Bug 32830

Summary: svg/dynamic-updates/SVGMaskElement-svgdom-height-prop.html crashed on Windows Test Bot
Product: WebKit Reporter: Eric Seidel (no email) <eric>
Component: Tools / TestsAssignee: Nobody <webkit-unassigned>
Status: RESOLVED FIXED    
Severity: Normal CC: krit, zimmermann
Priority: P2    
Version: 528+ (Nightly build)   
Hardware: PC   
OS: OS X 10.5   

Eric Seidel (no email)
Reported 2009-12-21 11:14:22 PST
svg/dynamic-updates/SVGMaskElement-svgdom-height-prop.html crashed on Windows Test Bot http://build.webkit.org/results/Windows%20Release%20(Tests)/r52447%20(7289)/results.html This is the first time I've seen it crash. It might be a recent regression.
Attachments
Nikolas Zimmermann
Comment 1 2009-12-21 12:10:08 PST
I wish there was a crashlog available :(
Dirk Schulze
Comment 2 2009-12-21 13:33:15 PST
It looks like random crashes on the Windows build bot. Some builds later svg/dynamic-updates/SVGLinearGradientElement-svgdom-y2-prop.html crashed, while Mask works again.
Nikolas Zimmermann
Comment 3 2009-12-21 14:48:04 PST
Hm, I thought it's a build problem, but maybe it's Cg/Win not handling 0x0 images properly. I see a warning on a lot of mask tests on Cg/Mac: "Mon Dec 21 23:40:05 nikolas-zimmermanns-macbook-pro-243.local DumpRenderTree[93233] <Error>: CGImageCreate: invalid image size: 0 x 0." in SVGMaskElement, there's a ImageBuffer object returned with 0x0 size. No problem on Cairo/Qt, according to Dirk. But it seems to lead to crashes on win, and warnings on mac with CoreGraphics. Dirk is aware, and will check tomorrow :-)
Dirk Schulze
Comment 4 2009-12-22 00:11:08 PST
(In reply to comment #3) > Hm, I thought it's a build problem, but maybe it's Cg/Win not handling 0x0 > images properly. > > I see a warning on a lot of mask tests on Cg/Mac: > "Mon Dec 21 23:40:05 nikolas-zimmermanns-macbook-pro-243.local > DumpRenderTree[93233] <Error>: CGImageCreate: invalid image size: 0 x 0." > > in SVGMaskElement, there's a ImageBuffer object returned with 0x0 size. No > problem on Cairo/Qt, according to Dirk. But it seems to lead to crashes on win, > and warnings on mac with CoreGraphics. > > Dirk is aware, and will check tomorrow :-) The problem occurred before the ImageBuffer changes in SVG Mask. And the changes don't affect Gradients.
Nikolas Zimmermann
Comment 5 2009-12-22 10:00:16 PST
(In reply to comment #4) > (In reply to comment #3) > > Hm, I thought it's a build problem, but maybe it's Cg/Win not handling 0x0 > > images properly. > > > > I see a warning on a lot of mask tests on Cg/Mac: > > "Mon Dec 21 23:40:05 nikolas-zimmermanns-macbook-pro-243.local > > DumpRenderTree[93233] <Error>: CGImageCreate: invalid image size: 0 x 0." > > > > in SVGMaskElement, there's a ImageBuffer object returned with 0x0 size. No > > problem on Cairo/Qt, according to Dirk. But it seems to lead to crashes on win, > > and warnings on mac with CoreGraphics. > > > > Dirk is aware, and will check tomorrow :-) > > The problem occurred before the ImageBuffer changes in SVG Mask. And the > changes don't affect Gradients. I take your word on that, didn't check on my own. Though it's still a bug affecting Cg that needs to be fixed
Dirk Schulze
Comment 6 2009-12-22 10:58:34 PST
(In reply to comment #5) > I take your word on that, didn't check on my own. Though it's still a bug > affecting Cg that needs to be fixed Yes. I'm working on it. :-)
Dirk Schulze
Comment 7 2009-12-30 14:54:55 PST
(In reply to comment #6) > I take your word on that, didn't check on my own. Though it's still a bug > affecting Cg that needs to be fixed The issue with one-dimensional sized ImageBuffer is fixed. The crash didn't appear on the build bots lately. I close this bug for now.
Note You need to log in before you can comment on or make changes to this bug.