[SR-Users] segfault in siptrace module using setflag?

[Dragos Dinu]

Hi Samuel,

Could you try to test again with the updated siptrace sources from 3.3 
branch? I had a patch recently commited for a crash in siptrace, so I'm 
pretty sure you have the same issue.
Please tell us if your problem was fixed.:)

Dragos

On 10/09/2012 12:00 PM, samuel wrote:
> Hi folks,
>
> I was making some tests with kamailio (both 3.3.0 and 3.3.1 tgz) with 
> siptrace module sending "homer" packets to a remote central agent 
> where the data is stored and presented to the user.
>
> The related config is:
> # ----- siptrace params from HOMER -----
> modparam("siptrace", "duplicate_uri", "sip:A.B.C.D:9060")
> modparam("siptrace", "trace_to_database", 0)
> modparam("siptrace", "trace_flag", 22)
> modparam("siptrace", "trace_on", 1)
> #HEPv2 == timestamp will be included to HEP header
> modparam("siptrace", "hep_version", 2)
> modparam("siptrace", "hep_mode_on", 1)
> modparam("siptrace", "hep_capture_id", 1)
>
> basically for all incoming messages, a route is called with the 
> following actions:
>
> route[HOMER] {
>         setflag(22);
>         sip_trace();
> }
>
> I've been making some tests and if I disable the setflag action, there 
> is no problem (well, no transaction messages are sent to the central 
> agent), but as soon as I activated setflag, kamailio crashes with the 
> following error:
>
> <core> [pass_fd.c:293]: ERROR: receive_fd: EOF on 28
>
> There's no concrete point in the config file where it fails so I think 
> the problem is more tm-internals related. I can provide further 
> information in case you need it.
>
> Just for adding information, and without any intention to start a 
> flame-war, with opensips 1.8.0-tls the server does not crash...
>
> Thanks in advance!
> Samuel.