook.status != AST_AUDIOHOOK_STATUS_RUNNING) {
/* Channel is already gone more than likely */
ast_audiohook_unlock(&csth->spy_audiohook);
return -1;
}
if (ast_test_flag(&csth->flags, OPTION_READONLY)) {
/* Option 'o' was set, so don't mix channel audio */
f = ast_audiohook_read_frame(&csth->spy_audiohook, samples, AST_AUDIOHOOK_DIRECTION_READ, &format_slin);
} else {
f = ast_audiohook_read_frame(&csth->spy_audiohook, samples, AST_AUDIOHOOK_DIRECTION_BOTH, &format_slin);
}
ast_audiohook_unlock(&csth->spy_audiohook);
if (!f)
return 0;
for (cur = f; cur; cur = AST_LIST_NEXT(cur, frame_list)) {
if (ast_write(chan, cur)) {
ast_frfree(f);
return -1;
}
if (csth->fd) {
if (write(csth->fd, cur->data.ptr, cur->datalen) < 0) {
ast_log(LOG_WARNING, "write() failed: %s\n", strerror(errno));
}
}
}
ast_frfree(f);
return 0;
}
static struct ast_generator spygen = {
.alloc = spy_alloc,
.release = spy_release,
.generate = spy_generate,
};
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
static int start_spying(struct ast_autochan *autochan, const char *spychan_name, struct ast_audiohook *audiohook)
{
int res = 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_NOTICE, "Attaching %s to %s\n", spychan_name, ast_channel_name(autochan->chan));
ast_set_flag(audiohook, AST_AUDIOHOOK_TRIGGER_SYNC | AST_AUDIOHOOK_SMALL_QUEUE);
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
res = ast_audiohook_attach(autochan->chan, audiohook);
if (!res) {
ast_channel_lock(autochan->chan);
if (ast_channel_is_bridged(autochan->chan)) {
ast_softhangup_nolock(autochan->chan, AST_SOFTHANGUP_UNBRIDGE);
}
ast_channel_unlock(autochan->chan);
}
return res;
}
static void change_spy_mode(const char digit, struct ast_flags *flags)
{
if (digit == '4') {
ast_clear_flag(flags, OPTION_WHISPER);
ast_clear_flag(flags, OPTION_BARGE);
} else if (digit == '5') {
ast_clear_flag(flags, OPTION_BARGE);
ast_set_flag(flags, OPTION_WHISPER);
} else if (digit == '6') {
ast_clear_flag(flags, OPTION_WHISPER);
ast_set_flag(flags, OPTION_BARGE);
}
}
static int pack_channel_into_message(struct ast_channel *chan, const char *role,
struct ast_multi_channel_blob *payload)
{
RAII_VAR(struct ast_channel_snapshot *, snapshot,
ast_channel_snapshot_get_latest(ast_channel_uniqueid(chan)),
ao2_cleanup);
if (!snapshot) {
return -1;
}
ast_multi_channel_blob_add_channel(payload, role, snapshot);
return 0;
}
/*! \internal
* \brief Publish the chanspy message over Stasis-Core
* \param spyer The channel doing the spying
* \param spyee Who is being spied upon
* \start start If non-zero, the spying is starting. Otherwise, the spyer is
* finishing
*/
static void publish_chanspy_message(struct ast_channel *spyer,
struct ast_channel *spyee,
int start)
{
RAII_VAR(struct ast_json *, blob, NULL, ast_json_unref);
RAII_VAR(struct ast_multi_channel_blob *, payload, NULL, ao2_cleanup);
RAII_VAR(struct stasis_message *, message, NULL, ao2_cleanup);
if (!spyer) {
ast_log(AST_LOG_WARNING, "Attempt to publish ChanSpy message for NULL spyer channel\n");
return;
}
blob = ast_json_null();
if (!blob) {
return;
}
payload = ast_multi_channel_blob_create(blob);
if (!payload) {
return;
}
if (pack_channel_into_message(spyer, "spyer_channel", payload)) {
return;
}
if (spyee) {
if (pack_channel_into_message(spyee, "spyee_channel", payload)) {
return;
}
}
message = stasis_message_create(
start ? ast_channel_chanspy_start_type(): ast_channel_chanspy_stop_type(),
payload);
if (!message) {
return;
}
stasis_publish(ast_channel_topic(spyer), message);
}
static int attach_barge(struct ast_autochan *spyee_autochan,
struct ast_autochan **spyee_bridge_autochan, struct ast_audiohook *bridge_whisper_audiohook,
