[Home]

Summary:ASTERISK-22581: AMI: ConfbridgeList has race condition causing crashes
Reporter:Timo Teräs (fabled)Labels:
Date Opened:2013-09-25 03:12:15Date Closed:2013-09-28 10:33:51
Priority:MajorRegression?
Status:Closed/CompleteComponents:Applications/app_confbridge
Versions:11.5.1 Frequency of
Occurrence		Occasional
Related
Issues:
is duplicated byASTERISK-21859 Confbridge doesn't tear down an empty conference bridge when all users were kicked via end_marked=yes. Also, side effect crashes.
Environment:Attachments:
Description:I have started to get random crashes on 11.5.1 (10.4.1 is known to be stable) with ConfBridge setups. I managed to get non-debug enabled core, and pinpointed the issue to handling ConfbridgeList AMI request.

The incomplete backtrace is:
{noformat}
(gdb) where                                                                    
#0  0x1170ee21 in ast_channel_name ()                                          
#1  0x4f8d2000 in ?? () from /usr/lib/asterisk/modules/app_confbridge.so        
#2  0x4f8d2296 in ?? () from /usr/lib/asterisk/modules/app_confbridge.so
#3  0x117764e8 in ?? ()                                                
#4  0x1177a159 in ?? ()                                                        
#5  0x117d1cf0 in ?? ()                                                        
#6  0x117dc72b in ?? ()                                                        
#7  0x4feacb8f in start_thread () from /lib/libpthread.so.0.9.32        
#8  0x501b2ffa in clone () from /lib/libc.so.0.9.32                    
{noformat}

And I managed to disasemble the the two functions from app_confbridge.so with matches, so the beginning of stack trace is:
{noformat}
 ast_channel_name(NULL)
 action_confbridgelist_item() in app_confbridge.c
 action_confbridgelist() in app_confbridge.c
{noformat}

Apparently there is some locking issue / race condition that triggers this.
Comments:By: Timo Teräs (fabled) 2013-09-25 03:45:22.275-0500

This might be actually caused by ASTERISK-21859. Will apply the fix locally, and try to reproduce. Will report back.
By: Matt Jordan (mjordan) 2013-09-26 14:42:16.598-0500

Let us know what you find out.
By: Timo Teräs (fabled) 2013-09-28 03:28:36.575-0500

I'm relatively sure that this is duplicate - the symptoms match exactly. You can close this as duplicate, I'll reopen in the unlikely case if this happens again. Thanks.