Fix clipping of squashed sibling layers with border radius masks

Fix clipping of squashed sibling layers with border radius masks

When a composited element with a border-radius overflow parent
has border radius and overflows itself, the element's clip rect
is included on the mask layer for the parent, which causes incorrect
clipping of sibling content.

Unfortunately a ref test fails due to a single pixel different on
the border radius.

R=chrishtr@chromium.org
BUG=709533
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2

Review-Url: https://codereview.chromium.org/2821733002
Cr-Commit-Position: refs/heads/master@{#465028}
Committed: https://chromium.googlesource.com/chromium/src/+/e50892ab6838d9359efbf309035f792c3e68b25b

[Stephen Chennney, 2017-04-14 21:03:50]

Description was changed from

==========
Fix clipping of sibling sqaushed layers with border radius masks

When a composited element with a border-radius overflow parent
has border radius and overflow itself, the element's clip rect
is included on the mask layer for the parent, which causes incorrect
clipping of sibling content.

Unfortunately a ref test fails due to a single pixel different on
the border radius.

R=chrishtr@chromium.org
BUG=709533
==========

to

==========
Fix clipping of sibling sqaushed layers with border radius masks

When a composited element with a border-radius overflow parent
has border radius and overflow itself, the element's clip rect
is included on the mask layer for the parent, which causes incorrect
clipping of sibling content.

Unfortunately a ref test fails due to a single pixel different on
the border radius.

R=chrishtr@chromium.org
BUG=709533
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2
==========

[Stephen Chennney, 2017-04-14 21:08:31]

Dumb mistake when I implemented border-radius for composited children. This is
an M-58 release blocker that I am hoping to commit as soon as the test results
come in, so that we have some Canary time before the Stable branch.

[chrishtr, 2017-04-14 21:26:23]

Description was changed from

==========
Fix clipping of sibling sqaushed layers with border radius masks

When a composited element with a border-radius overflow parent
has border radius and overflow itself, the element's clip rect
is included on the mask layer for the parent, which causes incorrect
clipping of sibling content.

Unfortunately a ref test fails due to a single pixel different on
the border radius.

R=chrishtr@chromium.org
BUG=709533
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2
==========

to

==========
Fix clipping of sibling squashed layers with border radius masks

When a composited element with a border-radius overflow parent
has border radius and overflow itself, the element's clip rect
is included on the mask layer for the parent, which causes incorrect
clipping of sibling content.

Unfortunately a ref test fails due to a single pixel different on
the border radius.

R=chrishtr@chromium.org
BUG=709533
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2
==========

[chrishtr, 2017-04-14 21:38:18]

Description was changed from

==========
Fix clipping of sibling squashed layers with border radius masks

When a composited element with a border-radius overflow parent
has border radius and overflow itself, the element's clip rect
is included on the mask layer for the parent, which causes incorrect
clipping of sibling content.

Unfortunately a ref test fails due to a single pixel different on
the border radius.

R=chrishtr@chromium.org
BUG=709533
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2
==========

to

==========
Fix clipping of sibling squashed layers with border radius masks

When a composited element with a border-radius overflow parent
has border radius and overflows itself, the element's clip rect
is included on the mask layer for the parent, which causes incorrect
clipping of sibling content.

Unfortunately a ref test fails due to a single pixel different on
the border radius.

R=chrishtr@chromium.org
BUG=709533
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2
==========

[chrishtr, 2017-04-14 21:38:40]

Description was changed from

==========
Fix clipping of sibling squashed layers with border radius masks

When a composited element with a border-radius overflow parent
has border radius and overflows itself, the element's clip rect
is included on the mask layer for the parent, which causes incorrect
clipping of sibling content.

Unfortunately a ref test fails due to a single pixel different on
the border radius.

R=chrishtr@chromium.org
BUG=709533
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2
==========

to

==========
Fix clipping of sibling layers with border radius masks

When a composited element with a border-radius overflow parent
has border radius and overflows itself, the element's clip rect
is included on the mask layer for the parent, which causes incorrect
clipping of sibling content.

Unfortunately a ref test fails due to a single pixel different on
the border radius.

R=chrishtr@chromium.org
BUG=709533
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2
==========

[chrishtr, 2017-04-14 21:41:33]

Description was changed from

==========
Fix clipping of sibling layers with border radius masks

When a composited element with a border-radius overflow parent
has border radius and overflows itself, the element's clip rect
is included on the mask layer for the parent, which causes incorrect
clipping of sibling content.

Unfortunately a ref test fails due to a single pixel different on
the border radius.

R=chrishtr@chromium.org
BUG=709533
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2
==========

to

==========
Fix clipping of squashed sibling layers with border radius masks

When a composited element with a border-radius overflow parent
has border radius and overflows itself, the element's clip rect
is included on the mask layer for the parent, which causes incorrect
clipping of sibling content.

Unfortunately a ref test fails due to a single pixel different on
the border radius.

R=chrishtr@chromium.org
BUG=709533
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2
==========

[chrishtr, 2017-04-14 21:43:40]

On my mac, B and C only composite and disappear if the window is narrow enough.
If you add backface-visibility: hidden to .outer, it always reproduces. I
suggest
you change the testcase to do that.