const char *spyer_name, const char *name)
{
int retval = 0;
struct ast_autochan *internal_bridge_autochan;
RAII_VAR(struct ast_channel *, bridged, ast_channel_bridge_peer(spyee_autochan->chan), ast_channel_cleanup);
if (!bridged) {
return -1;
}
ast_audiohook_init(bridge_whisper_audiohook, AST_AUDIOHOOK_TYPE_WHISPER, "Chanspy", 0);
internal_bridge_autochan = ast_autochan_setup(bridged);
if (!internal_bridge_autochan) {
return -1;
}
ast_channel_lock(internal_bridge_autochan->chan);
if (start_spying(internal_bridge_autochan, spyer_name, bridge_whisper_audiohook)) {
ast_log(LOG_WARNING, "Unable to attach barge audiohook on spyee '%s'. Barge mode disabled.\n", name);
retval = -1;
}
ast_channel_unlock(internal_bridge_autochan->chan);
*spyee_bridge_autochan = internal_bridge_autochan;
return retval;
}
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
static int channel_spy(struct ast_channel *chan, struct ast_autochan *spyee_autochan,
int *volfactor, int fd, struct spy_dtmf_options *user_options, struct ast_flags *flags,
char *exitcontext)
{
struct chanspy_translation_helper csth;
int running = 0, bridge_connected = 0, res, x = 0;
char inp[24] = {0};
char *name;
struct ast_frame *f;
struct ast_silence_generator *silgen = NULL;
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
struct ast_autochan *spyee_bridge_autochan = NULL;
const char *spyer_name;
if (ast_check_hangup(chan) || ast_check_hangup(spyee_autochan->chan) ||
ast_test_flag(ast_channel_flags(spyee_autochan->chan), AST_FLAG_ZOMBIE)) {
return 0;
}
ast_channel_lock(chan);
spyer_name = ast_strdupa(ast_channel_name(chan));
ast_channel_unlock(chan);
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
ast_channel_lock(spyee_autochan->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
name = ast_strdupa(ast_channel_name(spyee_autochan->chan));
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
ast_channel_unlock(spyee_autochan->chan);
ast_verb(2, "Spying on channel %s\n", name);
publish_chanspy_message(chan, spyee_autochan->chan, 1);
memset(&csth, 0, sizeof(csth));
ast_copy_flags(&csth.flags, flags, AST_FLAGS_ALL);
/* This is the audiohook which gives us the audio off the channel we are
spying on.
*/
ast_audiohook_init(&csth.spy_audiohook, AST_AUDIOHOOK_TYPE_SPY, "ChanSpy", 0);
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
if (start_spying(spyee_autochan, spyer_name, &csth.spy_audiohook)) {
ast_audiohook_destroy(&csth.spy_audiohook);
return 0;
}
if (ast_test_flag(flags, OPTION_WHISPER | OPTION_BARGE | OPTION_DTMF_SWITCH_MODES)) {
/* This audiohook will let us inject audio from our channel into the
channel we are currently spying on.
*/
ast_audiohook_init(&csth.whisper_audiohook, AST_AUDIOHOOK_TYPE_WHISPER, "ChanSpy", 0);
if (start_spying(spyee_autochan, spyer_name, &csth.whisper_audiohook)) {
ast_log(LOG_WARNING, "Unable to attach whisper audiohook to spyee %s. Whisper mode disabled!\n", name);
}
}
ast_channel_lock(chan);
ast_set_flag(ast_channel_flags(chan), AST_FLAG_END_DTMF_ONLY);
ast_channel_unlock(chan);
csth.volfactor = *volfactor;
if (csth.volfactor) {
csth.spy_audiohook.options.read_volume = csth.volfactor;
csth.spy_audiohook.options.write_volume = csth.volfactor;
}
csth.fd = fd;
if (ast_test_flag(flags, OPTION_PRIVATE))
silgen = ast_channel_start_silence_generator(chan);
else
ast_activate_generator(chan, &spygen, &csth);
/* We can no longer rely on 'spyee' being an actual channel;
it can be hung up and freed out from under us. However, the
channel destructor will put NULL into our csth.spy.chan
field when that happens, so that is our signal that the spyee
channel has gone away.
