[Serusers] RPID problem in 0.9.0 release

[Maxim Sobolev]

Francesco Bottà wrote:
> Hi all,
> 
> I don't know if this is related to this change in the code (or other last commit) but now the release 0.9.0 get terminated due to SIGCHLD:

Hmm, the problem isn't SIGCHLD, but SIGSEGV received by one of childrens:

child process 11345 exited by a signal 11

Can you please get a corefile and post a backtrace or enable more 
verbose debugging mode so that we can see what operation SER tries to 
make when this SIGSEGV happens?

-Maxim

> 
> Jan 13 16:06:40 spsexp /usr/local/sbin/ser[11333]: Maxfwd module- initializing 
> Jan 13 16:06:40 spsexp /usr/local/sbin/ser[11333]: permissions - initializing 
> Jan 13 16:06:40 spsexp /usr/local/sbin/ser[11333]: Default allow file (/usr/local/etc/ser/permissions.allow) not found => empty rule set 
> Jan 13 16:06:40 spsexp /usr/local/sbin/ser[11333]: Default deny file (/usr/local/etc/ser/permissions.deny) parsed 
> Jan 13 16:06:40 spsexp /usr/local/sbin/ser[11333]: PIKE - initializing 
> Jan 13 16:06:40 spsexp /usr/local/sbin/ser[11333]: INFO:pike:init_lock_set: probing 256 set size 
> Jan 13 16:06:40 spsexp /usr/local/sbin/ser[11333]: AVPops - initializing 
> Jan 13 16:06:40 spsexp /usr/local/sbin/ser[11333]: INFO: udp_init: SO_RCVBUF is initially 65535 
> Jan 13 16:06:40 spsexp /usr/local/sbin/ser[11333]: INFO: udp_init: SO_RCVBUF is finally 131070 
> Jan 13 16:06:40 spsexp /usr/local/sbin/ser[11343]: INFO: fifo process starting: 11343 
> Jan 13 16:06:40 spsexp /usr/local/sbin/ser[11343]: rtpp_test: RTP proxy found, support for it enabled 
> Jan 13 16:06:40 spsexp /usr/local/sbin/ser[11343]: SER: open_uac_fifo: fifo server up at /tmp/ser_fifo... 
> Jan 13 16:06:40 spsexp /usr/local/sbin/ser[11343]: WARNING: no fifo_db_url given - fifo DB commands disabled! 
> Jan 13 16:06:40 spsexp /usr/local/sbin/ser[11344]: rtpp_test: RTP proxy found, support for it enabled 
> Jan 13 16:06:40 spsexp /usr/local/sbin/ser[11345]: rtpp_test: RTP proxy found, support for it enabled 
> Jan 13 16:06:40 spsexp /usr/local/sbin/ser[11346]: rtpp_test: RTP proxy found, support for it enabled 
> Jan 13 16:06:40 spsexp /usr/local/sbin/ser[11348]: rtpp_test: RTP proxy found, support for it enabled 
> Jan 13 16:06:41 spsexp /usr/local/sbin/ser[11347]: rtpp_test: RTP proxy found, support for it enabled 
> Jan 13 16:06:41 spsexp /usr/local/sbin/ser[11349]: rtpp_test: RTP proxy found, support for it enabled 
> Jan 13 16:06:41 spsexp /usr/local/sbin/ser[11350]: rtpp_test: RTP proxy found, support for it enabled 
> Jan 13 16:06:41 spsexp /usr/local/sbin/ser[11352]: rtpp_test: RTP proxy found, support for it enabled 
> Jan 13 16:06:41 spsexp /usr/local/sbin/ser[11354]: rtpp_test: RTP proxy found, support for it enabled 
> Jan 13 16:06:41 spsexp /usr/local/sbin/ser[11351]: rtpp_test: RTP proxy found, support for it enabled 
> Jan 13 16:06:41 spsexp /usr/local/sbin/ser[11333]: rtpp_test: RTP proxy found, support for it enabled 
> Jan 13 16:07:12 spsexp /usr/local/sbin/ser[11333]: child process 11345 exited by a signal 11 
> Jan 13 16:07:12 spsexp /usr/local/sbin/ser[11333]: core was generated 
> Jan 13 16:07:12 spsexp /usr/local/sbin/ser[11333]: INFO: terminating due to SIGCHLD 
> Jan 13 16:07:12 spsexp /usr/local/sbin/ser[11348]: INFO: signal 15 received 
> Jan 13 16:07:12 spsexp /usr/local/sbin/ser[11343]: INFO: signal 15 received 
> Jan 13 16:07:12 spsexp /usr/local/sbin/ser[11349]: INFO: signal 15 received 
> Jan 13 16:07:12 spsexp /usr/local/sbin/ser[11350]: INFO: signal 15 received 
> Jan 13 16:07:12 spsexp /usr/local/sbin/ser[11351]: INFO: signal 15 received 
> Jan 13 16:07:12 spsexp /usr/local/sbin/ser[11352]: INFO: signal 15 received 
> Jan 13 16:07:12 spsexp /usr/local/sbin/ser[11354]: INFO: signal 15 received 
> Jan 13 16:07:15 spsexp /usr/local/sbin/ser[11333]: PIKE - destroying module 
> 
> How can I get more information to find the cause of this SIGCHLD???
> 
> Many thanx,
> 
> Verbal
> 
> ----- Original Message ----- 
> From: "Maxim Sobolev" <sobomax at portaone.com>
> To: "Richard" <richard at o-matrix.org>
> Cc: <Daniel-Constantin.Mierla at fokus.fraunhofer.de>; "'Francesco Bottà'" <francesco.botta at eutelia.it>; <serusers at lists.iptel.org>
> Sent: Thursday, January 13, 2005 10:15 AM
> Subject: Re: [Serusers] RPID problem in 0.9.0 release
> 
> 
> The fix is incorrect since it creates memory leak. I've just committed a 
> proper fix. Please test and let me know if you are still experiencing 
> this problem.
> 
> -Maxim
> 
> Richard wrote:
> 
>>>-----Original Message-----
>>>From: serusers-bounces at iptel.org [mailto:serusers-bounces at lists.iptel.org] On
>>>Behalf Of Daniel-Constantin.Mierla at fokus.fraunhofer.de
>>>Sent: Monday, December 27, 2004 8:01 AM
>>>To: Francesco Bottà
>>>Cc: serusers at lists.iptel.org; Maxim Sobolev
>>>Subject: Re: [Serusers] RPID problem in 0.9.0 release
>>>
>>>auth_db is one of the module which wasn't reviewed yet - reason to have a
>>>prelease (or testing release) by now. As far as I understood, with these
>>>two
>>>new parameters you load the values of the appropriate columns from
>>>database in
>>>AVP list, while append_rpid_hf() uses a static buffer which is filled when
>>>use_rpid=1. So you have to keep the value to 1 for now.
>>
>>
>>It appears to be a bug in the auth_db module. It saves rpid in the static
>>buffer as well as the avp rpid. The problem is the static buffer refers to a
>>local variable which probably no longer exists when append_rpid_hf() is
>>called.
>>
>>Apply the attached patch should solve the problem.
>>
>>
>>
>>>Otherwise, you can push a value from AVP list to a header using avpops
>>>module.
>>>
>>
>>The current avp functions can't manipulate the avp. So prefix and suffix
>>can't be added to the header field. Of course one can change the data in the
>>database, but it will not be backward compatible to the existing system.
>>
>>Richard
>>
> 
> 
> 
> 
>