[OpenSIPS-Users] Fwd: Segmentation fault when TM and DIALOG timer routines are not executed sequentially

[Mickael Marrache]

Any news regarding this issue?


---------- Forwarded message ----------
From: Mickael Marrache <mickaelmarrache at gmail.com>
Date: Mon, Jul 20, 2015 at 10:00 AM
Subject: Re: [OpenSIPS-Users] Segmentation fault when TM and DIALOG timer
routines are not executed sequentially
To: Bogdan-Andrei Iancu <bogdan at opensips.org>


Hi,

Sorry to reply late.

(gdb) print dlg->h_entry
$1 = 775173170
(gdb) print d_table->size
$2 = 4096

Mickael

On Fri, Jul 3, 2015 at 5:23 PM, Bogdan-Andrei Iancu <bogdan at opensips.org>
wrote:

> Hi Mickael,
>
> Could print in frame 0 the following:
>     dlg->h_entry
> and
>     d_table->size
>
> Best regards,
>
> Bogdan-Andrei Iancu
> OpenSIPS Founder and Developerhttp://www.opensips-solutions.com
>
> On 02.07.2015 16:52, Mickael Marrache wrote:
>
> Hi,
>
> I'm running version 1.11.2.
>
> Mickael
>
> On Thu, Jul 2, 2015 at 4:49 PM, Bogdan-Andrei Iancu <bogdan at opensips.org>
> wrote:
>
>> Hi Mickael,
>>
>> What OpenSIPS version are you using ?
>>
>> Best Regards,
>>
>> Bogdan-Andrei Iancu
>> OpenSIPS Founder and Developerhttp://www.opensips-solutions.com
>>
>> On 02.07.2015 13:25, Mickael Marrache wrote:
>>
>> Hi,
>>
>> We currently have OpenSIPS crashes that look to be caused by
>> synchronization issues between the TM and DIALOG modules.
>>
>> We started to get these crashes after setting the following modparam:
>>
>> modparam("tm", "own_timer_proc", 2)
>>
>> Note that we don't set the own_timer_proc parameter for the DIALOG
>> module, therefore it uses the global timer process.
>>
>> Do you have an idea what may cause the issue?
>>
>> The backtrace extracted from the core file:
>>
>> #0  0x00002b69aa157a28 in next_state_dlg (dlg=0x2b69b1271408, event=1,
>> dir=2, old_state=0x7fff7e830358, new_state=0x7fff7e83035c,
>> unref=0x7fff7e830354, is_replicated=0 '\0') at dlg_hash.c:777
>> #1  0x00002b69aa14aafc in dlg_onreply (t=0x0, type=<value optimized out>,
>> param=<value optimized out>) at dlg_handlers.c:457
>> #2  0x00002b69aa14bc67 in unreference_dialog_create
>> (dialog=0x2b69b1271408) at dlg_handlers.c:744
>> #3  0x00002b69a7fe2b59 in empty_tmcb_list (head=0x2b69b837c948) at
>> t_hooks.c:56
>> #4  0x00002b69a7fd057e in free_cell (dead_cell=0x2b69b837c8d8) at
>> h_table.c:140
>> #5  0x00002b69a7fe5e78 in delete_cell (p_cell=0x2b69b837c8d8, unlock=1)
>> at timer.c:240
>> #6  0x00002b69a7fe69ac in timer_routine (ticks=<value optimized out>,
>> set=<value optimized out>) at timer.c:416
>> #7  0x00000000004babe2 in start_timer_processes () at timer.c:384
>> #8  0x0000000000434711 in main_loop () at main.c:1016
>> #9  0x00000000004386f5 in main (argc=<value optimized out>,
>> argv=0x7fff7e830838) at main.c:1634
>>
>> The seg fault happens in the next_state_dlg function (line 777 - when
>> trying to lock the dialog entry):
>>
>> (gdb) list
>> 772             struct dlg_entry * d_entry;
>> 773
>> 774             d_entry = &(d_table->entries[dlg->h_entry]);
>> 775             *unref = 0;
>> 776
>> 777             dlg_lock(d_table, d_entry);
>> 778
>> 779             *old_state = dlg->state;
>> 780
>> 781             switch (event) {
>>
>> Thanks,
>> Mickael
>>
>>
>> _______________________________________________
>> Users mailing listUsers at lists.opensips.orghttp://lists.opensips.org/cgi-bin/mailman/listinfo/users
>>
>>
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opensips.org/pipermail/users/attachments/20150806/b554f4e4/attachment-0001.htm>