*/
/* Note: it is very important that the ast_waitfor() be the first
condition in this expression, so that if we wait for some period
of time before receiving a frame from our spying channel, we check
for hangup on the spied-on channel _after_ knowing that a frame
has arrived, since the spied-on channel could have gone away while
we were waiting
*/
while (ast_waitfor(chan, -1) > -1 && csth.spy_audiohook.status == AST_AUDIOHOOK_STATUS_RUNNING) {
if (!(f = ast_read(chan)) || ast_check_hangup(chan)) {
running = -1;
break;
}
if (ast_test_flag(flags, OPTION_BARGE) && f->frametype == AST_FRAME_VOICE) {
/* This hook lets us inject audio into the channel that the spyee is currently
* bridged with. If the spyee isn't bridged with anything yet, nothing will
* be attached and we'll need to continue attempting to attach the barge
* audio hook. */
if (!bridge_connected && attach_barge(spyee_autochan, &spyee_bridge_autochan,
&csth.bridge_whisper_audiohook, spyer_name, name) == 0) {
bridge_connected = 1;
}
ast_audiohook_lock(&csth.whisper_audiohook);
ast_audiohook_write_frame(&csth.whisper_audiohook, AST_AUDIOHOOK_DIRECTION_WRITE, f);
ast_audiohook_unlock(&csth.whisper_audiohook);
if (bridge_connected) {
ast_audiohook_lock(&csth.bridge_whisper_audiohook);
ast_audiohook_write_frame(&csth.bridge_whisper_audiohook, AST_AUDIOHOOK_DIRECTION_WRITE, f);
ast_audiohook_unlock(&csth.bridge_whisper_audiohook);
}
ast_frfree(f);
continue;
} else if (ast_test_flag(flags, OPTION_WHISPER) && f->frametype == AST_FRAME_VOICE) {
ast_audiohook_lock(&csth.whisper_audiohook);
ast_audiohook_write_frame(&csth.whisper_audiohook, AST_AUDIOHOOK_DIRECTION_WRITE, f);
ast_audiohook_unlock(&csth.whisper_audiohook);
ast_frfree(f);
continue;
}
res = (f->frametype == AST_FRAME_DTMF) ? f->subclass.integer : 0;
ast_frfree(f);
if (!res)
continue;
if (x == sizeof(inp))
x = 0;
if (res < 0) {
running = -1;
break;
}
if (ast_test_flag(flags, OPTION_EXIT)) {
char tmp[2];
tmp[0] = res;
tmp[1] = '\0';
if (!ast_goto_if_exists(chan, exitcontext, tmp, 1)) {
ast_debug(1, "Got DTMF %c, goto context %s\n", tmp[0], exitcontext);
pbx_builtin_setvar_helper(chan, "SPY_CHANNEL", name);
running = -2;
break;
} else {
ast_debug(2, "Exit by single digit did not work in chanspy. Extension %s does not exist in context %s\n", tmp, exitcontext);
}
} else if (res >= '0' && res <= '9') {
if (ast_test_flag(flags, OPTION_DTMF_SWITCH_MODES)) {
change_spy_mode(res, flags);
} else {
inp[x++] = res;
}
}
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
if (res == user_options->cycle) {
running = 0;
break;
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
} else if (res == user_options->exit) {
running = -2;
break;
} else if (res == user_options->volume) {
if (!ast_strlen_zero(inp)) {
running = atoi(inp);
break;
}
(*volfactor)++;
if (*volfactor > 4)
*volfactor = -4;
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, "Setting spy volume on %s to %d\n", ast_channel_name(chan), *volfactor);
csth.volfactor = *volfactor;
csth.spy_audiohook.options.read_volume = csth.volfactor;
csth.spy_audiohook.options.write_volume = csth.volfactor;
}
}
if (ast_test_flag(flags, OPTION_PRIVATE))
ast_channel_stop_silence_generator(chan, silgen);
else
ast_deactivate_generator(chan);
ast_channel_lock(chan);
ast_clear_flag(ast_channel_flags(chan), AST_FLAG_END_DTMF_ONLY);
ast_channel_unlock(chan);
if (ast_test_flag(flags, OPTION_WHISPER | OPTION_BARGE | OPTION_DTMF_SWITCH_MODES)) {
ast_audiohook_lock(&csth.whisper_audiohook);
ast_audiohook_detach(&csth.whisper_audiohook);
ast_audiohook_unlock(&csth.whisper_audiohook);
ast_audiohook_destroy(&csth.whisper_audiohook);
}
if (ast_test_flag(flags, OPTION_BARGE | OPTION_DTMF_SWITCH_MODES)) {
ast_audiohook_lock(&csth.bridge_whisper_audiohook);
ast_audiohook_detach(&csth.bridge_whisper_audiohook);
ast_audiohook_unlock(&csth.bridge_whisper_audiohook);
ast_audiohook_destroy(&csth.bridge_whisper_audiohook);
}
ast_audiohook_lock(&csth.spy_audiohook);
ast_audiohook_detach(&csth.spy_audiohook);
ast_audiohook_unlock(&csth.spy_audiohook);
ast_audiohook_destroy(&csth.spy_audiohook);
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
if (spyee_bridge_autochan) {
ast_autochan_destroy(spyee_bridge_autochan);
