Bug 29037 - media/controls-after-reload.html crashes occasionally
Summary: media/controls-after-reload.html crashes occasionally
Status: RESOLVED FIXED
Alias: None
Product: WebKit
Classification: Unclassified
Component: Layout and Rendering (show other bugs)
Version: 528+ (Nightly build)
Hardware: PC OS X 10.5
: P2 Normal
Assignee: Nobody
URL:
Keywords:
Depends on: 28845
Blocks:
  Show dependency treegraph
 
Reported: 2009-09-08 09:06 PDT by Eric Seidel (no email)
Modified: 2009-09-22 15:51 PDT (History)
3 users (show)

See Also:


Attachments
Stack trace from 9/3/09, while trying to land bug 28406 (56.55 KB, text/plain)
2009-09-08 09:06 PDT, Eric Seidel (no email)
no flags Details
Seen again on 9/08/09 with different stack trace while trying to land bug 28934. (46.32 KB, text/plain)
2009-09-08 10:02 PDT, Eric Seidel (no email)
no flags Details
Hang report from local run-webkit-tests controls-after-reload.html run 9/22/09 (17.96 KB, text/plain)
2009-09-22 14:54 PDT, Eric Seidel (no email)
no flags Details
Patch v1 (1.11 KB, patch)
2009-09-22 14:57 PDT, Eric Seidel (no email)
no flags Details | Formatted Diff | Diff
crashlog for posterity (64.01 KB, application/octet-stream)
2009-09-22 15:51 PDT, Eric Seidel (no email)
no flags Details

Note You need to log in before you can comment on or make changes to this bug.
Description Eric Seidel (no email) 2009-09-08 09:06:09 PDT
Created attachment 39192 [details]
Stack trace from 9/3/09, while trying to land bug 28406

media/controls-after-reload.html crashes occasionally

This is likely the same root cause as bug 28845, but since the stack trace looks different, I'm filing a separate bug.
Comment 1 Eric Seidel (no email) 2009-09-08 09:55:54 PDT
Hum... I wonder why bugzilla fails to detect crash-reports as text/plain. Maybe they have a BOM.
Comment 2 Eric Seidel (no email) 2009-09-08 10:02:08 PDT
Created attachment 39195 [details]
Seen again on 9/08/09 with different stack trace while trying to land bug 28934.
Comment 3 Eric Seidel (no email) 2009-09-09 16:19:34 PDT
I'm not sure the second stack trace is related.  But looking at these stack traces and hang reports from the other bugs, it looks like that the CVDisplayLink is deallocated already.  Possibly something further up the stack as well.
Comment 4 Eric Carlson 2009-09-09 17:00:32 PDT
I don't know about the CVDisplayLink, but it is *definitely* not normal to see 15 threads blocked in JVTCompEncodeFrame.
Comment 5 Eric Seidel (no email) 2009-09-18 15:18:49 PDT
Looks like we're still seeing this one:
http://build.webkit.org/results/Leopard%20Intel%20Release%20(Tests)/r48536%20(5140)/results.html

So bug 28845's workaround was probably not enough.  I don't have a stack trace from the buildbot sadly.
Comment 6 Eric Seidel (no email) 2009-09-21 18:52:02 PDT
This one is still crashing on the bots:
http://build.webkit.org/results/Leopard%20Intel%20Release%20(Tests)/r48615%20(5210)/results.html
Comment 7 Eric Seidel (no email) 2009-09-22 14:26:59 PDT
WebKitTools/Scripts/run-webkit-tests --iterations 100 media/controls-after-reload.html

Is 100% reproducible failure for me.  If Simon/Eric thinks this is the same issue as was seen in bug 29037 then we should just skip this test on Leopard.
Comment 8 Simon Fraser (smfr) 2009-09-22 14:29:51 PDT
Got a crash log?
Comment 9 Eric Seidel (no email) 2009-09-22 14:48:47 PDT
(In reply to comment #8)
> Got a crash log?

The failures I'm seeing today are timeouts.  I'll run it again and see if I can get you an up-to-date crash logs.  There are some crash logs already attached to this bug too.
Comment 10 Eric Seidel (no email) 2009-09-22 14:54:25 PDT
Created attachment 39947 [details]
Hang report from local run-webkit-tests controls-after-reload.html run 9/22/09
Comment 11 Simon Fraser (smfr) 2009-09-22 14:55:09 PDT
That's the same as the other CVDisplayLink issue. This bug should just be duped.
Comment 12 Eric Seidel (no email) 2009-09-22 14:57:40 PDT
Created attachment 39949 [details]
Patch v1
Comment 13 Eric Seidel (no email) 2009-09-22 14:58:11 PDT
Ok, posted a patch to skip this test on Leopard for the same reason as the other one.
Comment 14 WebKit Commit Bot 2009-09-22 15:33:40 PDT
Comment on attachment 39949 [details]
Patch v1

Clearing flags on attachment: 39949

Committed r48654: <http://trac.webkit.org/changeset/48654>
Comment 15 WebKit Commit Bot 2009-09-22 15:33:44 PDT
All reviewed patches have been landed.  Closing bug.
Comment 16 Eric Seidel (no email) 2009-09-22 15:51:07 PDT
Entertaining that the bot went out with a bang:
http://build.webkit.org/results/Leopard%20Intel%20Release%20(Tests)/r48653%20(5248)/results.html
That's the last revision before the checkin to skip this test. :)

I also have a crash log too which I'll upload for posterity.
Comment 17 Eric Seidel (no email) 2009-09-22 15:51:36 PDT
Created attachment 39955 [details]
crashlog for posterity