[asterisk-bugs] [JIRA] (ASTERISK-26669) PJSIP Segfault 13.13.1 (Bundled PJSIP)

[Richard Mudgett (JIRA)]

     [ https://issues.asterisk.org/jira/browse/ASTERISK-26669?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Richard Mudgett updated ASTERISK-26669:
---------------------------------------

    Attachment: jira_asterisk_26669_v13_tdata_trace.patch

[^jira_asterisk_26669_v13_tdata_trace.patch] - This patch adds some tdata ref tracing messages to pjproject.  It should be added in addition to the other patches you have installed.

The problem is somewhere in pjproject and it appears to be an extra unref of a tdata OPTIONS request message.  It has something to do with DNS resolution and how the messages are sent.  How it is triggered I still don't know and I haven't been able to reproduce it.

To apply:
# In the Asterisk source root directory {{patch -p1 -i jira_asterisk_26669_v13_tdata_trace.patch}}
# Use the project distclean method described in an earlier comment.
# Recompile and install.

To collect the needed debug data in addition to how you have collected the previous backtraces and Asterisk debug data enable {{pjproject set log level 5}}.  Optionally, you can see {{pjproject.conf.sample}} for an option to start the logging immediately on Asterisk start.  Adding the pjproject logging output will allow following tdata buffer ref changes and the DNS resolution steps.

The last six or so backtraces have all been in the same place.

Thanks

> PJSIP Segfault 13.13.1 (Bundled PJSIP)
> --------------------------------------
>
>                 Key: ASTERISK-26669
>                 URL: https://issues.asterisk.org/jira/browse/ASTERISK-26669
>             Project: Asterisk
>          Issue Type: Bug
>      Security Level: None
>          Components: pjproject/pjsip
>    Affects Versions: 13.13.1
>         Environment: Ubuntu 16.04.1 LTS, Linux 4.4.0-53-generic
>            Reporter: Nic Colledge
>            Assignee: Richard Mudgett
>         Attachments: backtrace-1482444465.txt, backtrace-1482482153.txt, backtrace-1482483920.txt, backtrace-1482496599.txt, backtrace-1482497243.txt, backtrace-1482498503.txt, backtrace.30388.ast3.1486008781-0413.txt, backtrace.30588.ast3.1486010460-0441.txt, backtrace.30719.ast3.1486015498-0604.txt, backtrace.36281.ast3.1486130000-1353.txt, backtrace.36368.ast3.1486140553-1649.txt, backtrace.core.asterisk.1107.ast3.1486474718-1338.txt, backtrace.core.asterisk.1215.ast3.1486495028-1917.txt, backtrace.core.asterisk.3079.ast3.1486566672-1511.txt, backtrace.core.asterisk.6222.ast3.1486609431-0303.txt, debug_log_0303.txt, debug_log_0413.txt, debug_log_0441.txt, debug_log_0604.txt, debug_log_1338.txt, debug_log_1353.txt, debug_log_1511.txt, debug_log_1649.txt, debug_log_1917.txt, jira_asterisk_26669_v13_tdata_trace.patch, jira_asterisk_26669_v13_test.patch
>
>
> Hi guys,
> Been getting an intermittent segfault with asterisk 13.13.1 since we upgraded. Its a low-load server but happens maybe a few times a day. 
> We jumped a few versions from 13.10 so not sure if this is a last-release regression or not.
> We have another higher-load server (with the same setup) that has not yet crashed in the same way.
> I will attach backtraces; the fault its self comes from pj_atomic_dec_and_get from /usr/lib/libasteriskpj.so



--
This message was sent by Atlassian JIRA
(v6.2#6252)