Bug#539681: More information

Fri Aug 28 16:36:05 UTC 2009

A member of the pkg-voip-maintainers team has provided me a link to an
archive of 1.6.1.0 packages that I believe are not affected by this
bug (thanks Jonas!). I'll be downgrading from 1.6.2beta3 (currently in
unstable) to these packages soon and will report back if the problem
still persists with stable 1.6.1.0. The links to the 1.6.1.0 packages
are:

----------------------------------------------------------------------
Pick one (and only one!) of the below depending on your system:

deb http://debian.jones.dk/ lenny asterisk
deb http://debian.jones.dk/ sid asterisk

Packages are available there for i386 and amd64.
----------------------------------------------------------------------

In the meantime, I obtained coredumps for the crashes and decoded them.
Since I've been collecting coredumps I've obtained these coredumps:

core.altamira-2009-08-10T22:42:48-0400
core.altamira-2009-08-14T15:00:35-0400
core.altamira-2009-08-14T15:02:24-0400
core.altamira-2009-08-14T15:49:26-0400
core.altamira-2009-08-14T16:14:15-0400
core.altamira-2009-08-14T16:51:22-0400
core.altamira-2009-08-14T20:39:29-0400
core.altamira-2009-08-14T21:32:36-0400
core.altamira-2009-08-14T21:34:14-0400
core.altamira-2009-08-15T20:20:05-0400
core.altamira-2009-08-16T20:56:16-0400
core.altamira-2009-08-17T22:14:26-0400
core.altamira-2009-08-19T19:06:10-0400
core.altamira-2009-08-21T12:03:51-0400
core.altamira-2009-08-21T17:36:15-0400
core.altamira-2009-08-21T20:03:38-0400
core.altamira-2009-08-22T20:27:47-0400
core.altamira-2009-08-25T13:54:13-0400
core.altamira-2009-08-25T21:47:32-0400
core.altamira-2009-08-25T21:48:54-0400
core.altamira-2009-08-27T12:30:40-0400
core.altamira-2009-08-27T13:25:18-0400
core.altamira-2009-08-28T07:45:27-0400

All these crashes are happening in the same place:

Core was generated by `/usr/sbin/asterisk -f -p -g -U asterisk -vvvg -c'.
Program terminated with signal 11, Segmentation fault.
#0  0xb6948093 in reqprep (req=0xb526688c, p=0xb778b748, sipmethod=5, seqno=102, newbranch=1) at chan_sip.c:8752
8752	chan_sip.c: No such file or directory.
	in chan_sip.c
(gdb) where
#0  0xb6948093 in reqprep (req=0xb526688c, p=0xb778b748, sipmethod=5, seqno=102, newbranch=1) at chan_sip.c:8752
#1  0xb694989a in transmit_reinvite_with_sdp (p=0xb778b748, t38version=0, oldsdp=0) at chan_sip.c:9791
#2  0xb694a471 in sip_set_rtp_peer (chan=0xb778efe8, rtp=0x0, vrtp=0x0, trtp=0x0, codecs=0, nat_active=0)
    at chan_sip.c:24322
#3  0x080fb704 in bridge_native_loop (c0=0x8a043c0, c1=0xb778efe8, flags=0, fo=0xb5266e98, rc=0xb5266e94, 
    timeoutms=-1) at rtp.c:4056
#4  ast_rtp_bridge (c0=0x8a043c0, c1=0xb778efe8, flags=0, fo=0xb5266e98, rc=0xb5266e94, timeoutms=-1) at rtp.c:4533
#5  0x0809572b in ast_channel_bridge (c0=0xb778efe8, c1=0x8a043c0, config=0xb5267d1c, fo=0xb5266e98, rc=0xb5266e94)
    at channel.c:4998
#6  0x080bc006 in ast_bridge_call (chan=0xb778efe8, peer=0x8a043c0, config=0xb5267d1c) at features.c:2558
#7  0xb705007c in dial_exec_full (chan=0xb778efe8, data=0xb5269f18, peerflags=0xb5267e90, continue_exec=0x0)
    at app_dial.c:2184
#8  0xb7052225 in dial_exec (chan=0xb778efe8, data=0xb5269f18) at app_dial.c:2258
#9  0x080f2676 in pbx_exec (c=0xb778efe8, app=0x8a029d0, data=0xb5269f18) at pbx.c:1349
#10 0x080f349b in pbx_extension_helper (c=0xb778efe8, con=0x0, context=0xb778f258 "default", exten=0xb778f2a8 "s", 
    priority=7, label=0x0, callerid=0xb7703ed8 "9193929118", action=E_SPAWN, found=0xb526c348, 
    combined_find_spawn=1) at pbx.c:3691
#11 0x080f4f79 in __ast_pbx_run (c=0xb778efe8, args=0x0) at pbx.c:4234
#12 0x080f61d0 in pbx_thread (data=0xb778efe8) at pbx.c:4521
#13 0x0812c267 in dummy_start (data=0xb77054b8) at utils.c:968
#14 0xb7b084b5 in start_thread () from /lib/i686/cmov/libpthread.so.0
#15 0xb7d3aa5e in clone () from /lib/i686/cmov/libc.so.6
(gdb) 

A quick Google search for "reqprep transmit_reinvite_with_sdp asterisk"
didn't return anything so I don't know if this is a problem that
upstream is aware of, and possibly fixed in beta4.

I have been running safe_asterisk so at least the asterisk process
restarts after a crash. Running asterisk doesn't not compensate for the
dropped calls when the asterisk process crashes, though, but helps a
little.

Anyway, I am downgrading in the hopes that I won't see the problem with
1.6.1.0.

Cheers,

Eloy Paris.-