[OpenSIPS-Users] db_mysql core dump

[Ovidiu Sas]

You are using mi to send a publish.  The body of that publish is taken
from the mi command directly.
I think the body should be copied into the shared memory and not used
directly from the mi structure.
Or maybe if it's used from the mi structure, the free is not necessary
(as it may be done by the underlaying mi implementation).
I will let Anca and Bogdan to comment more on this one.

An ugly workaraound here would be to remove the free (line 495,496 and 505,506).
This may lead to some memory leak, but it may be manageable until a
fix will be uploaded (should be pretty straight forward to fix this
issue).
Or maybe this is the proper fix (if the free is already done by the
underlaying mi implementation).


Regards,
Ovidiu Sas

On Fri, Dec 10, 2010 at 4:02 PM, Duane Larson <duane.larson at gmail.com> wrote:
> Thanks for working with me on this Ovidiu.  I think this is what you are
> wanting (sorry not good with debugging)???
>
> (gdb) frame 4
> #4  0x00007f1e935bc0b8 in send_publish_int (presentity=<value optimized
> out>,
>     publ=<value optimized out>, ev=0x7f1e88d57518, hash_index=332) at
> send_publish.c:496
> 496                             free(body->s);
>
> print body->s
> $1 = 0x83de41 "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n<xcap-diff
> xmlns=\"urn:ietf:params:xml:ns:xcap-diff\"
> xcap-root=\"http://xcap.aethercommunications.com/xcap-root\">\n<document
> new-etag=\"697c57e1c3c03ffe4febcc2c8905f"...
>
>
> On Fri, Dec 10, 2010 at 2:56 PM, Ovidiu Sas <osas at voipembedded.com> wrote:
>>
>> This is a crash in the pua module (different than the previous one).
>> send_publish.c:496 => free(body->s);
>>
>> Can you print the value and content of the body?
>>
>>
>> Thanks,
>> Ovidiu
>>
>> On Fri, Dec 10, 2010 at 3:28 PM, Duane Larson <duane.larson at gmail.com>
>> wrote:
>> > I did a total removal of the older OpenSIPS install and dropped and
>> > readded
>> > the Presence tables into MySQL.  That really helped thanks.  I still had
>> > some core dumps though not sure if it is going to happen more or not
>> > (will
>> > have to keep my eye on it), but this is what the core dump says
>> >
>> >
>> > (gdb) backtrace
>> > #0  0x00007f1e97c6e165 in raise () from /lib/libc.so.6
>> > #1  0x00007f1e97c70f70 in abort () from /lib/libc.so.6
>> > #2  0x00007f1e97ca427b in ?? () from /lib/libc.so.6
>> > #3  0x00007f1e97cadad6 in ?? () from /lib/libc.so.6
>> > #4  0x00007f1e935bc0b8 in send_publish_int (presentity=<value optimized
>> > out>, publ=<value optimized out>, ev=0x7f1e88d57518, hash_index=332) at
>> > send_publish.c:496
>> > #5  0x00007f1e935bcf12 in send_publish (publ=0x7fff97d6adc0) at
>> > send_publish.c:572
>> > #6  0x00007f1e92f968d3 in mi_pua_publish (cmd=0x84ddb0, param=<value
>> > optimized out>) at mi_func.c:237
>> > #7  0x00007f1e9540e388 in run_mi_cmd (rx_sock=<value optimized out>,
>> > tx_sock=<value optimized out>) at ../../mi/mi.h:104
>> > #8  mi_datagram_server (rx_sock=<value optimized out>, tx_sock=<value
>> > optimized out>) at datagram_fnc.c:529
>> > #9  0x00007f1e95410463 in datagram_process (rank=0) at mi_datagram.c:311
>> > #10 0x0000000000493f0c in start_module_procs () at sr_module.c:574
>> > #11 0x000000000042cefa in main_loop (argc=<value optimized out>,
>> > argv=<value
>> > optimized out>) at main.c:872
>> > #12 main (argc=<value optimized out>, argv=<value optimized out>) at
>> > main.c:1388
>> >
>> >
>> > libc.so.6 keeps popping up in the core dumps.
>> >
>> > On Fri, Dec 10, 2010 at 12:15 PM, Ovidiu Sas <osas at voipembedded.com>
>> > wrote:
>> >>
>> >> Have you also updated the layout of the presentity table?
>> >> Check for the extra_hdrs field.
>> >>
>> >> Regards,
>> >> Ovidiu Sas
>> >>
>> >> On Fri, Dec 10, 2010 at 12:52 PM, Duane Larson <duane.larson at gmail.com>
>> >> wrote:
>> >> > I did update the presentity table to be 5 instead of 4.  Perhaps i
>> >> > didn't
>> >> > clean up the previous install.  Let me try that.
>> >> >
>> >> > On Dec 10, 2010 11:46 AM, "Ovidiu Sas" <osas at voipembedded.com> wrote:
>> >> >
>> >> > Do you see any errors in the logs?
>> >> > Make sure that you clean up your previous install (remove all the
>> >> > opensips modules and all the opensips binaries) before re-installing
>> >> > opensips.
>> >> > Also, the presentity table has evolved (it has a new field:
>> >> > extra_hdrs).  The presentity version table should be 5.
>> >> >
>> >> >
>> >> > Regards,
>> >> > Ovidiu Sas
>> >> >
>> >> > On Fri, Dec 10, 2010 at 12:20 PM, Duane Larson
>> >> > <duane.larson at gmail.com>
>> >> > wrote:
>> >> >> I am not sure if I...
>> >> >
>> >> >> _______________________________________________
>> >> >
>> >> >> Users mailing list
>> >> >> Users at lists.opensips.org
>> >> >> http://lists.opensips.org/cgi-bin/mailman/listinfo...
>> >> >
>> >> > _______________________________________________
>> >> >
>> >> > Users mailing list
>> >> > Users at lists.opensips.org
>> >> > http://lists.opensips.org/cgi-bin/mailman/listinfo/users...
>> >> >
>> >> > _______________________________________________
>> >> > Users mailing list
>> >> > Users at lists.opensips.org
>> >> > http://lists.opensips.org/cgi-bin/mailman/listinfo/users
>> >> >
>> >> >
>> >>
>> >> _______________________________________________
>> >> Users mailing list
>> >> Users at lists.opensips.org
>> >> http://lists.opensips.org/cgi-bin/mailman/listinfo/users
>> >
>> >
>> >
>> > --
>> > --
>> > *--*--*--*--*--*
>> > Duane
>> > *--*--*--*--*--*
>> > --
>> >
>> > _______________________________________________
>> > Users mailing list
>> > Users at lists.opensips.org
>> > http://lists.opensips.org/cgi-bin/mailman/listinfo/users
>> >
>> >
>>
>> _______________________________________________
>> Users mailing list
>> Users at lists.opensips.org
>> http://lists.opensips.org/cgi-bin/mailman/listinfo/users
>
>
>
> --
> --
> *--*--*--*--*--*
> Duane
> *--*--*--*--*--*
> --
>
> _______________________________________________
> Users mailing list
> Users at lists.opensips.org
> http://lists.opensips.org/cgi-bin/mailman/listinfo/users
>
>