Summary: | [chromium] Resizing a 2d canvas to huge dimensions after compositing crashes with accelerated 2d canvas option enabled | ||||||
---|---|---|---|---|---|---|---|
Product: | WebKit | Reporter: | James Robinson <jamesr> | ||||
Component: | New Bugs | Assignee: | James Robinson <jamesr> | ||||
Status: | RESOLVED FIXED | ||||||
Severity: | Normal | CC: | alokp, enne, junov, kbr, senorblanco, twiz, vangelis | ||||
Priority: | P2 | ||||||
Version: | 528+ (Nightly build) | ||||||
Hardware: | Other | ||||||
OS: | OS X 10.5 | ||||||
Attachments: |
|
Description
James Robinson
2011-05-02 15:13:13 PDT
Created attachment 91991 [details]
Patch
This crash is showing up with moderate frequency in recent Chrome releases. It only shows up with the --enable-accelerated-2d-canvas flag on but can be triggered by doing something as simple as resizing the window with a fullscreen canvas demo opened. Comment on attachment 91991 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=91991&action=review > Source/WebCore/html/canvas/CanvasRenderingContext2D.cpp:207 > +#endif senorblanco should review this. He just added a short-circuit to DrawingBuffer::reset() to do no work if the size hasn't changed, and I have a feeling that making this contentChanged call in all cases may impact performance. Comment on attachment 91991 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=91991&action=review Looks good (unofficially). >> Source/WebCore/html/canvas/CanvasRenderingContext2D.cpp:207 >> +#endif > > senorblanco should review this. He just added a short-circuit to DrawingBuffer::reset() to do no work if the size hasn't changed, and I have a feeling that making this contentChanged call in all cases may impact performance. I think this is correct. Valid size changes are unaffected; this only changes the case when we're going from valid -> invalid size, in which case we won't be rendering Canvas2D on the GPU anyway and we should definitely tell the compositor. *** Bug 60056 has been marked as a duplicate of this bug. *** Comment on attachment 91991 [details]
Patch
OK, sounds good.
Comment on attachment 91991 [details] Patch Clearing flags on attachment: 91991 Committed r85661: <http://trac.webkit.org/changeset/85661> All reviewed patches have been landed. Closing bug. |