Re: [SR-Users] CRASH on qm_free for simultaneous calls (V3.1.0)

2010-12-20 Thread Jijo
Hi Daniel, Some more detail to this issue.. if we call t_release after failure route with append branch, a double free causes the crash because the transaction will be free again after the t_relay of ACK message. This happens because we compile the code with DBG_QM_MALLOC. wt_timer keeps the

Re: [SR-Users] CRASH on qm_free for simultaneous calls (V3.1.0)

2010-12-17 Thread Daniel-Constantin Mierla
Hello, On 12/17/10 12:58 AM, Jijo wrote: Hi Daniel.. This was due to an error in our config file. We were doing t_release and then t_relay. So we removed the t_release from config, which basically fixed the issue. thanks for further troubleshooting and reporting, I will check it and try to

[SR-Users] CRASH on qm_free for simultaneous calls (V3.1.0)

2010-12-15 Thread Jijo
I'm observing a crash on qm_free, when we made two outgoing calls to the same number. One SIP phone receives ring back tone and the other busy tone. The back trace is shown below which shows that it is crashed in abort(). How can the condition if (f-u.is_free){ could be true? Please let me know

Re: [SR-Users] CRASH on qm_free for simultaneous calls (V3.1.0)

2010-12-15 Thread Daniel-Constantin Mierla
Hello, look into syslog file (/var/log/syslog or /var/log/messages if you didn't set a custom one) and search for a message like BUG: qm_free: freeing already freed pointer . Send it here, seems to be a double free of same pointer. Thanks, Daniel On 12/16/10 12:34 AM, Jijo wrote: I'm