Summary:ASTERISK-20225: Segmentation Fault on manager_play_dtmf sip_senddigit_end
Reporter:Jeff Hoppe (jhoppebugs)Labels:
Date Opened:2012-08-13 14:30:35Date Closed:2013-04-10 09:07:00
Versions:10.6.1 10.8.0 10.10.0 Frequency of
Environment:Centos 5.7Attachments:( 0) backtrace11022012.txt
( 1) backtrace11142012.txt
( 2) backtrace80132012.txt
( 3) Debug.txt
( 4) fulllogsnippet.txt
Description:Program terminated with signal 11, Segmentation fault.
#0  0x04090bb4 in sip_senddigit_end (ast=0xb03ed00, digit=51 '3', duration=100) at chan_sip.c:6729
6729 switch (ast_test_flag(&p->flags[0], SIP_DTMF)) {
Comments:By: Matt Jordan (mjordan) 2012-08-13 15:29:55.524-0500

So, that's really weird.  There's no SIP dialog for that channel.

Do you mind getting a DEBUG log leading up to that point as well?  While we could certainly put something in to prevent the crash, it's possible that the crash in this location is a symptom of a much larger issue.

By: Jeff Hoppe (jhoppebugs) 2012-08-14 10:02:36.386-0500

Notice the ERROR right before the reboot.  This is not a debug but may be a clue.

Also, I searched the site and found nothing relating to my issue but I googled and got a file that you may be able to relate to a previous issue or request that is very similiar to my issue.


By: Jeff Hoppe (jhoppebugs) 2012-08-14 10:04:16.692-0500

I will try and get a DEBUG file, but this issue has only happened once so far.

By: Jeff Hoppe (jhoppebugs) 2012-08-30 10:53:27.370-0500

This happended again today.  I am now logging the DEBUG (as well as SIP DEBUG) to a file and the next time I get this error I will post the DEBUG information as requested.

This system is different than another one we have that does not get this error, and the difference is that in this system our agents VOIP soft phones register on a different Asterisk box which forwards the AGENTLOGIN to this Asterisk box.
I am not saying that this is an issue, but its the main difference between our two systems.

By: Sean Bright (seanbright) 2012-09-07 14:05:50.247-0500

I can confirm this with Asterisk 1.8 (r358484 which is painfully old but might help you narrow in on a commit).  BT is virtually identical - the sip_pvt is NULL.

By: Matt Jordan (mjordan) 2012-09-07 17:28:37.060-0500

Was the connection in that case TCP/TSL as well?

By: Sean Bright (seanbright) 2012-09-19 11:37:22.411-0500

Nope.  Good ol' unencrypted UDP.  If you meant the manager connection that sends the PlayDTMF action, that's just TCP, no TLS.

By: Jeff Hoppe (jhoppebugs) 2012-11-02 15:26:38.856-0500

This is the error on Asterisk 10.8.0 as well, attachment provided.

By: Jeff Hoppe (jhoppebugs) 2012-11-02 15:29:21.422-0500

From what I can see, this issue happens when my application sends a PLAYDTMF manager action to a channel that has already hung up.  When when I try and simulate this scenario, it just doesn't play DTMF tone.  There exists some window of opportunity that makes Asterisk crash in this scenario.

Here are last few lines of full log:

2:39:18 PM -- SIP/hq-ast-006-trunk-00016ec1 answered Agent/1290                                
2:41:02 PM -- Executing |h@manual-outbound:1| NoOp("Agent/1290", "Hangup Extension: ANSWER 1351885150.166414 Agent/1290 FREELANCE") in new stack                                                                                                                        
2:41:02 PM -- Executing |h@manual-outbound:2| UserEvent("Agent/1290", "MANOUTHANGUP,Channel:SIP/hq-ast-006-trunk/6097500777,DialStatus:ANSWER,CPLSFE:FREELANCE,AgentID:1290,ProgramID:10001505,CallerID:8554204495") in new stack                                                                                                                                                                                                                  
2:41:02 PM Hard hangup called by thread -1243485296 on SIP/hq-ast-006-trunk-00016ec1, while fd is blocked by thread -1234781296 in procedure ast_waitfor_nandfds!  Expect a failure                                                                                      

By: Jeff Hoppe (jhoppebugs) 2012-11-16 10:50:40.505-0600

Here is the debug information, for backtrace11142012.txt    
Let me know if you need to go farther back in the debug log.  I just went back to form a 1 meg file.

By: Jeff Hoppe (jhoppebugs) 2012-11-16 10:53:02.557-0600

Will you let me know what revision code this eventually goes under so I can keep an eye on it in the change log for future releases?

By: Matt Jordan (mjordan) 2012-11-20 14:51:52.624-0600

When an issue is committed, the revision it goes in is tracked under the "Subversion" tab.

By: Jeff Hoppe (jhoppebugs) 2012-11-27 10:20:24.912-0600

Is the debug file that you asked for and uploaded sufficient to figure out the issue or do you need more?

By: Jeff Hoppe (jhoppebugs) 2012-12-17 12:17:41.316-0600

I have another issue in the bug tracker that I am hearing feed back from and getting nothing since September on this one. I was wondering if this one is on the radar or if there is something I did not do correctly on giving the feedback you asked for.

By: Matt Jordan (mjordan) 2013-04-09 12:10:47.386-0500

Sorry for not getting back - I must have missed your last comment on this issue. The issue has been assigned (to me, no less :-)) and should get worked soon.

By: Jeff Hoppe (jhoppebugs) 2013-12-06 14:17:30.006-0600

I see that this issue is resolved for 1.8 and 11 but is it possible to get a patch for Asterisk 10?   We tried moving to Asterisk 11 (and would like to get there eventually) but there are some other issues and  bugs in Asterisk 11 for us.    Asterisk 10 has been a perfect fit with our system, other than this particular crash.

By: Matt Jordan (mjordan) 2013-12-06 15:49:26.986-0600

Asterisk 10 no longer receives bug fixes. It will soon be out of the Security Fix window as well.

You can attempt to backport the patch and apply it yourself, but it will not be committed to that branch.

{{svn diff -c 385173 http://svn.asterisk.org/svn/branches/11 > ASTERISK-20225.patch}}