[Home]

Summary:ASTERISK-15285: segfault error 4 in libpthread on Ubuntu
Reporter:alchemist (alchemist)Labels:
Date Opened:2009-12-07 09:37:15.000-0600Date Closed:2010-01-18 08:58:44.000-0600
Priority:CriticalRegression?No
Status:Closed/CompleteComponents:General
Versions:Frequency of
Occurrence
Related
Issues:
Environment:Attachments:
Description:Hi!
I've installed asterisk-1.6.2.0-rc3 & dahdi-linux-complete-2.2.0.2+2.2.0 on fresh Ubuntu server 9.04 x64
A few times a day my asterisk crashed with message:
kernel: [224188.892433] asterisk[10141]: segfault at 140 ip 00007f300e8b61a5 sp 00007f2fefbe24c8 error 4 in libpthread-2.9.so[7f300e8ac000+17000]
Yesterday i've installed asterisk-1.6.2.0-rc2 & dahdi-linux-complete-2.2.0.2+2.2.0 from repository on fresh Ubuntu Server 9.10 i386, but it still crashes:
kernel: [67103.338993] asterisk[17030]: segfault at a4 ip 00463111 sp b51847ac error 4 in libpthread-2.10.1.so[45a000+15000]


****** ADDITIONAL INFORMATION ******

Digium TE205P
Comments:By: Russell Bryant (russell) 2009-12-07 13:58:11.000-0600

Please get a backtrace of the crash.  There are some instructions in doc/backtrace.txt.
By: alchemist (alchemist) 2009-12-10 08:25:42.000-0600

(gdb) bt full
#0  0x00a93111 in __pthread_mutex_unlock_usercnt () from /lib/tls/i686/cmov/libpthread.so.0
No symbol table info available.
#1  0x001e9996 in pthread_mutex_unlock () from /lib/tls/i686/cmov/libc.so.6
No symbol table info available.
#2  0x03659c39 in ?? () from /usr/lib/asterisk/modules/chan_sip.so
No symbol table info available.
#3  0x0067a632 in ast_do_masquerade () from /usr/sbin/asterisk
No symbol table info available.
#4  0x0067dc6a in ?? () from /usr/sbin/asterisk
No symbol table info available.
ASTERISK-1  0x00680297 in ast_channel_bridge () from /usr/sbin/asterisk
No symbol table info available.
ASTERISK-2  0x006ab51e in ast_bridge_call () from /usr/sbin/asterisk
No symbol table info available.
ASTERISK-3  0x04fb82ed in ?? () from /usr/lib/asterisk/modules/app_dial.so
No symbol table info available.
ASTERISK-4  0x04fba655 in ?? () from /usr/lib/asterisk/modules/app_dial.so
No symbol table info available.
ASTERISK-5  0x006dbcb5 in pbx_exec () from /usr/sbin/asterisk
No symbol table info available.
ASTERISK-6 0x006e2f7e in ?? () from /usr/sbin/asterisk
No symbol table info available.
ASTERISK-7 0x006e347c in ast_spawn_extension () from /usr/sbin/asterisk
No symbol table info available.
ASTERISK-8 0x006e9336 in ?? () from /usr/sbin/asterisk
No symbol table info available.
ASTERISK-9 0x006eaa4d in ?? () from /usr/sbin/asterisk
No symbol table info available.
ASTERISK-10 0x0072d674 in ?? () from /usr/sbin/asterisk
No symbol table info available.
ASTERISK-11 0x00a8f80e in start_thread () from /lib/tls/i686/cmov/libpthread.so.0
No symbol table info available.
ASTERISK-12 0x001dc7ee in clone () from /lib/tls/i686/cmov/libc.so.6
No symbol table info available.
By: Leif Madsen (lmadsen) 2010-01-07 11:19:32.000-0600

And this is with DONT_OPTIMIZE enabled?
By: Leif Madsen (lmadsen) 2010-01-07 11:19:58.000-0600

Also, can you verify this is still an issue since 1.6.2.0 has now been released?
By: Mark Jacobsen (markj) 2010-01-11 13:06:21.000-0600

Alchemist: I'm having the same trouble on a server running 1.6.1.6 - please forgive me but where do I look for doc/backtrace.txt ?  Also  Where do I check to see if DONT_OPTIMIZE is enabled?

mj
By: Leif Madsen (lmadsen) 2010-01-11 14:27:30.000-0600

markj:  1.6.1.6 is pretty old -- please re-test with something newer.

DONT_OPTIMIZE is found in the Compiler Flags section of menuselect. Recompile and reinstall after changing this flag for the settings to take effect.

doc/backtrace.txt is a relative path to your Asterisk source directory, e.g. /usr/src/asterisk-current/doc/backtrace.txt
By: Mark Jacobsen (markj) 2010-01-16 12:31:44.000-0600

Installed 1.6.2.1-rc1 and checked "DONT_OPTIMIZE" on install - seems to be running good so far (about 3 or 4 days) - had a small problem with dahdi on the PRI the next morning after upgrade but restarting cleared it up and haven't had that come back so far (possibly un-reladed)
By: Leif Madsen (lmadsen) 2010-01-18 08:58:43.000-0600

I'm closing this issue for now as it appears this may be resolved for the reporter. If they have additional issues, please feel free to reopen this issue.