Set layers needsRepaint along the original compositingContainer chain if stacking status changes

Set layers needsRepaint along the original compositingContainer chain if stacking status changes

BUG=607768
TEST=paint/invalidation/relative-position-under-composited-scroll.html

Committed: https://crrev.com/89dcd6c67c48e8dfdb520a25556b50121cef75ba
Cr-Commit-Position: refs/heads/master@{#391107}

[Xianzhu, 2016-05-02 19:11:38]

wangxianzhu@chromium.org changed reviewers:
+ chrishtr@chromium.org

[chrishtr, 2016-05-02 21:13:32]

https://codereview.chromium.org/1937303003/diff/1/third_party/WebKit/LayoutTe...
File
third_party/WebKit/LayoutTests/paint/invalidation/relative-position-under-composited-scroll.html
(right):

https://codereview.chromium.org/1937303003/diff/1/third_party/WebKit/LayoutTe...
third_party/WebKit/LayoutTests/paint/invalidation/relative-position-under-composited-scroll.html:12:
internals.settings.setPreferCompositingToLCDTextEnabled(true);
How about using will-changeL transform instead of scrolling? Is there something
special about
scrolling?

[Xianzhu, 2016-05-02 21:59:29]

https://codereview.chromium.org/1937303003/diff/1/third_party/WebKit/LayoutTe...
File
third_party/WebKit/LayoutTests/paint/invalidation/relative-position-under-composited-scroll.html
(right):

https://codereview.chromium.org/1937303003/diff/1/third_party/WebKit/LayoutTe...
third_party/WebKit/LayoutTests/paint/invalidation/relative-position-under-composited-scroll.html:12:
internals.settings.setPreferCompositingToLCDTextEnabled(true);
On 2016/05/02 21:13:32, chrishtr wrote:
> How about using will-changeL transform instead of scrolling? Is there
something
> special about
> scrolling?

It's special because it establishes a composited layer but is not a stacking
context. When some of its descendants becomes stacked, the descendant will
change paint invalidation container to its stacking context.

[chrishtr, 2016-05-02 22:05:09]

On 2016/05/02 at 21:59:29, wangxianzhu wrote:
>
https://codereview.chromium.org/1937303003/diff/1/third_party/WebKit/LayoutTe...
> File
third_party/WebKit/LayoutTests/paint/invalidation/relative-position-under-composited-scroll.html
(right):
> 
>
https://codereview.chromium.org/1937303003/diff/1/third_party/WebKit/LayoutTe...
>
third_party/WebKit/LayoutTests/paint/invalidation/relative-position-under-composited-scroll.html:12:
internals.settings.setPreferCompositingToLCDTextEnabled(true);
> On 2016/05/02 21:13:32, chrishtr wrote:
> > How about using will-changeL transform instead of scrolling? Is there
something
> > special about
> > scrolling?
> 
> It's special because it establishes a composited layer but is not a stacking
context. When some of its descendants becomes stacked, the descendant will
change paint invalidation container to its stacking context.

backface-visibility:hidden?

[Xianzhu, 2016-05-02 23:20:56]

On 2016/05/02 22:05:09, chrishtr wrote:
> On 2016/05/02 at 21:59:29, wangxianzhu wrote:
> >
>
https://codereview.chromium.org/1937303003/diff/1/third_party/WebKit/LayoutTe...
> > File
>
third_party/WebKit/LayoutTests/paint/invalidation/relative-position-under-composited-scroll.html
> (right):
> > 
> >
>
https://codereview.chromium.org/1937303003/diff/1/third_party/WebKit/LayoutTe...
> >
>
third_party/WebKit/LayoutTests/paint/invalidation/relative-position-under-composited-scroll.html:12:
> internals.settings.setPreferCompositingToLCDTextEnabled(true);
> > On 2016/05/02 21:13:32, chrishtr wrote:
> > > How about using will-changeL transform instead of scrolling? Is there
> something
> > > special about
> > > scrolling?
> > 
> > It's special because it establishes a composited layer but is not a stacking
> context. When some of its descendants becomes stacked, the descendant will
> change paint invalidation container to its stacking context.
> 
> backface-visibility:hidden?

Tried that but failed to reproduce the bug. The 'item' div and the
backface-visibility:hidden layer are squashed together with the latter as the
squashing owner, causing setNeedsRepaint() happens to be correct.

[chrishtr, 2016-05-02 23:22:03]

The CQ bit was checked by chrishtr@chromium.org

[chrishtr, 2016-05-02 23:22:04]

lgtm

[commit-bot: I haz the power, 2016-05-02 23:22:32]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1937303003/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1937303003/1

[commit-bot: I haz the power, 2016-05-02 23:55:50]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2016-05-02 23:56:50]

Description was changed from

==========
Set layers needsRepaint along the original compositingContainer chain if
stacking status changes

BUG=607768
TEST=paint/invalidation/relative-position-under-composited-scroll.html
==========

to

==========
Set layers needsRepaint along the original compositingContainer chain if
stacking status changes

BUG=607768
TEST=paint/invalidation/relative-position-under-composited-scroll.html

Committed: https://crrev.com/89dcd6c67c48e8dfdb520a25556b50121cef75ba
Cr-Commit-Position: refs/heads/master@{#391107}
==========

[commit-bot: I haz the power, 2016-05-02 23:56:51]

Patchset 1 (id:??) landed as
https://crrev.com/89dcd6c67c48e8dfdb520a25556b50121cef75ba
Cr-Commit-Position: refs/heads/master@{#391107}