Summary: | [chromium] Layout test fast/canvas/setWidthResetAfterForcedRender.html failing on GPU after r93988 | ||||||
---|---|---|---|---|---|---|---|
Product: | WebKit | Reporter: | Stephen White <senorblanco> | ||||
Component: | Canvas | Assignee: | Stephen White <senorblanco> | ||||
Status: | RESOLVED DUPLICATE | ||||||
Severity: | Normal | CC: | jamesr, kbr, levin, mdelaney7 | ||||
Priority: | P2 | ||||||
Version: | 528+ (Nightly build) | ||||||
Hardware: | Unspecified | ||||||
OS: | Unspecified | ||||||
Attachments: |
|
Description
Stephen White
2011-08-29 11:42:19 PDT
Created attachment 105515 [details]
Patch
Comment on attachment 105515 [details]
Patch
Never mind; this fix didn't work. :(
This test is still really flaky FYI: even if it were "fixed" back to pre-93988 behaviour, I don't think the pixel results will be correct, since LayoutTestController::display() doesn't really handle the repaint mask correctly for composited layers. When the compositor is enabled, WebViewHost::displayRepaintMask() probably needs to paint into the composited layer rather than the framebuffer. See https://bugs.webkit.org/show_bug.cgi?id=54561 for reference. |