bugGnash - The GNU Flash player - Bugs: bug #19978, testsuite failure: deadlock in...

 
 

bug #19978: testsuite failure: deadlock in NetStream-SquareTestRunner (ffmpeg)

Submitted by:  Bastiaan Jacques <bjacques>
Submitted on:  Sun 27 May 2007 11:11:41 AM UTC  
 
Category: videoSeverity: 5 - Blocker
Release: NoneStatus: Fixed
Privacy: PublicAssigned to: Sandro Santilli <strk>
Open/Closed: Closed

Add a New Comment(Rich Markup)
   

You are not logged in

Please log in, so followups can be emailed to you.

 

Tue 29 May 2007 09:33:06 PM UTC, comment #5:

Confirmed.

Bastiaan Jacques <bjacques>
Project Member
Tue 29 May 2007 05:18:18 PM UTC, comment #4:

It turns out that the never signalled condition variable was due to the sound_handler_test not using a separate thread like sound_handler_sdl does, which is expected in the NetStreamFfmpeg code.

I committed a patch dropping the sound_handler_test and adding the testing methods to the sound_handler base class directly (and implementing them in the _sdl and _gst versions).

With this I dont' get the deadlock anymore, not with the testcase at least.

We might close this and continue the discussion on the other bug item (youtube stops after XXX seconds) since I can still reproduce a deadlock with youtube offline setup, just not with the automated testcase (just luck maybe).

Bastiaan, do you confirm you don't get the deadlock anymore on make check ?

Sandro Santilli <strk>
Project MemberIn charge of this item.
Tue 29 May 2007 08:15:30 AM UTC, comment #3:

So, the deadlock (FFMPEG) is due to a condition variable never signalled (decoder_wait). Use of that conditional variable is also bogus, as it is associated to a mutex that nobody uses except
the thread that calls the wait() method on the condition variable.

I'd remove the wait as a whole, as we need to release shortly.
Doing so I found other deadlocks, which makes me really depressed :/

Sandro Santilli <strk>
Project MemberIn charge of this item.
Mon 28 May 2007 08:55:10 AM UTC, comment #2:

I committed a patch rounding numbers for the matrix test, please try again.

For the NetStream failure I confirm I can reproduce with ffmpeg,
but not with gstreamer. Are you also using ffmpeg ?

Sandro Santilli <strk>
Project MemberIn charge of this item.
Sun 27 May 2007 02:43:19 PM UTC, comment #1:

The NetStream deadlock would be a blocker, as for the matrix_test it would be interesting to see how does the proprietary player react to the SWF, being it a self-contained test...
Not a big failure, the latter...

Anonymous
Sun 27 May 2007 11:11:41 AM UTC, original submission:

Running ./../simple.exp ...
FAIL: matrix_testrunner: expected: '-30.05,-30.05 30.05,30.05' obtained: -30.0499992370605,-30.0499992370605 30.0499992370605,30.0499992370605 [matrix_test.c:120]
FAIL: matrix_testrunner: expected: '-30.05,-30.05 30.05,30.05' obtained: -30.0499992370605,-30.0499992370605 30.0499992370605,30.0499992370605 [matrix_test.c:134]
FAIL: matrix_testrunner: expected: '-30.05,-30.05 30.05,30.05' obtained: -30.0499992370605,-30.0499992370605 30.0499992370605,30.0499992370605 [matrix_test.c:148]
FAIL: Test case NetStream-SquareTestRunner still running after 240 seconds, killing it (deadlock?)

=== Summary ===

# of expected passes 1867
# of unexpected failures 4
# of expected failures 61

Platform is GNU/Linux amd64. Using Ming CVS, Mtasc and swfmill are also installed.

Bastiaan Jacques <bjacques>
Project Member

 

(Note: upload size limit is set to 16384 kB, after insertion of the required escape characters.)

Attach File(s):
   
   
Comment:
   

No files currently attached

 

Depends on the following items: None found

Items that depend on this one: None found

 

Carbon-Copy List
  • -unavailable- added by strk (Posted a comment)
  • -unavailable- added by bjacques (Submitted the item)
  •  

    Do you think this task is very important?
    If so, you can click here to add your encouragement to it.
    This task has 0 encouragements so far.

    Only logged-in users can vote.

     

    Please enter the title of George Orwell's famous dystopian book (it's a date):

     

     

    Follow 6 latest changes.

    Date Changed By Updated Field Previous Value => Replaced By
    Tue 29 May 2007 09:33:06 PM UTCbjacquesStatusReady For Test=>Fixed
      Open/ClosedOpen=>Closed
    Tue 29 May 2007 05:18:18 PM UTCstrkStatusNone=>Ready For Test
    Mon 28 May 2007 08:55:10 AM UTCstrkCategorytestsuite=>video
      Severity2 - Minor=>5 - Blocker
      SummaryTestsuite failures=>testsuite failure: deadlock in NetStream-SquareTestRunner (ffmpeg)

    Back to the top


    Powered by Savane 3.1-cleanup1