Summary: | ASTERISK-11940: App Queue extension/context copying | ||
Reporter: | jeffery palmer (darren1713) | Labels: | |
Date Opened: | 2008-04-29 10:15:11 | Date Closed: | 2008-05-01 14:30:20 |
Priority: | Minor | Regression? | No |
Status: | Closed/Complete | Components: | Applications/app_queue |
Versions: | Frequency of Occurrence | ||
Related Issues: | |||
Environment: | Attachments: | ( 0) app_queue.c.macroextenpatch | |
Description: | The queue app copies the incoming context/exten from either its parent macro, or current context, but it behaves strangely. The macro app copies the context/extension to two places, one is on the chan struct (macrocontext/macroexten) and also copies it to the channel vars (MACRO_EXTEN, MACRO_CONTEXT). Why this is, I don't know, but it should be cleaned up in the future because they behave differently. The chan struct only receives the FIRST macro to run, and no successive ones. This means daisy chained macro's don't pass valid information off to app_queue. This patch is to read the correct, and modifiable, channel variables (MACRO_EXTEN, MACRO_CONTEXT) instead of the variables on the chan struct. | ||
Comments: | By: Mark Michelson (mmichelson) 2008-04-29 14:12:12 Good work. I'll get this merged in ASAP. Since this is a small patch and I can clearly see the modification (thanks in part to your good descriptive summary) I will go ahead and get this patch merged in, but for future patches, please upload the diff in unified format. Thanks! By: Digium Subversion (svnbot) 2008-04-29 14:34:34 Repository: asterisk Revision: 114848 U branches/1.4/apps/app_queue.c ------------------------------------------------------------------------ r114848 | mmichelson | 2008-04-29 14:34:31 -0500 (Tue, 29 Apr 2008) | 14 lines Use the MACRO_CONTEXT and MACRO_EXTEN channel variables instead of the channel's macrocontext and macroexten fields. This is needed because if macros are daisy-chained, the incorrect context and extension are placed on the new channel. I also added locking to the channel prior to accessing these variables as noted in trunk's janitor project file. (closes issue ASTERISK-11940) Reported by: darren1713 Patches: app_queue.c.macroextenpatch uploaded by darren1713 (license 116) (with modifications from me) Tested by: putnopvut ------------------------------------------------------------------------ http://svn.digium.com/view/asterisk?view=rev&revision=114848 By: Digium Subversion (svnbot) 2008-04-29 14:36:42 Repository: asterisk Revision: 114849 _U trunk/ U trunk/apps/app_queue.c ------------------------------------------------------------------------ r114849 | mmichelson | 2008-04-29 14:36:40 -0500 (Tue, 29 Apr 2008) | 22 lines Merged revisions 114848 via svnmerge from https://origsvn.digium.com/svn/asterisk/branches/1.4 ........ r114848 | mmichelson | 2008-04-29 14:40:06 -0500 (Tue, 29 Apr 2008) | 14 lines Use the MACRO_CONTEXT and MACRO_EXTEN channel variables instead of the channel's macrocontext and macroexten fields. This is needed because if macros are daisy-chained, the incorrect context and extension are placed on the new channel. I also added locking to the channel prior to accessing these variables as noted in trunk's janitor project file. (closes issue ASTERISK-11940) Reported by: darren1713 Patches: app_queue.c.macroextenpatch uploaded by darren1713 (license 116) (with modifications from me) Tested by: putnopvut ........ ------------------------------------------------------------------------ http://svn.digium.com/view/asterisk?view=rev&revision=114849 By: Digium Subversion (svnbot) 2008-04-29 14:37:24 Repository: asterisk Revision: 114850 _U branches/1.6.0/ U branches/1.6.0/apps/app_queue.c ------------------------------------------------------------------------ r114850 | mmichelson | 2008-04-29 14:37:21 -0500 (Tue, 29 Apr 2008) | 30 lines Merged revisions 114849 via svnmerge from https://origsvn.digium.com/svn/asterisk/trunk ................ r114849 | mmichelson | 2008-04-29 14:42:04 -0500 (Tue, 29 Apr 2008) | 22 lines Merged revisions 114848 via svnmerge from https://origsvn.digium.com/svn/asterisk/branches/1.4 ........ r114848 | mmichelson | 2008-04-29 14:40:06 -0500 (Tue, 29 Apr 2008) | 14 lines Use the MACRO_CONTEXT and MACRO_EXTEN channel variables instead of the channel's macrocontext and macroexten fields. This is needed because if macros are daisy-chained, the incorrect context and extension are placed on the new channel. I also added locking to the channel prior to accessing these variables as noted in trunk's janitor project file. (closes issue ASTERISK-11940) Reported by: darren1713 Patches: app_queue.c.macroextenpatch uploaded by darren1713 (license 116) (with modifications from me) Tested by: putnopvut ........ ................ ------------------------------------------------------------------------ http://svn.digium.com/view/asterisk?view=rev&revision=114850 By: Digium Subversion (svnbot) 2008-04-30 07:19:26 Repository: asterisk Revision: 114878 _U team/mvanbaak/skinny-realtime/ U team/mvanbaak/skinny-realtime/CHANGES U team/mvanbaak/skinny-realtime/Makefile.rules U team/mvanbaak/skinny-realtime/UPGRADE.txt U team/mvanbaak/skinny-realtime/apps/app_chanspy.c U team/mvanbaak/skinny-realtime/apps/app_directory.c U team/mvanbaak/skinny-realtime/apps/app_queue.c U team/mvanbaak/skinny-realtime/apps/app_voicemail.c U team/mvanbaak/skinny-realtime/channels/iax2-provision.c U team/mvanbaak/skinny-realtime/contrib/scripts/get_ilbc_source.sh U team/mvanbaak/skinny-realtime/include/asterisk/app.h U team/mvanbaak/skinny-realtime/main/app.c U team/mvanbaak/skinny-realtime/main/features.c U team/mvanbaak/skinny-realtime/phoneprov/polycom.xml U team/mvanbaak/skinny-realtime/res/res_config_pgsql.c ------------------------------------------------------------------------ r114878 | mvanbaak | 2008-04-30 07:19:24 -0500 (Wed, 30 Apr 2008) | 151 lines Merged revisions 114813,114824,114830,114832,114834,114841,114845,114849,114852,114857,114866,114874,114876 via svnmerge from https://origsvn.digium.com/svn/asterisk/trunk ................ r114813 | mmichelson | 2008-04-29 00:38:07 +0200 (Tue, 29 Apr 2008) | 10 lines Adding a new option 'n' to app_chanspy. This option allows for the name of the spied-on party to be spoken instead of the channel name or number. This was accomplished by adding a new function pointer to point to a function in app_voicemail which retrieves the name file and plays it. This makes for an easy way that applications may play a user's name should it be necessary. app_directory, in particular, can be simplified greatly by this change. This change comes as a suggestion from Switchvox, which already has this feature. AST-23 ................ r114824 | kpfleming | 2008-04-29 14:54:31 +0200 (Tue, 29 Apr 2008) | 18 lines Merged revisions 114823 via svnmerge from https://origsvn.digium.com/svn/asterisk/branches/1.4 ................ r114823 | kpfleming | 2008-04-29 07:53:12 -0500 (Tue, 29 Apr 2008) | 10 lines Merged revisions 114822 via svnmerge from https://origsvn.digium.com/svn/asterisk/branches/1.2 ........ r114822 | kpfleming | 2008-04-29 07:52:32 -0500 (Tue, 29 Apr 2008) | 2 lines stop script from appending source code if run multiple times ........ ................ ................ r114830 | qwell | 2008-04-29 19:10:55 +0200 (Tue, 29 Apr 2008) | 9 lines Merged revisions 114829 via svnmerge from https://origsvn.digium.com/svn/asterisk/branches/1.4 ........ r114829 | qwell | 2008-04-29 12:08:55 -0500 (Tue, 29 Apr 2008) | 1 line Change warning message to debug, since there are cases where 0 results is perfectly fine. ........ ................ r114832 | mmichelson | 2008-04-29 19:31:26 +0200 (Tue, 29 Apr 2008) | 3 lines Fix a crash happening in app_directory. This crash would occur if a users.conf existed. ................ r114834 | mmichelson | 2008-04-29 19:56:13 +0200 (Tue, 29 Apr 2008) | 8 lines Since there is now a globally available function for saying someone's name, a LOT of functions in app_directory can be removed since the ODBC-specific lookups are accomplished within app_voicemail. This change greatly reduces the amount of lines in app_directory that were solely for the purpose of looking up a name when ODBC_STORAGE is specified for voicemail. This commit also makes the name-saying interruptable via DTMF. ................ r114841 | mmichelson | 2008-04-29 20:48:26 +0200 (Tue, 29 Apr 2008) | 4 lines Make app_directory dependent on app_voicemail. This is because the function which says the person's name is handled inside app_voicemail now. ................ r114845 | kpfleming | 2008-04-29 20:58:48 +0200 (Tue, 29 Apr 2008) | 3 lines fix this logic to actually be correct... the fd can't be *both* -1 and an array index to be checked in rfds/efds (bug found by gcc-4.3) ................ r114849 | mmichelson | 2008-04-29 21:42:04 +0200 (Tue, 29 Apr 2008) | 22 lines Merged revisions 114848 via svnmerge from https://origsvn.digium.com/svn/asterisk/branches/1.4 ........ r114848 | mmichelson | 2008-04-29 14:40:06 -0500 (Tue, 29 Apr 2008) | 14 lines Use the MACRO_CONTEXT and MACRO_EXTEN channel variables instead of the channel's macrocontext and macroexten fields. This is needed because if macros are daisy-chained, the incorrect context and extension are placed on the new channel. I also added locking to the channel prior to accessing these variables as noted in trunk's janitor project file. (closes issue ASTERISK-11940) Reported by: darren1713 Patches: app_queue.c.macroextenpatch uploaded by darren1713 (license 116) (with modifications from me) Tested by: putnopvut ........ ................ r114852 | qwell | 2008-04-29 22:05:48 +0200 (Tue, 29 Apr 2008) | 1 line Fix formatting ................ r114857 | mmichelson | 2008-04-29 23:07:36 +0200 (Tue, 29 Apr 2008) | 15 lines Patching app_chanspy to jibe better with what is documented. This allows for a colon-delimited list of spygroups to be specified when calling the ChanSpy application with the 'g' option. Prior to this, you could only specify a single group when using the 'g' option. I also have upped the maximum number of spygroups to 128 and added a #define so that this can be easily increased or decreased later. (closes issue ASTERISK-11894) Reported by: jsmith Patches: app_chanspy_multiple_groups_v2.patch uploaded by jsmith (license 15) Tested by: atis, jvandal ................ r114866 | jpeeler | 2008-04-30 00:54:14 +0200 (Wed, 30 Apr 2008) | 2 lines Fixes a problem where all the templates were marked as dead no matter what. The templates should only be marked as dead if a configuration file has been successfully loaded and has changes. Bug found while making API documentation for 1.6.0. ................ r114874 | tilghman | 2008-04-30 07:05:25 +0200 (Wed, 30 Apr 2008) | 2 lines Document the Incomplete application addition. ................ r114876 | kpfleming | 2008-04-30 14:15:43 +0200 (Wed, 30 Apr 2008) | 10 lines Merged revisions 114875 via svnmerge from https://origsvn.digium.com/svn/asterisk/branches/1.4 ........ r114875 | kpfleming | 2008-04-30 07:14:07 -0500 (Wed, 30 Apr 2008) | 2 lines pay attention to *all* header files for dependency tracking, not just the local ones (inspired by r578 of asterisk-addons by tilghman) ........ ................ ------------------------------------------------------------------------ http://svn.digium.com/view/asterisk?view=rev&revision=114878 By: Digium Subversion (svnbot) 2008-05-01 12:06:48 Repository: asterisk Revision: 114930 _U team/group/taskprocessors/ U team/group/taskprocessors/CHANGES U team/group/taskprocessors/Makefile.rules U team/group/taskprocessors/UPGRADE.txt U team/group/taskprocessors/apps/app_chanspy.c U team/group/taskprocessors/apps/app_directory.c U team/group/taskprocessors/apps/app_externalivr.c U team/group/taskprocessors/apps/app_macro.c U team/group/taskprocessors/apps/app_meetme.c U team/group/taskprocessors/apps/app_minivm.c U team/group/taskprocessors/apps/app_morsecode.c U team/group/taskprocessors/apps/app_queue.c U team/group/taskprocessors/apps/app_speech_utils.c U team/group/taskprocessors/apps/app_stack.c U team/group/taskprocessors/apps/app_voicemail.c U team/group/taskprocessors/apps/app_while.c U team/group/taskprocessors/channels/chan_iax2.c U team/group/taskprocessors/channels/chan_sip.c U team/group/taskprocessors/channels/chan_zap.c U team/group/taskprocessors/channels/iax2-provision.c U team/group/taskprocessors/channels/iax2.h U team/group/taskprocessors/configs/queues.conf.sample U team/group/taskprocessors/configs/sip.conf.sample U team/group/taskprocessors/contrib/scripts/get_ilbc_source.sh A team/group/taskprocessors/funcs/func_speex.c U team/group/taskprocessors/include/asterisk/audiohook.h U team/group/taskprocessors/main/audiohook.c U team/group/taskprocessors/main/features.c U team/group/taskprocessors/phoneprov/polycom.xml U team/group/taskprocessors/res/res_config_pgsql.c ------------------------------------------------------------------------ r114930 | dhubbard | 2008-05-01 12:06:46 -0500 (Thu, 01 May 2008) | 266 lines Merged revisions 114824,114830,114832,114834,114841,114845,114849,114852,114857,114866,114874,114876,114884,114888,114892,114899,114904,114906,114912,114922,114926 via svnmerge from https://origsvn.digium.com/svn/asterisk/trunk ................ r114824 | kpfleming | 2008-04-29 07:54:31 -0500 (Tue, 29 Apr 2008) | 18 lines Merged revisions 114823 via svnmerge from https://origsvn.digium.com/svn/asterisk/branches/1.4 ................ r114823 | kpfleming | 2008-04-29 07:53:12 -0500 (Tue, 29 Apr 2008) | 10 lines Merged revisions 114822 via svnmerge from https://origsvn.digium.com/svn/asterisk/branches/1.2 ........ r114822 | kpfleming | 2008-04-29 07:52:32 -0500 (Tue, 29 Apr 2008) | 2 lines stop script from appending source code if run multiple times ........ ................ ................ r114830 | qwell | 2008-04-29 12:10:55 -0500 (Tue, 29 Apr 2008) | 9 lines Merged revisions 114829 via svnmerge from https://origsvn.digium.com/svn/asterisk/branches/1.4 ........ r114829 | qwell | 2008-04-29 12:08:55 -0500 (Tue, 29 Apr 2008) | 1 line Change warning message to debug, since there are cases where 0 results is perfectly fine. ........ ................ r114832 | mmichelson | 2008-04-29 12:31:26 -0500 (Tue, 29 Apr 2008) | 3 lines Fix a crash happening in app_directory. This crash would occur if a users.conf existed. ................ r114834 | mmichelson | 2008-04-29 12:56:13 -0500 (Tue, 29 Apr 2008) | 8 lines Since there is now a globally available function for saying someone's name, a LOT of functions in app_directory can be removed since the ODBC-specific lookups are accomplished within app_voicemail. This change greatly reduces the amount of lines in app_directory that were solely for the purpose of looking up a name when ODBC_STORAGE is specified for voicemail. This commit also makes the name-saying interruptable via DTMF. ................ r114841 | mmichelson | 2008-04-29 13:48:26 -0500 (Tue, 29 Apr 2008) | 4 lines Make app_directory dependent on app_voicemail. This is because the function which says the person's name is handled inside app_voicemail now. ................ r114845 | kpfleming | 2008-04-29 13:58:48 -0500 (Tue, 29 Apr 2008) | 3 lines fix this logic to actually be correct... the fd can't be *both* -1 and an array index to be checked in rfds/efds (bug found by gcc-4.3) ................ r114849 | mmichelson | 2008-04-29 14:42:04 -0500 (Tue, 29 Apr 2008) | 22 lines Merged revisions 114848 via svnmerge from https://origsvn.digium.com/svn/asterisk/branches/1.4 ........ r114848 | mmichelson | 2008-04-29 14:40:06 -0500 (Tue, 29 Apr 2008) | 14 lines Use the MACRO_CONTEXT and MACRO_EXTEN channel variables instead of the channel's macrocontext and macroexten fields. This is needed because if macros are daisy-chained, the incorrect context and extension are placed on the new channel. I also added locking to the channel prior to accessing these variables as noted in trunk's janitor project file. (closes issue ASTERISK-11940) Reported by: darren1713 Patches: app_queue.c.macroextenpatch uploaded by darren1713 (license 116) (with modifications from me) Tested by: putnopvut ........ ................ r114852 | qwell | 2008-04-29 15:05:48 -0500 (Tue, 29 Apr 2008) | 1 line Fix formatting ................ r114857 | mmichelson | 2008-04-29 16:07:36 -0500 (Tue, 29 Apr 2008) | 15 lines Patching app_chanspy to jibe better with what is documented. This allows for a colon-delimited list of spygroups to be specified when calling the ChanSpy application with the 'g' option. Prior to this, you could only specify a single group when using the 'g' option. I also have upped the maximum number of spygroups to 128 and added a #define so that this can be easily increased or decreased later. (closes issue ASTERISK-11894) Reported by: jsmith Patches: app_chanspy_multiple_groups_v2.patch uploaded by jsmith (license 15) Tested by: atis, jvandal ................ r114866 | jpeeler | 2008-04-29 17:54:14 -0500 (Tue, 29 Apr 2008) | 2 lines Fixes a problem where all the templates were marked as dead no matter what. The templates should only be marked as dead if a configuration file has been successfully loaded and has changes. Bug found while making API documentation for 1.6.0. ................ r114874 | tilghman | 2008-04-30 00:05:25 -0500 (Wed, 30 Apr 2008) | 2 lines Document the Incomplete application addition. ................ r114876 | kpfleming | 2008-04-30 07:15:43 -0500 (Wed, 30 Apr 2008) | 10 lines Merged revisions 114875 via svnmerge from https://origsvn.digium.com/svn/asterisk/branches/1.4 ........ r114875 | kpfleming | 2008-04-30 07:14:07 -0500 (Wed, 30 Apr 2008) | 2 lines pay attention to *all* header files for dependency tracking, not just the local ones (inspired by r578 of asterisk-addons by tilghman) ........ ................ r114884 | kpfleming | 2008-04-30 09:49:51 -0500 (Wed, 30 Apr 2008) | 10 lines Merged revisions 114880 via svnmerge from https://origsvn.digium.com/svn/asterisk/branches/1.4 ........ r114880 | kpfleming | 2008-04-30 09:46:57 -0500 (Wed, 30 Apr 2008) | 2 lines use the ARRAY_LEN macro for indexing through the iaxs/iaxsl arrays so that the size of the arrays can be adjusted in one place, and change the size of the arrays from 32768 calls to 2048 calls when LOW_MEMORY is defined ........ ................ r114888 | jpeeler | 2008-04-30 11:14:43 -0500 (Wed, 30 Apr 2008) | 3 lines Fixes a bug where if a stream monitor thread was not created (caused from failure of opening or starting the stream) pthread_cancel was called with an invalid thread ID. ................ r114892 | russell | 2008-04-30 11:34:24 -0500 (Wed, 30 Apr 2008) | 36 lines Merged revisions 114891 via svnmerge from https://origsvn.digium.com/svn/asterisk/branches/1.4 ........ r114891 | russell | 2008-04-30 11:30:01 -0500 (Wed, 30 Apr 2008) | 28 lines Merge changes from team/russell/iax2_find_callno and iax2_find_callno_1.4 These changes address a critical performance issue introduced in the latest release. The fix for the latest security issue included a change that made Asterisk randomly choose call numbers to make them more difficult to guess by attackers. However, due to some inefficient (this is by far, an understatement) code, when Asterisk chose high call numbers, chan_iax2 became unusable after just a small number of calls. On a small embedded platform, it would not be able to handle a single call. On my Intel Core 2 Duo @ 2.33 GHz, I couldn't run more than about 16 IAX2 channels. Ouch. These changes address some performance issues of the find_callno() function that have bothered me for a very long time. On every incoming media frame, it iterated through every possible call number trying to find a matching active call. This involved a mutex lock and unlock for each call number checked. So, if the random call number chosen was 20000, then every media frame would cause 20000 locks and unlocks. Previously, this problem was not as obvious since Asterisk always chose the lowest call number it could. A second container for IAX2 pvt structs has been added. It is an astobj2 hash table. When we know the remote side's call number, the pvt goes into the hash table with a hash value of the remote side's call number. Then, lookups for incoming media frames are a very fast hash lookup instead of an absolutely insane array traversal. In a quick test, I was able to get more than 3600 By: Digium Subversion (svnbot) 2008-05-01 14:30:20 Repository: asterisk Revision: 115069 _U team/seanbright/resolve-shadow-warnings/ U team/seanbright/resolve-shadow-warnings/CHANGES U team/seanbright/resolve-shadow-warnings/Makefile.rules U team/seanbright/resolve-shadow-warnings/UPGRADE.txt U team/seanbright/resolve-shadow-warnings/apps/app_chanspy.c U team/seanbright/resolve-shadow-warnings/apps/app_directory.c U team/seanbright/resolve-shadow-warnings/apps/app_externalivr.c U team/seanbright/resolve-shadow-warnings/apps/app_macro.c U team/seanbright/resolve-shadow-warnings/apps/app_meetme.c U team/seanbright/resolve-shadow-warnings/apps/app_minivm.c U team/seanbright/resolve-shadow-warnings/apps/app_morsecode.c U team/seanbright/resolve-shadow-warnings/apps/app_queue.c U team/seanbright/resolve-shadow-warnings/apps/app_speech_utils.c U team/seanbright/resolve-shadow-warnings/apps/app_stack.c U team/seanbright/resolve-shadow-warnings/apps/app_voicemail.c U team/seanbright/resolve-shadow-warnings/apps/app_while.c U team/seanbright/resolve-shadow-warnings/channels/chan_iax2.c U team/seanbright/resolve-shadow-warnings/channels/chan_sip.c U team/seanbright/resolve-shadow-warnings/channels/chan_skinny.c U team/seanbright/resolve-shadow-warnings/channels/chan_zap.c U team/seanbright/resolve-shadow-warnings/channels/iax2-provision.c U team/seanbright/resolve-shadow-warnings/channels/iax2.h U team/seanbright/resolve-shadow-warnings/configs/queues.conf.sample U team/seanbright/resolve-shadow-warnings/configs/sip.conf.sample A team/seanbright/resolve-shadow-warnings/funcs/func_speex.c U