See https://bugs.webkit.org/show_bug.cgi?id=33416 This random failure might be caused by a hardware/driver bug, likely some unwanted competition is going on.
Observation (not sure if it's of any value): I used to be able to reproduce this failure on my Mac the first run after I just rebuilt the webkit, but not any later runs; right now I am unable to reproduce it on my machine anymore.
Another report of the same issue. http://lists.apple.com/archives/mac-opengl/2009/Dec/msg00010.html
If this is a bug in Mac OS X software, you need to report it at http://radar.apple.com/ to have it looked at. This tracker is only for WebKit bugs.
Yes, we will file a bug there.
Seems like the issue is not hardware/driver bug, but the uninitialized variables in Mac port. See https://bugs.webkit.org/show_bug.cgi?id=36908 Going to turn on anti-alias / stencil again as default canvas attributes for tests.
Created attachment 54012 [details] patch Anti-alias and stencil were turned off from default canvas attribute set when we try to pin-point the cause of these random failures; now seems like they are not the cause, we turn them back on.
Comment on attachment 54012 [details] patch OK.
Comment on attachment 54012 [details] patch Clearing flags on attachment: 54012 Committed r58090: <http://trac.webkit.org/changeset/58090>
All reviewed patches have been landed. Closing bug.