RESOLVED FIXED 87913
[chromium] Work around lost context assertion failures in compositor startup 
https://bugs.webkit.org/show_bug.cgi?id=87913
Summary [chromium] Work around lost context assertion failures in compositor startup
Kenneth Russell
Reported 2012-05-30 18:14:26 PDT
In order to fully test and backport the fix for Chromium bug http://code.google.com/p/chromium/issues/detail?id=121780 , a workaround is needed for the several compositor assertion failures seen when rendering a frame with a lost context. This workaround will be very small, amounting to checking for a lost context event at the beginning of compositing rather than just at the end, and can be removed later.
Attachments
Patch (7.43 KB, patch)
2012-05-30 20:05 PDT, Kenneth Russell 		jamesr: review-
webkit.review.bot: commit-queue-
DetailsFormatted DiffDiff
Archive of layout-test-results from ec2-cr-linux-02 (533.13 KB, application/zip)
2012-05-31 01:10 PDT, WebKit Review Bot 		no flags
Details
View All Add attachment proposed patch, testcase, etc.
Kenneth Russell
Comment 1 2012-05-30 20:05:21 PDT
Created attachment 144976 [details] Patch
WebKit Review Bot
Comment 2 2012-05-31 01:10:53 PDT
Comment on attachment 144976 [details] Patch Attachment 144976 [details] did not pass chromium-ews (chromium-xvfb): Output: http://queues.webkit.org/results/12864139 New failing tests: platform/chromium/compositing/webgl-loses-compositor-context-2.html
WebKit Review Bot
Comment 3 2012-05-31 01:10:57 PDT
Created attachment 145012 [details] Archive of layout-test-results from ec2-cr-linux-02 The attached test failures were seen while running run-webkit-tests on the chromium-ews. Bot: ec2-cr-linux-02 Port: <class 'webkitpy.common.config.ports.ChromiumXVFBPort'> Platform: Linux-2.6.35-28-virtual-x86_64-with-Ubuntu-10.10-maverick
James Robinson
Comment 4 2012-05-31 10:20:33 PDT
Comment on attachment 144976 [details] Patch Test failure needs to be looked at. It's one thing for the failure to not show up in DRT without a patch, it's another for the test to fail in DRT. Why does the expected png appear to have red boxes in it? It looks like this doesn't really solve any problems, just moves things around - we can lose the context at any point and need to be robust to that.
James Robinson
Comment 5 2012-05-31 10:20:59 PDT
Do you expect that this patch will have any impact on release builds, or are you just seeing debug-only ASSERT()s without it?
James Robinson
Comment 6 2012-05-31 15:32:36 PDT
https://bugs.webkit.org/show_bug.cgi?id=87912 has a patch that fixes the ASSERT()s to take a context loss into account and has a much tighter test. I don't think we need any other changes on the compositor side. I don't think this workaround provides any additional value. If other systems (like ganesh) are behaving badly with lost context we should definitely investigate and fix any issues that come up.
James Robinson
Comment 7 2012-05-31 15:46:55 PDT
Committed r119160: <http://trac.webkit.org/changeset/119160>
Note You need to log in before you can comment on or make changes to this bug.