}
ast_verb(2, "Done Spying on channel %s\n", name);
publish_chanspy_message(chan, NULL, 0);
return running;
}
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
static struct ast_autochan *next_channel(struct ast_channel_iterator *iter,
struct ast_autochan *autochan, struct ast_channel *chan)
{
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
struct ast_channel *next;
struct ast_autochan *autochan_store;
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
const size_t pseudo_len = strlen("DAHDI/pseudo");
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
if (!iter) {
return NULL;
}
for (; (next = ast_channel_iterator_next(iter)); ast_channel_unref(next)) {
if (!strncmp(ast_channel_name(next), "DAHDI/pseudo", pseudo_len)
|| next == chan) {
continue;
}
autochan_store = ast_autochan_setup(next);
ast_channel_unref(next);
return autochan_store;
}
return NULL;
}
static int common_exec(struct ast_channel *chan, struct ast_flags *flags,
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
int volfactor, const int fd, struct spy_dtmf_options *user_options,
const char *mygroup, const char *myenforced, const char *spec, const char *exten,
const char *context, const char *mailbox, const char *name_context)
{
char nameprefix[AST_NAME_STRLEN];
char exitcontext[AST_MAX_CONTEXT] = "";
signed char zero_volume = 0;
int waitms;
int res;
int num_spyed_upon = 1;
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
struct ast_channel_iterator *iter = NULL;
if (ast_test_flag(flags, OPTION_EXIT)) {
const char *c;
ast_channel_lock(chan);
if ((c = pbx_builtin_getvar_helper(chan, "SPY_EXIT_CONTEXT"))) {
ast_copy_string(exitcontext, c, sizeof(exitcontext));
} else if (!ast_strlen_zero(ast_channel_macrocontext(chan))) {
ast_copy_string(exitcontext, ast_channel_macrocontext(chan), sizeof(exitcontext));
} else {
ast_copy_string(exitcontext, ast_channel_context(chan), sizeof(exitcontext));
}
ast_channel_unlock(chan);
}
if (ast_channel_state(chan) != AST_STATE_UP)
ast_answer(chan);
ast_set_flag(ast_channel_flags(chan), AST_FLAG_SPYING); /* so nobody can spy on us while we are spying */
waitms = 100;
for (;;) {
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
struct ast_autochan *autochan = NULL, *next_autochan = NULL;
struct ast_channel *prev = NULL;
if (!ast_test_flag(flags, OPTION_QUIET) && num_spyed_upon) {
res = ast_streamfile(chan, "beep", ast_channel_language(chan));
if (!res)
res = ast_waitstream(chan, "");
else if (res < 0) {
ast_clear_flag(ast_channel_flags(chan), AST_FLAG_SPYING);
break;
}
if (!ast_strlen_zero(exitcontext)) {
char tmp[2];
tmp[0] = res;
tmp[1] = '\0';
if (!ast_goto_if_exists(chan, exitcontext, tmp, 1))
goto exit;
else
ast_debug(2, "Exit by single digit did not work in chanspy. Extension %s does not exist in context %s\n", tmp, exitcontext);
}
}
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
/* Set up the iterator we'll be using during this call */
if (!ast_strlen_zero(spec)) {
if (ast_test_flag(flags, OPTION_UNIQUEID)) {
struct ast_channel *unique_chan;
unique_chan = ast_channel_get_by_name(spec);
if (!unique_chan) {
res = -1;
goto exit;
}
iter = ast_channel_iterator_by_name_new(ast_channel_name(unique_chan), 0);
ast_channel_unref(unique_chan);
} else {
iter = ast_channel_iterator_by_name_new(spec, strlen(spec));
}
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
} else if (!ast_strlen_zero(exten)) {
iter = ast_channel_iterator_by_exten_new(exten, context);
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
} else {
iter = ast_channel_iterator_all_new();
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
}
if (!iter) {
res = -1;
goto exit;
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
}
res = ast_waitfordigit(chan, waitms);
if (res < 0) {
iter = ast_channel_iterator_destroy(iter);
ast_clear_flag(ast_channel_flags(chan), AST_FLAG_SPYING);
break;
}
if (!ast_strlen_zero(exitcontext)) {
char tmp[2];
tmp[0] = res;
tmp[1] = '\0';
if (!ast_goto_if_exists(chan, exitcontext, tmp, 1)) {
iter = ast_channel_iterator_destroy(iter);
goto exit;
} else {
ast_debug(2, "Exit by single digit did not work in chanspy. Extension %s does not exist in context %s\n", tmp, exitcontext);
