[asterisk-dev] [Code Review] 3255: testsuite: chan_sip ice crash test for ASTERISK-22911

Jonathan Rose reviewboard at asterisk.org
Fri Feb 28 14:30:12 CST 2014


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviewboard.asterisk.org/r/3255/
-----------------------------------------------------------

(Updated Feb. 28, 2014, 2:30 p.m.)


Review request for Asterisk Developers, Joshua Colp, Kevin Harwell, and Matt Jordan.


Changes
-------

Add checks to this patch to make sure that ICE check lists are built successfully on two separate occasions during the call. Fail on ICE check list creation failure.

Test events are added to supply these checks in the following patch:
https://reviewboard.asterisk.org/r/3275/

With those test events added to 11 currently without the rest of the patch, this test will fail because the check list will only be built once.
11.7 will fail this test because the check list will never be built successfully and Asterisk will crash.
My previous patch for ASTERISK-22911 would have failed this test because while the check lists would be built, they would fail on the pruning phase.

With the patch from the above review completely in place, this test is successful.

I'm currently working on a chan_pjsip equivalent to this test.


Bugs: ASTERISK-22911
    https://issues.asterisk.org/jira/browse/ASTERISK-22911


Repository: testsuite


Description
-------

Tests for a crash occurring in 11.7 when ICE on SIP calls (reproduced from a SIPML5 with 6 total candidate fields in SDP) when holding. The crash is caused by an initially failed ICE session startup followed by a second attempt at doing the startup when a HOLD is received. The crash was resolved in 11.8, but in the process some people lost audio at the start of similar calls that previously worked in 11.7

The test itself is a near duplicate of the existing SIP hold tests, but slightly simplified and with some changes in the call flow. A starts the call and engages hold and unhold. At the conclusion of unhold if all MOH events went through the test will be considered successful.


Diffs (updated)
-----

  /asterisk/trunk/tests/channels/SIP/tests.yaml 4726 
  /asterisk/trunk/tests/channels/SIP/sip_hold_ice/test-config.yaml PRE-CREATION 
  /asterisk/trunk/tests/channels/SIP/sip_hold_ice/sipp/phone_B.xml PRE-CREATION 
  /asterisk/trunk/tests/channels/SIP/sip_hold_ice/sipp/phone_A.xml PRE-CREATION 
  /asterisk/trunk/tests/channels/SIP/sip_hold_ice/sipp/inject_bridge.csv PRE-CREATION 
  /asterisk/trunk/tests/channels/SIP/sip_hold_ice/run-test PRE-CREATION 
  /asterisk/trunk/tests/channels/SIP/sip_hold_ice/configs/ast1/sip.conf PRE-CREATION 
  /asterisk/trunk/tests/channels/SIP/sip_hold_ice/configs/ast1/rtp.conf PRE-CREATION 
  /asterisk/trunk/tests/channels/SIP/sip_hold_ice/configs/ast1/extensions.conf PRE-CREATION 

Diff: https://reviewboard.asterisk.org/r/3255/diff/


Testing
-------

Ran test against 11.7. Crashed
Ran test against 11.8. Did not crash.

Ran test against a diagnostic patch I made to check the PJ_NATH errors that were occurring to make sure everything mirrored the results I had been seeing in my reproduction efforts.


Thanks,

Jonathan Rose

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.digium.com/pipermail/asterisk-dev/attachments/20140228/0da6b224/attachment.html>


More information about the asterisk-dev mailing list