|Summary:||[Chromium] videos are not painted for compositing/ tests|
|Product:||WebKit||Reporter:||Mihai Parparita <mihai>|
|Component:||Layout and Rendering||Assignee:||Victoria Kirst <vrk>|
|Severity:||Normal||CC:||commit-queue, kbr, vangelis, vrk|
|Version:||528+ (Nightly build)|
|Bug Depends on:||55269|
Description Mihai Parparita 2011-02-17 15:46:42 PST
The following tests all fail with new-run-webkit-tests --platform=chromium-gpu (once bug 47231 is fixed) with image diffs on the Mac (haven't verified other platforms yet): compositing/geometry/clipped-video-controller.html compositing/geometry/video-fixed-scrolling.html compositing/geometry/video-opacity-overlay.html compositing/layers-inside-overflow-scroll.html compositing/overflow/overflow-compositing-descendant.html compositing/overflow/scroll-ancestor-update.html compositing/reflections/load-video-in-reflection.html compositing/self-painting-layers.html compositing/self-painting-layers2.html compositing/video/video-background-color.html compositing/video/video-poster.html The video does not appear to get displayed when run under the DRT (w/ Mesa), though it does get displayed when run under Chrome.
Comment 1 Mihai Parparita 2011-02-22 13:19:19 PST
Victoria, these are the tests that I was talking about in the testing meeting. The expected result is to have a video in the clipped rectangle: http://svn.webkit.org/repository/webkit/trunk/LayoutTests/platform/mac/compositing/geometry/clipped-video-controller-expected.png But though we do get pixel results, the video is not displayed: http://build.chromium.org/f/chromium/layout_test_results/Webkit_Win_-_GPU/results/layout-test-results/compositing/geometry/clipped-video-controller-actual.png The dashboard has results from the other bots: http://test-results.appspot.com/dashboards/flakiness_dashboard.html#showExpectations=true&group=%40ToT%20GPU%20-%20chromium.org&tests=compositing%2Fgeometry%2Fclipped-video-controller.html
Comment 2 Victoria Kirst 2011-02-23 17:00:24 PST
Hi Mihai! I am having a similar problem with some of our media tests. I am imagining they are the same problem. Once I figure out what's going on in media land, I will see if the same thing is happening here.
Comment 3 Victoria Kirst 2011-02-24 18:06:42 PST
OK, figured out the problem. VideoLayerChromium uses mapTexSubImage2DCHROMIUM to update textures and that is not supported in MESA it seems (or it fails anyway). There is a TODO to have an appropriate fall back when this fails, but TODO wasn't implemented :) But if I fall back to normal texSubImage2D, videos are painted with DRT and Mesa! I'll upload that patch and rebaselined images.
Comment 5 Kenneth Russell 2011-03-01 15:50:33 PST
Comment on attachment 84311 [details] Patch Looks good to me.
Comment 6 WebKit Commit Bot 2011-03-01 21:50:28 PST
Comment on attachment 84311 [details] Patch Clearing flags on attachment: 84311 Committed r80091: <http://trac.webkit.org/changeset/80091>
Comment 7 WebKit Commit Bot 2011-03-01 21:50:35 PST
All reviewed patches have been landed. Closing bug.