[asterisk-dev] [Code Review] 4382: stasis bridge: handle early hangup of swap channel

[Scott Griepentrog]

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviewboard.asterisk.org/r/4382/
-----------------------------------------------------------

(Updated Jan. 29, 2015, 11:20 a.m.)


Review request for Asterisk Developers.


Changes
-------

Addressed Richard's findings.


Bugs: ASTERISK-24649
    https://issues.asterisk.org/jira/browse/ASTERISK-24649


Repository: Asterisk


Description
-------

During an attended transfer of a channel in stasis, where a local channel is used to replace (swap in for) a PJSIP channel, there is a race condition.  The PJSIP REFER transfer has been initiated (ast_bridge_impart) but will be completed later on another thread (bridge_channel_ind_thread), however, prior to the stasis bridging callback (bridge_stasis_push) being called the PJSIP channel is hungup, which allows the stasis app loop to exit, deleting the stasis control.  That prevents the stasis bridge callback from getting the app name from the channel being replaced (swap).

This patch adds a new stasis bridge callback push_peek, which is called from the ast_bridge_impart() thread.  This allows the new bridge_stasis_push_peek() function to copy the stasis app name from the originating channel before the PJSIP channel can be hungup.


Diffs (updated)
-----

  /branches/13/res/stasis/stasis_bridge.c 431296 
  /branches/13/main/bridge.c 431296 
  /branches/13/include/asterisk/bridge.h 431296 

Diff: https://reviewboard.asterisk.org/r/4382/diff/


Testing
-------

Running tests/rest_api/external_interaction/attended_transfer/stasis_bridge_to_stasis_app now does not result in an occasional failure.


Thanks,

Scott Griepentrog

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.digium.com/pipermail/asterisk-dev/attachments/20150129/8f9e3b76/attachment.html>