Summary:ASTERISK-07963: Module unload causes segfault
Reporter:jmls (jmls)Labels:
Date Opened:2006-10-19 11:58:29Date Closed:2007-03-06 12:25:31.000-0600
Versions:Frequency of
Environment:Attachments:( 0) btfull-20061019-1755.txt
Description:entering "module unload res_jabber.so" causes a segfault .

bt full attached
Comments:By: Joshua C. Colp (jcolp) 2006-10-19 12:28:27

Here mog, have a bug.

By: jmls (jmls) 2006-11-01 11:11:53.000-0600

nudge nudge

By: Anthony LaMantia (alamantia) 2006-11-01 14:01:09.000-0600

after looking at this problem, it seems to be a failure inside of iksemel a library we use to impliment jabber, i have submited a bugreport to the team
issue ASTERISK-5192 in their bugtracker.

also after following the code path in stream.c that leads to this bug it seems you can avoid it by turning off tls support in jabber.conf this would be the only fix i can see working until iksemel is updated.

By: Anthony LaMantia (alamantia) 2006-11-01 14:03:22.000-0600

place this in your jabber.conf for now   usetls=no

i am going to close this bug, please reopen it or report another one with a new backtrace if it still happens for you with tls disabled.

By: jmls (jmls) 2006-11-01 14:34:51.000-0600

sorry guys, I have always had usetls=no in my jabber.conf

By: Anthony LaMantia (alamantia) 2006-11-01 15:26:11.000-0600

in that case i am going to see see the console output with jabber debug  turned on.

By: jmls (jmls) 2006-11-01 16:01:28.000-0600

sure thing:

Connected to Asterisk SVN-trunk-r46319M currently running on foxtrot (pid = 4751)
   -- Remote UNIX connection
Verbosity is at least 10
foxtrot*CLI> jabber debug
Jabber Debugging Enabled.
foxtrot*CLI> core verbose 10
Verbosity is at least 10
foxtrot*CLI> module unload res_jabber.so
Disconnected from Asterisk server
asterisk -r
Connected to Asterisk SVN-trunk-r46319M currently running on foxtrot (pid = 28925)
   -- Remote UNIX connection
Verbosity is at least 3
foxtrot*CLI> show uptime
System uptime: 12 seconds

By: jmls (jmls) 2006-11-07 13:08:08.000-0600

housekeeping: ping :)

By: Anthony LaMantia (alamantia) 2006-11-09 11:47:50.000-0600

jmls, can you provide a bt full?

By: jmls (jmls) 2006-11-09 17:48:39.000-0600

urm, I did when I reported the bug ...

By: Anthony LaMantia (alamantia) 2006-11-13 15:42:42.000-0600

was it a bit longer then the one you uploaded i really need more infomration to trace the source of the problem

By: jmls (jmls) 2006-12-05 04:36:15.000-0600

just tried against a recent 1.4 svn (48223) and got the same error.

my bt full is this: (there is literally nothing else)

(gdb) bt full
#0  0x00290aad in iks_recv (prs=0x9e716bc, timeout=1) at stream.c:459
       data = (struct stream_data *) 0x9e7167c
       len = Variable "len" is not available.

By: Paul Cadach (pcadach) 2006-12-22 15:35:58.000-0600

Looks like the problem is because there is no shutdown of service/socket on module unload...

By: jmls (jmls) 2007-01-28 02:44:01.000-0600

still present on r51360 (1.4)

(gdb) bt full
#0  0x00903aad in iks_recv (prs=0x9f8e7dc, timeout=1) at stream.c:459
       data = (struct stream_data *) 0x9f8e79c
       len = Variable "len" is not available.

By: Serge Vecher (serge-v) 2007-02-26 13:48:22.000-0600

following alamantia's note 49334, a problem is in iksemel library. Has it been fixed there since november?

By: jmls (jmls) 2007-03-06 10:45:56.000-0600

I don't think that note is part of this bug report .. however, problem still exists with svn 1.4-r57914

By: Russell Bryant (russell) 2007-03-06 12:25:31.000-0600

I'm very sorry, but there is just nothing we can do for this bug.  The bug is within the iksemel library.  Because of this, among various other problems, we will probably end up completely replacing it with something else for 1.6.