Summary:ASTERISK-20783: segfault on hangup
Reporter:Birger "WIMPy" Harzenetter (wimpy)Labels:
Date Opened:2012-12-11 15:53:02.000-0600Date Closed:2012-12-31 15:04:45.000-0600
Versions:SVN Frequency of
Environment:Attachments:( 0) asterisk-20783-smoother-fed-check.diff
( 1) core.asterisk.1355261085
Description:This happened when switching off speakerphone with an active call and a 2nd call coming in.

Not sure if that's a general RTP issue or a chan_skinny thing.
Comments:By: Birger "WIMPy" Harzenetter (wimpy) 2012-12-11 15:53:50.910-0600


By: Michael L. Young (elguero) 2012-12-11 19:53:55.912-0600

Was there anything captured in the logs in regards to the smoother?

"Huh?  Can't smooth a non-voice frame!"
"Smoother was working on %s format frames, now trying to feed %s?"

I am looking at the code and am curious if it is due to us not checking the return code from ast_smoother_feed() in ast_smoother_write() before calling ast_smoother_read().  The len being -1235221544 in ast_smoother_read() doesn't look right since that would be used when calling memcpy().

By: Birger "WIMPy" Harzenetter (wimpy) 2012-12-11 20:15:40.573-0600

No occurrences of "smooth" in the logs at all.

By: Michael L. Young (elguero) 2012-12-11 21:56:17.584-0600

If you are able to reproduce this, give this patch a try and see if it stops the segfault.

By: Damien Wedhorn (wedhorn) 2012-12-12 01:26:24.202-0600

For info, fired up my 30vip, set up a call, sent another call in and hit the speaker button and worked as expected (ie first call ended, device started ringing for second call).

By: Birger "WIMPy" Harzenetter (wimpy) 2012-12-12 19:21:30.371-0600

I'm not able to reproduce the segfault.

By: Rusty Newton (rnewton) 2012-12-28 14:38:20.891-0600

Is anyone able to reproduce this issue? Or are the two last comments indicating that the patch fixes the issue?

By: Damien Wedhorn (wedhorn) 2012-12-28 15:53:42.483-0600

My comment was without the patch applied, so have no idea if the patch does anything.

By: Birger "WIMPy" Harzenetter (wimpy) 2012-12-29 13:42:19.922-0600

I wasn't able to reproduce the crash even without the patch.

By: Rusty Newton (rnewton) 2012-12-31 15:04:45.238-0600

Thanks for the responses. Closing out as "Cannot Reproduce". We are trying to avoid having too many one-off issues hanging out there when they can't be produced and we can't confirm they are a persistent issue.