Summary:	ASTERISK-12146: iax2 deadlock asterisk
Reporter:	Francesco Romano (francesco_r)	Labels:
Date Opened:	2008-06-05 06:32:14	Date Closed:	2008-06-25 20:11:03
Priority:	Blocker	Regression?	No
Status:	Closed/Complete	Components:	Channels/chan_iax2
Versions:		Frequency of Occurrence
Related Issues:
Environment:		Attachments:	( 0) chan_iax2-patch.txt ( 1) locks.txt ( 2) threads.txt
Description:	I have upgraded two servers conected via iax to asterisk 1.4.21-rc1 and after some hours, asterisk deadlocked and the CPU usage jumped high in both. This problem appeared many times in office hours and i reverted immediately to 1.20.1 because there are productions environvment. So i don't have any informations. But i'll try to reproduce the bug in test enviroment so i'll post the necessary info later. Only to advice that something is wrong in chan_iax2 with the lastest changes from Russel.
Comments:	By: Tilghman Lesher (tilghman) 2008-06-05 08:21:52 You're going to need to compile with DEBUG_THREADS and give us the output of 'core show locks' while this is occurring. By: Peter Whisker (whiskerp) 2008-06-06 10:05:45 I'm seeing something similar - midway through an IAX2 call, the CPU goes to 100%. My version is Asterisk SVN-branch-1.4-r120226M Peter By: Russell Bryant (russell) 2008-06-06 11:16:48 If there is a 100% CPU problem, then I would need the output of "core show locks", as well as a "thread apply all bt" from gdb. By: tbsky (tbsky) 2008-06-16 01:24:23 i upgrade two  asterisk box to 1.4.21. they use iax2 trunk. the trunk is dead after talking for a few minutes. and i need to restart asterisk to recover. however i didn't notice the high cpu usage. just the trunk become dead suddenly. i switch back to 1.4.20.1 and everything is fine. By: Nils Hammar (ehsnils) 2008-06-16 01:34:23 I have experienced the same thing, I also filed more details about this in bug 12864 (sorry I missed this bug). By: Rangi Biddle (rangib) 2008-06-17 13:11:54 I have also been experiencing this same issue.  I have attached three files to this bug (locks.txt, threads.txt, chan_iax2-patch.txt).  The patch that I have attached was from here http://bugs.digium.com/view.php?id=12717 By: Digium Subversion (svnbot) 2008-06-17 13:50:00 Repository: asterisk Revision: 123391 U   branches/1.4/channels/chan_iax2.c ------------------------------------------------------------------------ r123391 | tilghman | 2008-06-17 13:49:54 -0500 (Tue, 17 Jun 2008) | 3 lines Fix 3 more places where failing to lock a structure could cause the wrong lock to be unlocked.  (Closes issue ASTERISK-12146) ------------------------------------------------------------------------ http://svn.digium.com/view/asterisk?view=rev&revision=123391 By: Digium Subversion (svnbot) 2008-06-17 13:51:40 Repository: asterisk Revision: 123392 _U  trunk/ U   trunk/channels/chan_iax2.c ------------------------------------------------------------------------ r123392 | tilghman | 2008-06-17 13:51:35 -0500 (Tue, 17 Jun 2008) | 11 lines Merged revisions 123391 via svnmerge from https://origsvn.digium.com/svn/asterisk/branches/1.4 ........ r123391 | tilghman | 2008-06-17 13:56:53 -0500 (Tue, 17 Jun 2008) | 3 lines Fix 3 more places where failing to lock a structure could cause the wrong lock to be unlocked.  (Closes issue ASTERISK-12146) ........ ------------------------------------------------------------------------ http://svn.digium.com/view/asterisk?view=rev&revision=123392 By: Digium Subversion (svnbot) 2008-06-17 13:54:33 Repository: asterisk Revision: 123394 _U  branches/1.6.0/ U   branches/1.6.0/channels/chan_iax2.c ------------------------------------------------------------------------ r123394 | tilghman | 2008-06-17 13:54:31 -0500 (Tue, 17 Jun 2008) | 19 lines Merged revisions 123392 via svnmerge from https://origsvn.digium.com/svn/asterisk/trunk ................ r123392 | tilghman | 2008-06-17 13:57:45 -0500 (Tue, 17 Jun 2008) | 11 lines Merged revisions 123391 via svnmerge from https://origsvn.digium.com/svn/asterisk/branches/1.4 ........ r123391 | tilghman | 2008-06-17 13:56:53 -0500 (Tue, 17 Jun 2008) | 3 lines Fix 3 more places where failing lock a structure could cause the wrong lock to be unlocked.  (Closes issue ASTERISK-12146) ........ ................ ------------------------------------------------------------------------ http://svn.digium.com/view/asterisk?view=rev&revision=123394 By: Tilghman Lesher (tilghman) 2008-06-17 13:57:38 Just to be clear, rangib's upload of 'core show locks' was what enabled me to find and fix this issue so quickly.