|Summary:||ASTERISK-17681: Deadlock between rtp_engine.c (ast_rtp_instance_early_bridge) and chan_sip (handle_incoming)|
|Reporter:||Michel Verbraak (astmiv)||Labels:|
|Date Opened:||2011-04-12 10:22:28||Date Closed:||2011-05-23 09:47:28|
|Environment:||Attachments:||( 0) backtrace_for_lock2.txt|
( 1) backtrace4.txt
( 2) lock2.txt
( 3) locks.txt
|Description:||Deadlock situation. See locks.txt|
Produced this with a test setup of asterisk 126.96.36.199 and sipp.
|Comments:||By: Michel Verbraak (astmiv) 2011-04-12 10:24:39|
also attached backtrace file.
By: Michel Verbraak (astmiv) 2011-04-12 11:20:42
Added second deadlock files (lock2.txt and backtrace_for_lock2.txt)
By: Gregory Hinton Nietsky (irroot) 2011-04-13 04:43:19
By: Michel Verbraak (astmiv) 2011-04-13 11:00:48
Created a test patch see https://reviewboard.asterisk.org/r/1177/
By: Private Name (falves11) 2011-04-14 22:10:16
Version 1.6.2 has the same issue. We started using for a Dialer application, and after 10-15 mins Asterisk locked up. It could not even be killed with killall, just with kill -9. The application generates 150 channels if calls and plays a message when somebody answers the phone. We had to go back 1.4 and it works perfectly.
By: Private Name (falves11) 2011-04-14 22:11:20
If somebody creates a patch for 1.6.2 I will go back and try again.
By: Leif Madsen (lmadsen) 2011-05-19 14:03:10
It's likely this has already been resolved in Asterisk 1.8. I have related issue ASTERISK-17891 which has a patch attached that may resolve it in a different way for Asterisk 1.4 and 1.6.2.
By: Leif Madsen (lmadsen) 2011-05-23 09:47:28
I'm closing this issue per a previous note that this should already be resolved in the Asterisk 1.8 branch. Thanks!