Use unfragmented clip bounds when fragmentation is inhibited due to compositing

Use unfragmented clip bounds when fragmentation is inhibited due to compositing

This CL fixes one case we missed to patch up in:
https://codereview.chromium.org/2028893002

When a PaintLayer has opacity it will use the (incorrectly) fragmented bounds
for CompositingRecorder to save Skia layer. The bound is soft so it is not
always reproducible depends on internal implementation.

Committed: https://crrev.com/533c7f586be0bb8fc64705a82cdc8133d455b17f
Cr-Commit-Position: refs/heads/master@{#401504}

[trchen, 2016-06-22 23:47:08]

trchen@chromium.org changed reviewers:
+ chrishtr@chromium.org, wkorman@chromium.org

[trchen, 2016-06-22 23:47:09]

FYI: found this bug with my CL https://codereview.chromium.org/2075873002/
because it made culling stricter.

fast/multicol/composited-with-child-layer-in-next-column.html will catch this
bug once my CL gets submitted.

[trchen, 2016-06-22 23:47:14]

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

[commit-bot: I haz the power, 2016-06-22 23:48:00]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/2088333002/1

[chrishtr, 2016-06-22 23:54:00]

Did some more digging. Looks like this call site also needs the conditional?

https://cs.chromium.org/chromium/src/third_party/WebKit/Source/core/paint/Pai...

[trchen, 2016-06-22 23:58:24]

On 2016/06/22 23:54:00, chrishtr wrote:
> Did some more digging. Looks like this call site also needs the conditional?
> 
>
https://cs.chromium.org/chromium/src/third_party/WebKit/Source/core/paint/Pai...

I'm not sure. It is non-trivial to fix though, because we don't have
'rootLayer'(painting root) in that context. It may be easier to add a
fragmentationInhibited flag on PaintLayer which will be updated during
compositing updates.

[chrishtr, 2016-06-23 00:01:24]

On 2016/06/22 at 23:58:24, trchen wrote:
> On 2016/06/22 23:54:00, chrishtr wrote:
> > Did some more digging. Looks like this call site also needs the conditional?
> > 
> >
https://cs.chromium.org/chromium/src/third_party/WebKit/Source/core/paint/Pai...
> 
> I'm not sure. It is non-trivial to fix though, because we don't have
'rootLayer'(painting root) in that context. It may be easier to add a
fragmentationInhibited flag on PaintLayer which will be updated during
compositing updates.

Hmm. Yeah this looks like a chicken-egg issue with overlap testing actually,
since compositing affects the way pagination is painted. Please add a comment
instead then.

[trchen, 2016-06-23 00:13:08]

On 2016/06/23 00:01:24, chrishtr wrote:
> On 2016/06/22 at 23:58:24, trchen wrote:
> > On 2016/06/22 23:54:00, chrishtr wrote:
> > > Did some more digging. Looks like this call site also needs the
conditional?
> > > 
> > >
>
https://cs.chromium.org/chromium/src/third_party/WebKit/Source/core/paint/Pai...
> > 
> > I'm not sure. It is non-trivial to fix though, because we don't have
> 'rootLayer'(painting root) in that context. It may be easier to add a
> fragmentationInhibited flag on PaintLayer which will be updated during
> compositing updates.
> 
> Hmm. Yeah this looks like a chicken-egg issue with overlap testing actually,
> since compositing affects the way pagination is painted. Please add a comment
> instead then.

Done

[chrishtr, 2016-06-23 00:18:05]

The CQ bit was checked by chrishtr@chromium.org

[chrishtr, 2016-06-23 00:18:06]

lgtm

[commit-bot: I haz the power, 2016-06-23 00:18:30]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/2088333002/20001

[commit-bot: I haz the power, 2016-06-23 01:33:27]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2016-06-23 01:38:26]

Description was changed from

==========
Use unfragmented clip bounds when fragmentation is inhibited due to compositing

This CL fixes one case we missed to patch up in:
https://codereview.chromium.org/2028893002

When a PaintLayer has opacity it will use the (incorrectly) fragmented bounds
for CompositingRecorder to save Skia layer. The bound is soft so it is not
always reproducible depends on internal implementation.
==========

to

==========
Use unfragmented clip bounds when fragmentation is inhibited due to compositing

This CL fixes one case we missed to patch up in:
https://codereview.chromium.org/2028893002

When a PaintLayer has opacity it will use the (incorrectly) fragmented bounds
for CompositingRecorder to save Skia layer. The bound is soft so it is not
always reproducible depends on internal implementation.

Committed: https://crrev.com/533c7f586be0bb8fc64705a82cdc8133d455b17f
Cr-Commit-Position: refs/heads/master@{#401504}
==========

[commit-bot: I haz the power, 2016-06-23 01:38:27]

Patchset 2 (id:??) landed as
https://crrev.com/533c7f586be0bb8fc64705a82cdc8133d455b17f
Cr-Commit-Position: refs/heads/master@{#401504}