Summary:ASTERISK-06263: Crash in chan_sip
Reporter:paradise (paradise)Labels:
Date Opened:2006-02-07 00:50:13.000-0600Date Closed:2008-01-15 16:53:24.000-0600
Versions:Frequency of
Environment:Attachments:( 0) crash3.txt
( 1) patch.sip.crash
Description:this is the first time i've faced this type of crash.
there is no clue!


bt is attached
Comments:By: Serge Vecher (serge-v) 2006-02-07 08:36:17.000-0600

paradise: did you capture the sip log as well, if not, can you reproduce the issue and get the sip log?

By: paradise (paradise) 2006-02-08 03:23:45.000-0600

no, unfotunatly i can't reproduce it.

By: Peng Yong (ppyy) 2006-02-08 06:09:09.000-0600

patch uploaded.

it's a bug introduced in:


static int handle_request(...)
     case SIP_ACK:
             __sip_ack(p, seqno, FLAG_RESPONSE, -1);

then sipmethod = -1 and msg = 0x0, and segment fault

By: Peng Yong (ppyy) 2006-02-08 06:10:43.000-0600

it should be fixed in 1.2 and trunk

By: Matt O'Gorman (mogorman) 2006-02-08 10:41:06.000-0600

is ppyy correct has upgrading solved your issue.

By: paradise (paradise) 2006-02-10 01:57:35.000-0600

ppyy: but in my BT "sipmethod" is "0" !!

By: paradise (paradise) 2006-02-10 01:59:10.000-0600

btw, i've upgraded to trunk-r9349 (latest) and waiting if this occurs again or not.

By: Peng Yong (ppyy) 2006-02-10 06:13:22.000-0600

it's so strange. there is only one __sip_ack() in function handle_request:

__sip_ack(p, seqno, FLAG_RESPONSE, -1);

how does the "sipmethod" become 0.

and it's indeed a bug when we call __sip_ack with "sipmethod" set to -1.

By: Mark Spencer (markster) 2006-02-19 23:18:56.000-0600

Fixed in SVN trunk and 1.2...  The -1 should be a 0...

By: Digium Subversion (svnbot) 2008-01-15 16:53:24.000-0600

Repository: asterisk
Revision: 10536

U   trunk/channels/chan_sip.c

r10536 | markster | 2008-01-15 16:53:24 -0600 (Tue, 15 Jan 2008) | 2 lines

Remove gratuitous -1 (bug ASTERISK-6263)