Root layer scrolling: set correct size for document's main GraphicsLayer.

Root layer scrolling: set correct size for document's main GraphicsLayer.

The main GraphicsLayer of a composited scroller should be the size of the
scroller's frame.  Report this size in PaintLayer::boundingBoxForCompositing
and update ViewPainter to use the same rect when painting the background fill.

This fixes placement of scrollbar layers in RTL documents, which were confused
by a bogus m_offsetFromLayoutObject because the rect returned by
LayoutView::documentRect is relative to the scroll origin.

This patch makes compositing/rtl/rtl-overflow-invalidation.html pass (discovered
through flag-specific expectations!)

Sadly we regress several repaint-tracking tests due to the new invalidation in
FrameView::viewportSizeChanged.  These will need flag-specific baselines (not
yet implemented) at least until http://crbug.com/568847 is addressed.

BUG=542432
Committed: https://crrev.com/027a020589e4b35c14a40193e4b953733e70f410
Cr-Commit-Position: refs/heads/master@{#367034}

[skobes, 2015-12-17 03:20:33]

Description was changed from

==========
Update PaintLayer::boundingBoxForCompositing for root layer scrolling.

BUG=542432
==========

to

==========
Root layer scrolling: set correct size for document's main GraphicsLayer.

The main GraphicsLayer of a composited scroller should be the size of the
scroller's frame.  Report this size in PaintLayer::boundingBoxForCompositing
and update ViewPainter to use the same rect when painting the background fill.

This fixes placement of scrollbar layers in RTL documents, which were confused
by a bogus m_offsetFromLayoutObject because the rect returned by
LayoutView::documentRect is relative to the scroll origin.

Sadly we regress several repaint-tracking tests due to the new invalidation in
FrameView::viewportSizeChanged.  These will need flag-specific baselines until
http://crbug.com/568847 is addressed.

BUG=542432
==========

[skobes, 2015-12-17 03:24:50]

Description was changed from

==========
Root layer scrolling: set correct size for document's main GraphicsLayer.

The main GraphicsLayer of a composited scroller should be the size of the
scroller's frame.  Report this size in PaintLayer::boundingBoxForCompositing
and update ViewPainter to use the same rect when painting the background fill.

This fixes placement of scrollbar layers in RTL documents, which were confused
by a bogus m_offsetFromLayoutObject because the rect returned by
LayoutView::documentRect is relative to the scroll origin.

Sadly we regress several repaint-tracking tests due to the new invalidation in
FrameView::viewportSizeChanged.  These will need flag-specific baselines until
http://crbug.com/568847 is addressed.

BUG=542432
==========

to

==========
Root layer scrolling: set correct size for document's main GraphicsLayer.

The main GraphicsLayer of a composited scroller should be the size of the
scroller's frame.  Report this size in PaintLayer::boundingBoxForCompositing
and update ViewPainter to use the same rect when painting the background fill.

This fixes placement of scrollbar layers in RTL documents, which were confused
by a bogus m_offsetFromLayoutObject because the rect returned by
LayoutView::documentRect is relative to the scroll origin.

This patch makes compositing/rtl/rtl-overflow-invalidation.html pass (discovered
through flag-specific expectations!)

Sadly we regress several repaint-tracking tests due to the new invalidation in
FrameView::viewportSizeChanged.  These will need flag-specific baselines until
http://crbug.com/568847 is addressed.

BUG=542432
==========

[skobes, 2015-12-17 03:25:20]

Description was changed from

==========
Root layer scrolling: set correct size for document's main GraphicsLayer.

The main GraphicsLayer of a composited scroller should be the size of the
scroller's frame.  Report this size in PaintLayer::boundingBoxForCompositing
and update ViewPainter to use the same rect when painting the background fill.

This fixes placement of scrollbar layers in RTL documents, which were confused
by a bogus m_offsetFromLayoutObject because the rect returned by
LayoutView::documentRect is relative to the scroll origin.

This patch makes compositing/rtl/rtl-overflow-invalidation.html pass (discovered
through flag-specific expectations!)

Sadly we regress several repaint-tracking tests due to the new invalidation in
FrameView::viewportSizeChanged.  These will need flag-specific baselines until
http://crbug.com/568847 is addressed.