Also, are B and C actually squashed in to a CLM::squashing_layer_ ?

https://codereview.chromium.org/2821733002/diff/20001/third_party/WebKit/Layo...
File
third_party/WebKit/LayoutTests/compositing/composited-flex-siblings-with-border-radius.html
(right):

https://codereview.chromium.org/2821733002/diff/20001/third_party/WebKit/Layo...
third_party/WebKit/LayoutTests/compositing/composited-flex-siblings-with-border-radius.html:32:
<div class="seperator">A</div>
nit: separator

[Stephen Chennney, 2017-04-14 22:08:57]

On 2017/04/14 21:43:40, chrishtr wrote:
> On my mac, B and C only composite and disappear if the window is narrow
enough.
> If you add backface-visibility: hidden to .outer, it always reproduces. I
> suggest
> you change the testcase to do that.

Good to know. Maybe we squash differently when the parent is composited.

> 
> Also, are B and C actually squashed in to a CLM::squashing_layer_ ?

Yes, that's the source of the bug because the squashing layer gets added as a
sibling of the composited buttons, after them in paint order.

Actually, that just made me realize that the test case might not fail on content
shell, because it was not squashing the separators. Maybe this will have to wait
until Monday.
 
>
https://codereview.chromium.org/2821733002/diff/20001/third_party/WebKit/Layo...
> File
>
third_party/WebKit/LayoutTests/compositing/composited-flex-siblings-with-border-radius.html
> (right):
> 
>
https://codereview.chromium.org/2821733002/diff/20001/third_party/WebKit/Layo...
>
third_party/WebKit/LayoutTests/compositing/composited-flex-siblings-with-border-radius.html:32:
> <div class="seperator">A</div>
> nit: separator

I'll fix that - came from the original content.

[Stephen Chennney, 2017-04-15 17:26:37]

This is not the right fix. The issue is with the client pointers on the ancestor
clip and mask layers. In the basic case of a single composited child of a
border-radius overflow, the client of the clip and mask layers is the overflow
element. But if there are more than 1 child, the clip and mask client somehow
ends up being the child layer.

In the first case we need to include the "client's" clip, while in the second we
do not want to include the "client's" clip.

The underlying issue is that the test case creates 2 ancestor clip+mask layers
even though both have the same effect and only one is really needed. I think the
client is the child so that any given CLM only has one ancestor clip layer.

We create this (AC == ancestor clip, AM == ancestor mask):

A (overflow hidden and border radius)
|    \
AC   AC
|     |
AM   AM
|    |  \
1    2  3

But, as far as I can tell, we could just as well have

A (overflow hidden and border radius)
  \
  AC
   |
  AM
/  |  \
1  2  3

[chrishtr, 2017-04-17 17:25:00]

https://codereview.chromium.org/2821733002/diff/20001/third_party/WebKit/Sour...
File third_party/WebKit/Source/core/paint/PaintLayerPainter.cpp (right):

https://codereview.chromium.org/2821733002/diff/20001/third_party/WebKit/Sour...
third_party/WebKit/Source/core/paint/PaintLayerPainter.cpp:965: case
kPaintPhaseClippingMask:
Your fix is generally correct, just a small error. kPaintPhaseClippingMask can
be true
for more than just ancestor mask clips. If you use this code instead, it should
work:

case kPaintPhaseClippingMask:
        if (paint_flags & kPaintLayerPaintingAncestorClippingMaskPhase) {
          clipping_rule = LayerClipRecorder::kDoNotIncludeSelfForBorderRadius;
          break;
        }

[Stephen Chennney, 2017-04-17 17:43:41]

The CQ bit was checked by schenney@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2017-04-17 17:43:56]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[chrishtr, 2017-04-17 19:06:09]

lgtm

[Stephen Chennney, 2017-04-17 19:40:45]

The CQ bit was checked by schenney@chromium.org

[Stephen Chennney, 2017-04-17 19:40:45]

The patchset sent to the CQ was uploaded after l-g-t-m from
chrishtr@chromium.org
Link to the patchset: https://codereview.chromium.org/2821733002/#ps120001
(title: "Fix comments")

[commit-bot: I haz the power, 2017-04-17 19:41:05]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2017-04-17 21:33:24]

CQ is committing da patch.

Bot data: {"patchset_id": 120001, "attempt_start_ts": 1492458045078400,
"parent_rev": "3781c029a68e4ea22f3e36db45a561723b610667", "commit_rev":
"e50892ab6838d9359efbf309035f792c3e68b25b"}

[commit-bot: I haz the power, 2017-04-17 21:33:37]

Description was changed from

==========
Fix clipping of squashed sibling layers with border radius masks

When a composited element with a border-radius overflow parent
has border radius and overflows itself, the element's clip rect
is included on the mask layer for the parent, which causes incorrect
clipping of sibling content.

Unfortunately a ref test fails due to a single pixel different on
the border radius.

R=chrishtr@chromium.org
BUG=709533
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2
==========

to

==========
Fix clipping of squashed sibling layers with border radius masks

When a composited element with a border-radius overflow parent
has border radius and overflows itself, the element's clip rect
is included on the mask layer for the parent, which causes incorrect
clipping of sibling content.

Unfortunately a ref test fails due to a single pixel different on
the border radius.

R=chrishtr@chromium.org
BUG=709533
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2

Review-Url: https://codereview.chromium.org/2821733002
Cr-Commit-Position: refs/heads/master@{#465028}
Committed:
https://chromium.googlesource.com/chromium/src/+/e50892ab6838d9359efbf309035f...
==========

[commit-bot: I haz the power, 2017-04-17 21:33:43]

Committed patchset #7 (id:120001) as
https://chromium.googlesource.com/chromium/src/+/e50892ab6838d9359efbf309035f...