|Summary:||ASTERISK-07811: app_queue keeps crashing|
|Reporter:||Roy Sigurd Karlsbakk (rkarlsba)||Labels:|
|Date Opened:||2006-09-24 09:07:47||Date Closed:||2011-06-07 14:00:37|
|Environment:||Attachments:||( 0) app_queue_crash.txt|
I'm not sure why, but app_queue keeps crashing. I have had two crashes in app_queue.c:501 and one in app_queue.c:666 (EVIL!) the latest week, and I cannot understand why. Backtraces are attached. I've looked through the logs, but I can't find _anything_ and full debug logging is always enabled to another box with enough diskspace... I haven't tried later than 1.2.10 yet, but AFAICS from the -svn lists, no relevant changes exist.
|Comments:||By: Roy Sigurd Karlsbakk (rkarlsba) 2006-09-24 09:08:36|
ops. sorry. this was filed into the wrong project. can someone fix this, please?
By: Serge Vecher (serge-v) 2006-09-25 08:42:50
1. Was asterisk built with 'make don-optimize'
2. Please do try to reproduce this on 184.108.40.206
By: Roy Sigurd Karlsbakk (rkarlsba) 2006-09-25 09:31:46
I changed to -O0 in the Makefile
I'll try to reproduce it with 220.127.116.11, but I really don't see why it should matter, since I can't find any updates of relevance....
By: Anthony LaMantia (alamantia) 2006-09-26 15:34:58
the modifcations you are making to one makefile don't have the same effect as make don-optimize
please recompile after select this from the menuselect menu. and submit the backtrace(s)
By: Roy Sigurd Karlsbakk (rkarlsba) 2006-09-27 03:30:23
the modifications do indeed have the same effects unless there is some really nasty magic around. From the makefile
ifneq ($(findstring dont-optimize,$(MAKECMDGOALS)),dont-optimize)
######### More GSM codec optimization
######### Uncomment to enable MMXTM optimizations for x86 architecture CPU's
######### which support MMX instructions. This should be newer pentiums,
######### ppro's, etc, as well as the AMD K6 and K7.
#K6OPT = -DK6OPT
#Tell gcc to optimize the code
...and further down
meaning when I changed OPTIMIZE+=-O6 to OPTIMIZE+=-O0, this has the very same effect
By: Anthony LaMantia (alamantia) 2006-09-28 10:41:12
is there anyway you can provide us with a core dump file .. or information on what the values being passed to strcasecmp are? this would be a huge help is tracking down the source of this issue some. atm i can't figure out of some of the pointers being returned AST_LIST_TRAVERSE are correct or not and having this information would be a huge help.
By: BJ Weschke (bweschke) 2006-09-28 11:17:02
MAYBE related to M7818. Not 100% sure...
By: Anthony LaMantia (alamantia) 2006-09-28 15:08:47
also posting any debug logs you may have would also be very helpfull is iron'ing out this issue.
By: Roy Sigurd Karlsbakk (rkarlsba) 2006-09-29 06:56:51
It is probably related to, or quite possible the same bug as 7818, since it occured right after a reload
By: Serge Vecher (serge-v) 2006-09-29 09:15:43
rkarlsba: jsmith is reporting in 7818 that the issue appears to be resolved. I think Kevin's commit may have also contributed http://lists.digium.com/pipermail/svn-commits/2006-September/017331.html. Can you please check out the latest 1.2 branch code?
By: Anthony LaMantia (alamantia) 2006-10-09 11:44:24
Did updating to the latest subversion release resolve this problem for you, or do we need to continue our investigation into the issue?
By: jmls (jmls) 2006-11-02 12:55:46.000-0600
rkarlsba, 7818 has been closed because the problem described in there was fixed. As previously mentioned, could you please retest against the latest 1.2 code ? thanks
By: Roy Sigurd Karlsbakk (rkarlsba) 2006-11-03 03:44:30.000-0600
Problem seems to be solved. Please close this. Thanks :)
By: jmls (jmls) 2006-11-03 03:52:34.000-0600
reporter states that the problem has been fixed. Thanks