Summary:	ASTERISK-16769: Asterisk always crashes when linked against glibc 2.12.1
Reporter:	Andrew Walrond (awalrond)	Labels:
Date Opened:	2010-10-05 05:13:46	Date Closed:	2011-06-07 14:00:54
Priority:	Critical	Regression?	No
Status:	Closed/Complete	Components:	General
Versions:		Frequency of Occurrence
Related Issues:
Environment:		Attachments:	( 0) backtrace2.txt
Description:	Always segfaults, same place every time in pthread_cond_timedwait I've tried the latest from the svn 1.6.2 branch, same result. Works just fine with glibc 11.2.2
Comments:	By: Leif Madsen (lmadsen) 2010-10-05 15:34:39 Thank you for your bug report. In order to move your issue forward, we require a backtrace from the core file produced after the crash. Please see the doc/backtrace.txt file in your Asterisk source directory. Also, be sure you have DONT_OPTIMIZE enabled in menuselect within the Compiler Flags section, then: make install after enabling, reproduce the crash, and then execute the instructions in doc/backtrace.txt. When complete, attach that file to this issue report. Thanks! By: Bradley Watkins (marquis) 2010-10-05 17:57:28 Can you provide a brief description about how to reproduce the crash?  I have tried compiling and doing some very basic things with the latest 1.6.2 branch on Fedora 14 Beta (which comes with glibc 2.12) and have thus far not seen a crash. By: Andrew Walrond (awalrond) 2010-10-05 18:23:13 I just run asterisk. either no config files or the default samples and it crashes within a second or two. But like I said, it works fine against glibc 2.11.2. thanks for trying it on fedora; I am using the vanilla 2.12.1 release of glibc here. I'll check out the fedora branch of glibc 2.12 to see if there have been any patches in this area, and also to test asterisk against it. (I see there was some chatter on the glibc ML regarding this pthread function recently) I'll come back with my findings tomorrow. By: Bradley Watkins (marquis) 2010-10-05 18:41:40 Yes, I'd be interested to know if there is something peculiar about Fedora's glibc versus vanilla.  As per most distros, I suspect there is. However, I'd be more inclined to think it's a bugfix or set of bugfixes that resolve the issue you're seeing. FWIW, I ran exactly the same tests you did (no configs, untouched samples) and a couple other minor ones and did not have a crash. By: Sébastien Cramatte (scramatte) 2010-10-07 05:47:48 I'm running debian squeeze  with glibc 2.11.2 and asterisk crash every minutes. Before I was running debian lenny with older glibc (2.7.x) and asterisk crash every 5mn... Might be is an issue on x86_64 platform only ??? By: Leif Madsen (lmadsen) 2011-01-06 14:42:02.000-0600 Is this still an issue? By: Andrew Walrond (awalrond) 2011-01-07 06:44:52.000-0600 probably; my todo stack got swamped recently and this issue hasn't floated to the top yet :( Leave it active for another week or so and I'll get to it. Thanks for your patience! By: Tilghman Lesher (tilghman) 2011-01-07 12:49:12.000-0600 Based upon the backtrace, I suspect that this version of glibc changed how it handles va_arg style functions and the change wasn't completely tested.  I agree that there's probably a bugfix to glibc that you're missing. By: Leif Madsen (lmadsen) 2011-01-19 10:14:44.000-0600 This appears to be a glibc issue and not an Asterisk issue, so I'm going to close this issue.