}
}
/* reset for the next loop around, unless overridden later */
waitms = 100;
num_spyed_upon = 0;
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
for (autochan = next_channel(iter, autochan, chan);
autochan;
prev = autochan->chan, ast_autochan_destroy(autochan),
autochan = next_autochan ? next_autochan :
next_channel(iter, autochan, chan), next_autochan = NULL) {
int igrp = !mygroup;
int ienf = !myenforced;
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
if (autochan->chan == prev) {
ast_autochan_destroy(autochan);
break;
}
if (ast_check_hangup(chan)) {
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
ast_autochan_destroy(autochan);
break;
}
if (ast_test_flag(flags, OPTION_BRIDGED) && !ast_channel_is_bridged(autochan->chan)) {
continue;
}
if (ast_check_hangup(autochan->chan) || ast_test_flag(ast_channel_flags(autochan->chan), AST_FLAG_SPYING)) {
continue;
}
if (mygroup) {
int num_groups = 0;
int num_mygroups = 0;
char dup_group[512];
char dup_mygroup[512];
char *groups[NUM_SPYGROUPS];
char *mygroups[NUM_SPYGROUPS];
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
const char *group = NULL;
int x;
int y;
ast_copy_string(dup_mygroup, mygroup, sizeof(dup_mygroup));
num_mygroups = ast_app_separate_args(dup_mygroup, ':', mygroups,
ARRAY_LEN(mygroups));
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
/* Before dahdi scan was part of chanspy, it would use the "GROUP" variable
* rather than "SPYGROUP", this check is done to preserve expected behavior */
if (ast_test_flag(flags, OPTION_DAHDI_SCAN)) {
group = pbx_builtin_getvar_helper(autochan->chan, "GROUP");
} else {
group = pbx_builtin_getvar_helper(autochan->chan, "SPYGROUP");
}
if (!ast_strlen_zero(group)) {
ast_copy_string(dup_group, group, sizeof(dup_group));
num_groups = ast_app_separate_args(dup_group, ':', groups,
ARRAY_LEN(groups));
}
for (y = 0; y < num_mygroups; y++) {
for (x = 0; x < num_groups; x++) {
if (!strcmp(mygroups[y], groups[x])) {
igrp = 1;
break;
}
}
}
}
if (!igrp) {
continue;
}
if (myenforced) {
char ext[AST_CHANNEL_NAME + 3];
char buffer[512];
char *end;
snprintf(buffer, sizeof(buffer) - 1, ":%s:", myenforced);
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_copy_string(ext + 1, ast_channel_name(autochan->chan), sizeof(ext) - 1);
if ((end = strchr(ext, '-'))) {
*end++ = ':';
*end = '\0';
}
ext[0] = ':';
if (strcasestr(buffer, ext)) {
ienf = 1;
}
}
if (!ienf) {
continue;
}
if (!ast_test_flag(flags, OPTION_QUIET)) {
char peer_name[AST_NAME_STRLEN + 5];
char *ptr, *s;
strcpy(peer_name, "spy-");
strncat(peer_name, ast_channel_name(autochan->chan), AST_NAME_STRLEN - 4 - 1);
if ((ptr = strchr(peer_name, '/'))) {
*ptr++ = '\0';
for (s = peer_name; s < ptr; s++) {
*s = tolower(*s);
}
if ((s = strchr(ptr, '-'))) {
*s = '\0';
}
}
if (ast_test_flag(flags, OPTION_NAME)) {
const char *local_context = S_OR(name_context, "default");
const char *local_mailbox = S_OR(mailbox, ptr);
if (local_mailbox) {
res = ast_app_sayname(chan, local_mailbox, local_context);
} else {
res = -1;
}
}
if (!ast_test_flag(flags, OPTION_NAME) || res < 0) {
int num;
if (!ast_test_flag(flags, OPTION_NOTECH)) {
if (ast_fileexists(peer_name, NULL, NULL) > 0) {
res = ast_streamfile(chan, peer_name, ast_channel_language(chan));
if (!res) {
res = ast_waitstream(chan, "");
}
if (res) {
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
ast_autochan_destroy(autochan);
break;
}
} else {
res = ast_say_character_str(chan, peer_name, "", ast_channel_language(chan), AST_SAY_CASE_NONE);
}
}
if (ptr && (num = atoi(ptr))) {
ast_say_digits(chan, num, "", ast_channel_language(chan));
}
}
}
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
res = channel_spy(chan, autochan, &volfactor, fd, user_options, flags, exitcontext);
num_spyed_upon++;
if (res == -1) {
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
ast_autochan_destroy(autochan);
iter = ast_channel_iterator_destroy(iter);
goto exit;
} else if (res == -2) {
res = 0;
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
ast_autochan_destroy(autochan);
iter = ast_channel_iterator_destroy(iter);
goto exit;
} else if (res > 1 && spec && !ast_test_flag(flags, OPTION_UNIQUEID)) {
struct ast_channel *next;
