WebKit Bugzilla
New
Browse
Search+
Log In
×
Sign in with GitHub
or
Remember my login
Create Account
·
Forgot Password
Forgotten password account recovery
RESOLVED FIXED
65360
[Chromium][GPU] REGRESSION(
r90646
): Cannot render big sliver blocks.
https://bugs.webkit.org/show_bug.cgi?id=65360
Summary
[Chromium][GPU] REGRESSION(r90646): Cannot render big sliver blocks.
Hajime Morrita
Reported
2011-07-28 22:05:14 PDT
When GPU rendering is enabled, long div is not drawn in some circumstances. (Reduction attached.) This bug happens in a real site which has a style like following: div p { height:30px; width: 30px; list-style-type:none; text-indent:-9999px; -webkit-perspective:300; border-style: solid; } The text-indent line can make long sliver blocks. Attached reduction is more straightforward.
Attachments
Patch
(8.05 KB, patch)
2011-07-28 23:24 PDT
,
Hajime Morrita
no flags
Details
Formatted Diff
Diff
Show Obsolete
(1)
View All
Add attachment
proposed patch, testcase, etc.
Hajime Morrita
Comment 1
2011-07-28 23:24:54 PDT
Created
attachment 102334
[details]
Patch
Stephen White
Comment 2
2011-07-29 06:29:21 PDT
Thanks for the patch (and the test!) I'm not really the right person to review compositor changes, but I've added a few people who would be.
Hajime Morrita
Comment 3
2011-08-02 21:54:55 PDT
(In reply to
comment #2
)
> Thanks for the patch (and the test!) I'm not really the right person to review compositor changes, but I've added a few people who would be.
Hi Stephen, thanks you for responding! Enne, Could you give a review when you have time?
Adrienne Walker
Comment 4
2011-08-03 11:54:53 PDT
(In reply to
comment #3
)
> (In reply to
comment #2
) > > Thanks for the patch (and the test!) I'm not really the right person to review compositor changes, but I've added a few people who would be. > > Hi Stephen, thanks you for responding! > Enne, Could you give a review when you have time?
I'm not actually a reviewer yet. ;) Oh, I see why this is happening. I added the logic so that long, skinny layers didn't get tiled. This gets passed implicitly as an empty tile size and calculated in growLayerToContain, but then we don't clamp it there. I realize this isn't part of your patch, but you're touching that code. I wonder if it'd be cleaner to just do all the clamping internal to LayerTilerChromium? i.e. move the call in TiledLayerChromium to LayerTilerChromium::setTileSize so that any other callers in the future don't have to remember to clamp as well? Other than that minor thought, it unofficially looks good to me. Maybe jamesr can take a look. :)
Hajime Morrita
Comment 5
2011-08-03 21:25:28 PDT
Hi Adrienne, thanks for taking a look!
> > Other than that minor thought, it unofficially looks good to me. Maybe jamesr can take a look. :)
Well, Actually I noticed that
r92255
(for
Bug 64942
) fixed this. So I'll close this.
Note
You need to
log in
before you can comment on or make changes to this bug.
Top of Page
Format For Printing
XML
Clone This Bug