Bug 70641 – Layout Test fast/frames/sandboxed-iframe-scripting.html is flaky

• Status: RESOLVED
• Resolution: WORKSFORME
• Component: Tools / Tests

[Julien Chaffraix]

The following layout test is flaky on snow-leopard:

fast/frames/sandboxed-iframe-scripting.html

Probable cause:

There is an infinite loop in toV8Event.

[Julien Chaffraix]

Created attachment 112014 [details]
Crash dump

[Adam Barth]

This was a bug, but I fixed it.  Is this still happening?

[Julien Chaffraix]

(In reply to comment #2)
> This was a bug, but I fixed it.  Is this still happening?

I saw it on the bot this morning and it was suggested by garden-o-matic. Unfortunately it does not appear anymore as we silenced it.

[Adam Barth]

I think there's a bug with RWT picking up old crash dumps.  It's unlikely the crash exists anymore.  If it recurs, you might check whether the date changes:

Date/Time:       2011-10-20 18:06:24.036 -0700

[Julien Chaffraix]

(In reply to comment #4)
> I think there's a bug with RWT picking up old crash dumps.  It's unlikely the crash exists anymore.  If it recurs, you might check whether the date changes:
> 
> Date/Time:       2011-10-20 18:06:24.036 -0700

FYI there is another test that is showing the same stack-trace for 2 bots: http://test-results.appspot.com/dashboards/flakiness_dashboard.html#showExpectations=true&tests=storage%2Fdomstorage%2Fevents%2Fbasic-body-attribute.html

I downloaded the test_results.zip from the bot and the _modify_ stamp is:

Date/Time: 2011-10-21 15:34:27.000000000 -0700

This is before the time I extracted the archive but it's possible that the script touched the file so no conclusion can be drawn.

[Adam Barth]

> > Date/Time:       2011-10-20 18:06:24.036 -0700

From:

http://build.chromium.org/f/chromium/layout_test_results/Webkit_Mac10_6__dbg_/results/layout-test-results/storage/domstorage/events/basic-body-attribute-crash-log.txt

Date/Time:       2011-10-20 18:06:24.036 -0700

Notice that the date in the crash log is the same as the previous.

The tool is just picking up old crash logs.  It's entirely like that this test is crashing sporadically, but it's not crashing with that stack.

[Julien Chaffraix]

(In reply to comment #6)
> > > Date/Time:       2011-10-20 18:06:24.036 -0700
> 
> From:
> 
> http://build.chromium.org/f/chromium/layout_test_results/Webkit_Mac10_6__dbg_/results/layout-test-results/storage/domstorage/events/basic-body-attribute-crash-log.txt
> 
> Date/Time:       2011-10-20 18:06:24.036 -0700
> 
> Notice that the date in the crash log is the same as the previous.
> 
> The tool is just picking up old crash logs.  It's entirely like that this test is crashing sporadically, but it's not crashing with that stack.

Ah, I looked at the wrong date. This means that this bug is very likely invalid.

[Julien Chaffraix]

> Ah, I looked at the wrong date. This means that this bug is very likely invalid.

Just following-up, Dimitri showed me that the bots had some issue uploading the new stacktraces to the server on Friday and that the test did crash for real (NRWT does not check the stacktrace to mark a test as crashing).

Also the bots seems more stable today and does not crash so I may have been too eager by marking the test as crashing. See http://test-results.appspot.com/dashboards/flakiness_dashboard.html#tests=fast%2Fframes%2Fsandboxed-iframe-scripting.html

[Ryosuke Niwa]

No longer observing this on bots.