Summary:ASTERISK-12265: Segmentation fault in ast_atomic_fetchadd_int
Reporter:Grant Murray (grant)Labels:
Date Opened:2008-06-26 13:16:41Date Closed:2011-06-07 14:01:06
Versions:Frequency of
Environment:Attachments:( 0) 20080626__bug12934.diff.txt
( 1) bt-01.txt
( 2) bt-02.txt
( 3) list-01.txt
Description:Crash occurs after running for some time. I am using the manager API to originate calls. Originating at the rate of about 2 per second with aprrox 60 simultaneous sessions.
Comments:By: Grant Murray (grant) 2008-06-26 13:46:46

crashed again in the same place, it ran for only 3 minutes (about 200 originations) before croaking.

Edit - seem to be a pattern, it crashes everytime after 3 minutes

By: Tilghman Lesher (tilghman) 2008-06-26 15:28:58

Patch uploaded; please test and let me know if it fixes your issue.

By: Grant Murray (grant) 2008-06-26 15:46:45

Applied your patch - it crashed in a different place this time. see bt-02.txt

By: Tilghman Lesher (tilghman) 2008-06-26 16:37:36

I'm going to need the output of "list" in gdb.

By: Grant Murray (grant) 2008-06-26 19:15:03

added the list output as requested

By: Tilghman Lesher (tilghman) 2008-06-26 22:45:53

This indicates corrupt memory.  You're going to need to follow the instructions in doc/valgrind.txt.

By: Grant Murray (grant) 2008-06-27 12:49:00

I tried running it under valgrind, but that interferes with the process too much, I was getting 98% non-connects from the carrier. I cannot run asterisk like that.

However, running asterisk from the command line has allowed it to run for more than an hour so far. So that has also changed the nature of the problem.

Unfortunately, observing this problem seems to interfere with it.

By: Grant Murray (grant) 2008-10-20 16:32:29

I have not been able to repeat this under 1.6
Please close this bug report.

By: Tilghman Lesher (tilghman) 2008-10-21 09:35:56

Closed on request of reporter.