snprintf(nameprefix, AST_NAME_STRLEN, "%s/%d", spec, res);
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
if ((next = ast_channel_get_by_name_prefix(nameprefix, strlen(nameprefix)))) {
next_autochan = ast_autochan_setup(next);
next = ast_channel_unref(next);
} else {
/* stay on this channel, if it is still valid */
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
if (!ast_check_hangup(autochan->chan)) {
next_autochan = ast_autochan_setup(autochan->chan);
} else {
/* the channel is gone */
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
next_autochan = NULL;
}
}
} else if (res == 0 && ast_test_flag(flags, OPTION_EXITONHANGUP)) {
ast_autochan_destroy(autochan);
iter = ast_channel_iterator_destroy(iter);
goto exit;
}
}
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
iter = ast_channel_iterator_destroy(iter);
if (res == -1 || ast_check_hangup(chan))
break;
if (ast_test_flag(flags, OPTION_STOP) && !next_autochan) {
break;
}
}
exit:
ast_clear_flag(ast_channel_flags(chan), AST_FLAG_SPYING);
ast_channel_setoption(chan, AST_OPTION_TXGAIN, &zero_volume, sizeof(zero_volume), 0);
return res;
}
static int chanspy_exec(struct ast_channel *chan, const char *data)
{
char *myenforced = NULL;
char *mygroup = NULL;
char *recbase = NULL;
int fd = 0;
struct ast_flags flags;
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
struct spy_dtmf_options user_options = {
.cycle = '*',
.volume = '#',
.exit = '\0',
};
struct ast_format oldwf;
int volfactor = 0;
int res;
char *mailbox = NULL;
char *name_context = NULL;
AST_DECLARE_APP_ARGS(args,
AST_APP_ARG(spec);
AST_APP_ARG(options);
);
char *opts[OPT_ARG_ARRAY_SIZE];
char *parse = ast_strdupa(data);
AST_STANDARD_APP_ARGS(args, parse);
ast_format_clear(&oldwf);
if (args.spec && !strcmp(args.spec, "all"))
args.spec = NULL;
if (args.options) {
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
char tmp;
ast_app_parse_options(spy_opts, &flags, opts, args.options);
if (ast_test_flag(&flags, OPTION_GROUP))
mygroup = opts[OPT_ARG_GROUP];
if (ast_test_flag(&flags, OPTION_RECORD) &&
!(recbase = opts[OPT_ARG_RECORD]))
recbase = "chanspy";
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
if (ast_test_flag(&flags, OPTION_DTMF_EXIT) && opts[OPT_ARG_EXIT]) {
tmp = opts[OPT_ARG_EXIT][0];
if (strchr("0123456789*#", tmp) && tmp != '\0') {
user_options.exit = tmp;
} else {
ast_log(LOG_NOTICE, "Argument for option 'x' must be a valid DTMF digit.\n");
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
}
}
if (ast_test_flag(&flags, OPTION_DTMF_CYCLE) && opts[OPT_ARG_CYCLE]) {
tmp = opts[OPT_ARG_CYCLE][0];
if (strchr("0123456789*#", tmp) && tmp != '\0') {
user_options.cycle = tmp;
} else {
ast_log(LOG_NOTICE, "Argument for option 'c' must be a valid DTMF digit.\n");
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
}
}
if (ast_test_flag(&flags, OPTION_VOLUME) && opts[OPT_ARG_VOLUME]) {
int vol;
if ((sscanf(opts[OPT_ARG_VOLUME], "%30d", &vol) != 1) || (vol > 4) || (vol < -4))
ast_log(LOG_NOTICE, "Volume factor must be a number between -4 and 4\n");
else
volfactor = vol;
}
if (ast_test_flag(&flags, OPTION_PRIVATE))
ast_set_flag(&flags, OPTION_WHISPER);
if (ast_test_flag(&flags, OPTION_ENFORCED))
myenforced = opts[OPT_ARG_ENFORCED];
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
if (ast_test_flag(&flags, OPTION_NAME)) {
if (!ast_strlen_zero(opts[OPT_ARG_NAME])) {
char *delimiter;
if ((delimiter = strchr(opts[OPT_ARG_NAME], '@'))) {
mailbox = opts[OPT_ARG_NAME];
*delimiter++ = '\0';
name_context = delimiter;
} else {
mailbox = opts[OPT_ARG_NAME];
}
}
}
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
} else {
ast_clear_flag(&flags, AST_FLAGS_ALL);
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
}
ast_format_copy(&oldwf, ast_channel_writeformat(chan));
if (ast_set_write_format_by_id(chan, AST_FORMAT_SLINEAR) < 0) {
ast_log(LOG_ERROR, "Could Not Set Write Format.\n");
return -1;
}
if (recbase) {
char filename[PATH_MAX];
snprintf(filename, sizeof(filename), "%s/%s.%d.raw", ast_config_AST_MONITOR_DIR, recbase, (int) time(NULL));
if ((fd = open(filename, O_CREAT | O_WRONLY | O_TRUNC, AST_FILE_MODE)) <= 0) {
ast_log(LOG_WARNING, "Cannot open '%s' for recording\n", filename);