team/seanbright/resolve-shadow-warnings/include/asterisk/audiohook.h U team/seanbright/resolve-shadow-warnings/main/audiohook.c U team/seanbright/resolve-shadow-warnings/main/features.c U team/seanbright/resolve-shadow-warnings/main/utils.c U team/seanbright/resolve-shadow-warnings/phoneprov/polycom.xml U team/seanbright/resolve-shadow-warnings/res/res_config_pgsql.c ------------------------------------------------------------------------ r115069 | seanbright | 2008-05-01 14:30:19 -0500 (Thu, 01 May 2008) | 278 lines Merged revisions 114830,114832,114834,114841,114845,114849,114852,114857,114866,114874,114876,114884,114888,114892,114899,114904,114906,114912,114922,114926,114931,114977,115018 via svnmerge from https://origsvn.digium.com/svn/asterisk/trunk ................ r114830 | qwell | 2008-04-29 13:10:55 -0400 (Tue, 29 Apr 2008) | 9 lines Merged revisions 114829 via svnmerge from https://origsvn.digium.com/svn/asterisk/branches/1.4 ........ r114829 | qwell | 2008-04-29 12:08:55 -0500 (Tue, 29 Apr 2008) | 1 line Change warning message to debug, since there are cases where 0 results is perfectly fine. ........ ................ r114832 | mmichelson | 2008-04-29 13:31:26 -0400 (Tue, 29 Apr 2008) | 3 lines Fix a crash happening in app_directory. This crash would occur if a users.conf existed. ................ r114834 | mmichelson | 2008-04-29 13:56:13 -0400 (Tue, 29 Apr 2008) | 8 lines Since there is now a globally available function for saying someone's name, a LOT of functions in app_directory can be removed since the ODBC-specific lookups are accomplished within app_voicemail. This change greatly reduces the amount of lines in app_directory that were solely for the purpose of looking up a name when ODBC_STORAGE is specified for voicemail. This commit also makes the name-saying interruptable via DTMF. ................ r114841 | mmichelson | 2008-04-29 14:48:26 -0400 (Tue, 29 Apr 2008) | 4 lines Make app_directory dependent on app_voicemail. This is because the function which says the person's name is handled inside app_voicemail now. ................ r114845 | kpfleming | 2008-04-29 14:58:48 -0400 (Tue, 29 Apr 2008) | 3 lines fix this logic to actually be correct... the fd can't be *both* -1 and an array index to be checked in rfds/efds (bug found by gcc-4.3) ................ r114849 | mmichelson | 2008-04-29 15:42:04 -0400 (Tue, 29 Apr 2008) | 22 lines Merged revisions 114848 via svnmerge from https://origsvn.digium.com/svn/asterisk/branches/1.4 ........ r114848 | mmichelson | 2008-04-29 14:40:06 -0500 (Tue, 29 Apr 2008) | 14 lines Use the MACRO_CONTEXT and MACRO_EXTEN channel variables instead of the channel's macrocontext and macroexten fields. This is needed because if macros are daisy-chained, the incorrect context and extension are placed on the new channel. I also added locking to the channel prior to accessing these variables as noted in trunk's janitor project file. (closes issue ASTERISK-11940) Reported by: darren1713 Patches: app_queue.c.macroextenpatch uploaded by darren1713 (license 116) (with modifications from me) Tested by: putnopvut ........ ................ r114852 | qwell | 2008-04-29 16:05:48 -0400 (Tue, 29 Apr 2008) | 1 line Fix formatting ................ r114857 | mmichelson | 2008-04-29 17:07:36 -0400 (Tue, 29 Apr 2008) | 15 lines Patching app_chanspy to jibe better with what is documented. This allows for a colon-delimited list of spygroups to be specified when calling the ChanSpy application with the 'g' option. Prior to this, you could only specify a single group when using the 'g' option. I also have upped the maximum number of spygroups to 128 and added a #define so that this can be easily increased or decreased later. (closes issue ASTERISK-11894) Reported by: jsmith Patches: app_chanspy_multiple_groups_v2.patch uploaded by jsmith (license 15) Tested by: atis, jvandal ................ r114866 | jpeeler | 2008-04-29 18:54:14 -0400 (Tue, 29 Apr 2008) | 2 lines Fixes a problem where all the templates were marked as dead no matter what. The templates should only be marked as dead if a configuration file has been successfully loaded and has changes. Bug found while making API documentation for 1.6.0. ................ r114874 | tilghman | 2008-04-30 01:05:25 -0400 (Wed, 30 Apr 2008) | 2 lines Document the Incomplete application addition. ................ r114876 | kpfleming | 2008-04-30 08:15:43 -0400 (Wed, 30 Apr 2008) | 10 lines Merged revisions 114875 via svnmerge from https://origsvn.digium.com/svn/asterisk/branches/1.4 ........ r114875 | kpfleming | 2008-04-30 07:14:07 -0500 (Wed, 30 Apr 2008) | 2 lines pay attention to *all* header files for dependency tracking, not just the local ones (inspired by r578 of asterisk-addons by tilghman) ........ ................ r114884 | kpfleming | 2008-04-30 10:49:51 -0400 (Wed, 30 Apr 2008) | 10 lines Merged revisions 114880 via svnmerge from https://origsvn.digium.com/svn/asterisk/branches/1.4 ........ r114880 | kpfleming | 2008-04-30 09:46:57 -0500 (Wed, 30 Apr 2008) | 2 lines use the ARRAY_LEN macro for indexing through the iaxs/iaxsl arrays so that the size of the arrays can be adjusted in one place, and change the size of the arrays from 32768 calls to 2048 calls when LOW_MEMORY is defined ........ ................ r114888 | jpeeler | 2008-04-30 12:14:43 -0400 (Wed, 30 Apr 2008) | 3 lines Fixes a bug where if a stream monitor thread was not created (caused from failure of opening or starting the stream) pthread_cancel was called with an invalid thread ID. ................ r114892 | russell | 2008-04-30 12:34:24 -0400 (Wed, 30 Apr 2008) | 36 lines Merged revisions 114891 via svnmerge from https://origsvn.digium.com/svn/asterisk/branches/1.4 ........ r114891 | russell | 2008-04-30 11:30:01 -0500 (Wed, 30 Apr 2008) | 28 lines Merge changes from team/russell/iax2_find_callno and iax2_find_callno_1.4 These changes address a critical performance issue introduced in the latest release. The fix for the latest security issue included a change that made Asterisk randomly choose call numbers to make them more difficult to guess by attackers. However, due to some inefficient (this is by far, an understatement) code, when Asterisk chose high call numbers, chan_iax2 became unusable after just a small number of calls. On a small embedded platform, it would not be able to handle a single call. On my Intel Core 2 Duo @ 2.33 GHz, I couldn't run more than about 16 IAX2 channels. Ouch. These changes address some performance issues of the find_callno() function that have bothered me for a very long time. On every incoming media frame, it iterated through every possible call number trying to find a matching active call. This involved a mutex lock and unlock for each call number checked. So, if the random call number chosen was 20000, then every media frame would cause 20000 locks and unlocks. Previously, this problem was not as obvious since Asterisk always chose the lowest call number it could. A second container for IAX2 pvt structs has been added. It is an astobj2 hash table. When we know the remote side's call number, the pvt goes into the hash table with a hash value of the remote side's call number. Then, lookups for incoming media frames are a very fast hash lookup instead of an absolutely insane array traversal. In a quick test, I was able to get more than 3600 |