[asterisk-bugs] [Asterisk 0016175]: Revision 202007 Introduces Deadlock
Asterisk Bug Tracker
noreply at bugs.digium.com
Wed Nov 4 17:45:00 CST 2009
A NOTE has been added to this issue.
======================================================================
https://issues.asterisk.org/view.php?id=16175
======================================================================
Reported By: paul-tg
Assigned To:
======================================================================
Project: Asterisk
Issue ID: 16175
Category: Channels/chan_sip/CodecHandling
Reproducibility: random
Severity: major
Priority: normal
Status: feedback
Asterisk Version: 1.6.2.0-rc3
JIRA: SWP-335
Regression: Yes
Reviewboard Link: https://reviewboard.asterisk.org/r/287/
SVN Branch (only for SVN checkouts, not tarball releases): N/A
SVN Revision (number only!): 202007
Request Review:
======================================================================
Date Submitted: 2009-11-03 19:23 CST
Last Modified: 2009-11-04 17:45 CST
======================================================================
Summary: Revision 202007 Introduces Deadlock
Description:
After upgrading from 1.6.0-beta9 to 1.6.2-rc2, we are experiencing
deadlocks in try_suggested_sip_codec(). "core show locks" points to code
introduced in revision 202007, which ironically was intended to protect
against deadlocks. The Asterisk build is completely stock except for a
modified SIP response code in chan_sip.c:handle_request_invite().
We have had two deadlocks so far in as many weeks, both times during
periods of lower than average call load.
======================================================================
----------------------------------------------------------------------
(0113162) paul-tg (reporter) - 2009-11-04 17:45
https://issues.asterisk.org/view.php?id=16175#c113162
----------------------------------------------------------------------
Apologies for that -- I marked it as a crash because that seemed to be the
most accurate description. The symptom is a deadlock; Asterisk stops
responding, but there is no crash so unfortunately I haven't a core dump
for you. I was mulling over instructing Operations to force a crash via
SIGSEGV, but decided that a core wouldn't be useful; would it be?
I would verify this in 1.6.2.0-rc3, but the issue occurs on a production
server and its deadlocking costs us money :-). I believe that
try_suggested_sip_codec() has not changed in the interim though (although
it has changed in trunk), so unless the other "side" of the deadlock has
been fixed since, this issue is unlikely to have been resolved.
In order to avoid this in future, I have reverted
try_suggested_sip_codec() to how it was before 1.6.2.0-beta4, and expect
the problem not to recur.
Issue History
Date Modified Username Field Change
======================================================================
2009-11-04 17:45 paul-tg Note Added: 0113162
======================================================================
More information about the asterisk-bugs
mailing list