fd = 0;
}
}
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
res = common_exec(chan, &flags, volfactor, fd, &user_options, mygroup, myenforced, args.spec, NULL, NULL, mailbox, name_context);
if (fd)
close(fd);
if (oldwf.id && ast_set_write_format(chan, &oldwf) < 0)
ast_log(LOG_ERROR, "Could Not Set Write Format.\n");
if (ast_test_flag(&flags, OPTION_EXITONHANGUP)) {
ast_verb(3, "Stopped spying due to the spied-on channel hanging up.\n");
}
return res;
}
static int extenspy_exec(struct ast_channel *chan, const char *data)
{
char *ptr, *exten = NULL;
char *mygroup = NULL;
char *recbase = NULL;
int fd = 0;
struct ast_flags flags;
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
struct spy_dtmf_options user_options = {
.cycle = '*',
.volume = '#',
.exit = '\0',
};
struct ast_format oldwf;
int volfactor = 0;
int res;
char *mailbox = NULL;
char *name_context = NULL;
AST_DECLARE_APP_ARGS(args,
AST_APP_ARG(context);
AST_APP_ARG(options);
);
char *parse = ast_strdupa(data);
AST_STANDARD_APP_ARGS(args, parse);
ast_format_clear(&oldwf);
if (!ast_strlen_zero(args.context) && (ptr = strchr(args.context, '@'))) {
exten = args.context;
*ptr++ = '\0';
args.context = ptr;
}
if (ast_strlen_zero(args.context))
args.context = ast_strdupa(ast_channel_context(chan));
if (args.options) {
char *opts[OPT_ARG_ARRAY_SIZE];
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
char tmp;
ast_app_parse_options(spy_opts, &flags, opts, args.options);
if (ast_test_flag(&flags, OPTION_GROUP))
mygroup = opts[OPT_ARG_GROUP];
if (ast_test_flag(&flags, OPTION_RECORD) &&
!(recbase = opts[OPT_ARG_RECORD]))
recbase = "chanspy";
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
if (ast_test_flag(&flags, OPTION_DTMF_EXIT) && opts[OPT_ARG_EXIT]) {
tmp = opts[OPT_ARG_EXIT][0];
if (strchr("0123456789*#", tmp) && tmp != '\0') {
user_options.exit = tmp;
} else {
ast_log(LOG_NOTICE, "Argument for option 'x' must be a valid DTMF digit.\n");
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
}
}
if (ast_test_flag(&flags, OPTION_DTMF_CYCLE) && opts[OPT_ARG_CYCLE]) {
tmp = opts[OPT_ARG_CYCLE][0];
if (strchr("0123456789*#", tmp) && tmp != '\0') {
user_options.cycle = tmp;
} else {
ast_log(LOG_NOTICE, "Argument for option 'c' must be a valid DTMF digit.\n");
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
}
}
if (ast_test_flag(&flags, OPTION_VOLUME) && opts[OPT_ARG_VOLUME]) {
int vol;
if ((sscanf(opts[OPT_ARG_VOLUME], "%30d", &vol) != 1) || (vol > 4) || (vol < -4))
ast_log(LOG_NOTICE, "Volume factor must be a number between -4 and 4\n");
else
volfactor = vol;
}
if (ast_test_flag(&flags, OPTION_PRIVATE))
ast_set_flag(&flags, OPTION_WHISPER);
if (ast_test_flag(&flags, OPTION_NAME)) {
if (!ast_strlen_zero(opts[OPT_ARG_NAME])) {
char *delimiter;
if ((delimiter = strchr(opts[OPT_ARG_NAME], '@'))) {
mailbox = opts[OPT_ARG_NAME];
*delimiter++ = '\0';
name_context = delimiter;
} else {
mailbox = opts[OPT_ARG_NAME];
}
}
}
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
} else {
/* Coverity - This uninit_use should be ignored since this macro initializes the flags */
ast_clear_flag(&flags, AST_FLAGS_ALL);
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
}
ast_format_copy(&oldwf, ast_channel_writeformat(chan));
if (ast_set_write_format_by_id(chan, AST_FORMAT_SLINEAR) < 0) {
ast_log(LOG_ERROR, "Could Not Set Write Format.\n");
return -1;
}
if (recbase) {
char filename[PATH_MAX];
snprintf(filename, sizeof(filename), "%s/%s.%d.raw", ast_config_AST_MONITOR_DIR, recbase, (int) time(NULL));
if ((fd = open(filename, O_CREAT | O_WRONLY | O_TRUNC, AST_FILE_MODE)) <= 0) {
ast_log(LOG_WARNING, "Cannot open '%s' for recording\n", filename);
fd = 0;
}
}
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
res = common_exec(chan, &flags, volfactor, fd, &user_options, mygroup, NULL, NULL, exten, args.context, mailbox, name_context);
if (fd)
close(fd);
if (oldwf.id && ast_set_write_format(chan, &oldwf) < 0)
ast_log(LOG_ERROR, "Could Not Set Write Format.\n");
return res;
}
static int dahdiscan_exec(struct ast_channel *chan, const char *data)
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
{
const char *spec = "DAHDI";
struct ast_flags flags;
struct spy_dtmf_options user_options = {
.cycle = '#',
.volume = '\0',
.exit = '*',
};
struct ast_format oldwf;
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
int res;
char *mygroup = NULL;
/* Coverity - This uninit_use should be ignored since this macro initializes the flags */
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
ast_clear_flag(&flags, AST_FLAGS_ALL);
ast_format_clear(&oldwf);
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
if (!ast_strlen_zero(data)) {
mygroup = ast_strdupa(data);
}
ast_set_flag(&flags, OPTION_DTMF_EXIT);
ast_set_flag(&flags, OPTION_DTMF_CYCLE);
ast_set_flag(&flags, OPTION_DAHDI_SCAN);
ast_format_copy(&oldwf, ast_channel_writeformat(chan));
if (ast_set_write_format_by_id(chan, AST_FORMAT_SLINEAR) < 0) {
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
ast_log(LOG_ERROR, "Could Not Set Write Format.\n");
return -1;
}
res = common_exec(chan, &flags, 0, 0, &user_options, mygroup, NULL, spec, NULL, NULL, NULL, NULL);
if (oldwf.id && ast_set_write_format(chan, &oldwf) < 0)
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
ast_log(LOG_ERROR, "Could Not Set Write Format.\n");
return res;
}
static int unload_module(void)
{
int res = 0;
res |= ast_unregister_application(app_chan);
res |= ast_unregister_application(app_ext);
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
res |= ast_unregister_application(app_dahdiscan);
return res;
}
static int load_module(void)
{
int res = 0;
res |= ast_register_application_xml(app_chan, chanspy_exec);
res |= ast_register_application_xml(app_ext, extenspy_exec);
Convert the ast_channel data structure over to the astobj2 framework. There is a lot that could be said about this, but the patch is a big improvement for performance, stability, code maintainability, and ease of future code development. The channel list is no longer an unsorted linked list. The main container for channels is an astobj2 hash table. All of the code related to searching for channels or iterating active channels has been rewritten. Let n be the number of active channels. Iterating the channel list has gone from O(n^2) to O(n). Searching for a channel by name went from O(n) to O(1). Searching for a channel by extension is still O(n), but uses a new method for doing so, which is more efficient. The ast_channel object is now a reference counted object. The benefits here are plentiful. Some benefits directly related to issues in the previous code include: 1) When threads other than the channel thread owning a channel wanted access to a channel, it had to hold the lock on it to ensure that it didn't go away. This is no longer a requirement. Holding a reference is sufficient. 2) There are places that now require less dealing with channel locks. 3) There are places where channel locks are held for much shorter periods of time. 4) There are places where dealing with more than one channel at a time becomes _MUCH_ easier. ChanSpy is a great example of this. Writing code in the future that deals with multiple channels will be much easier. Some additional information regarding channel locking and reference count handling can be found in channel.h, where a new section has been added that discusses some of the rules associated with it. Mark Michelson also assisted with the development of this patch. He did the conversion of ChanSpy and introduced a new API, ast_autochan, which makes it much easier to deal with holding on to a channel pointer for an extended period of time and having it get automatically updated if the channel gets masqueraded. Mark was also a huge help in the code review process. Thanks to David Vossel for his assistance with this branch, as well. David did the conversion of the DAHDIScan application by making it become a wrapper for ChanSpy internally. The changes come from the svn/asterisk/team/russell/ast_channel_ao2 branch. Review: http://reviewboard.digium.com/r/203/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@190423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2009-04-24 14:04:26 +00:00
res |= ast_register_application_xml(app_dahdiscan, dahdiscan_exec);
return res;
}
AST_MODULE_INFO_STANDARD(ASTERISK_GPL_KEY, "Listen to the audio of an active channel");