BUG=542432
==========

to

==========
Root layer scrolling: set correct size for document's main GraphicsLayer.

The main GraphicsLayer of a composited scroller should be the size of the
scroller's frame.  Report this size in PaintLayer::boundingBoxForCompositing
and update ViewPainter to use the same rect when painting the background fill.

This fixes placement of scrollbar layers in RTL documents, which were confused
by a bogus m_offsetFromLayoutObject because the rect returned by
LayoutView::documentRect is relative to the scroll origin.

This patch makes compositing/rtl/rtl-overflow-invalidation.html pass (discovered
through flag-specific expectations!)

Sadly we regress several repaint-tracking tests due to the new invalidation in
FrameView::viewportSizeChanged.  These will need flag-specific baselines (not
yet implemented) at least until http://crbug.com/568847 is addressed.

BUG=542432
==========

[skobes, 2015-12-17 05:04:27]

skobes@chromium.org changed reviewers:
+ trchen@chromium.org

[skobes, 2015-12-18 18:01:26]

skobes@chromium.org changed reviewers:
+ vollick@chromium.org
- trchen@chromium.org

[skobes, 2015-12-18 18:01:50]

I think trchen is OOO.  @vollick can you take a look (if you're around)?

[Ian Vollick, 2015-12-22 03:21:50]

https://codereview.chromium.org/1472083006/diff/120001/third_party/WebKit/Sou...
File third_party/WebKit/Source/core/paint/PaintLayer.cpp (right):

https://codereview.chromium.org/1472083006/diff/120001/third_party/WebKit/Sou...
third_party/WebKit/Source/core/paint/PaintLayer.cpp:2190: if (isRootLayer() &&
!needsCompositedScrolling())
If the root layer does need composited scrolling, would it be correct to return
the viewRect here? I.e., can the code here match that in ViewPainter (and can it
be shared)?

[skobes, 2015-12-28 22:39:14]

Patchset #1 (id:1) has been deleted

[skobes, 2015-12-28 22:39:18]

Patchset #1 (id:20001) has been deleted

[skobes, 2015-12-28 22:39:21]

Patchset #1 (id:40001) has been deleted

[skobes, 2015-12-28 22:39:26]

Patchset #1 (id:60001) has been deleted

[skobes, 2015-12-28 22:39:30]

Patchset #1 (id:80001) has been deleted

[skobes, 2015-12-28 22:39:35]

Patchset #1 (id:100001) has been deleted

[skobes, 2015-12-28 23:33:17]

https://codereview.chromium.org/1472083006/diff/120001/third_party/WebKit/Sou...
File third_party/WebKit/Source/core/paint/PaintLayer.cpp (right):

https://codereview.chromium.org/1472083006/diff/120001/third_party/WebKit/Sou...
third_party/WebKit/Source/core/paint/PaintLayer.cpp:2190: if (isRootLayer() &&
!needsCompositedScrolling())
On 2015/12/22 at 03:21:50, vollick wrote:
> If the root layer does need composited scrolling, would it be correct to
return the viewRect here? I.e., can the code here match that in ViewPainter (and
can it be shared)?

It does return the viewRect, indirectly, since the result of localClipRect() is
computed by PaintLayerClipper::calculateRects, which was updated to use
viewRect() in http://crrev.com/1262063002.

I've updated ViewPainter to call PaintLayer::boundingBoxForCompositing so that
this logic is reused.

[Ian Vollick, 2015-12-29 00:54:29]

https://codereview.chromium.org/1472083006/diff/120001/third_party/WebKit/Sou...
File third_party/WebKit/Source/core/paint/PaintLayer.cpp (right):

https://codereview.chromium.org/1472083006/diff/120001/third_party/WebKit/Sou...
third_party/WebKit/Source/core/paint/PaintLayer.cpp:2190: if (isRootLayer() &&
!needsCompositedScrolling())
On 2015/12/28 23:33:17, skobes wrote:
> On 2015/12/22 at 03:21:50, vollick wrote:
> > If the root layer does need composited scrolling, would it be correct to
> return the viewRect here? I.e., can the code here match that in ViewPainter
(and
> can it be shared)?
> 
> It does return the viewRect, indirectly, since the result of localClipRect()
is
> computed by PaintLayerClipper::calculateRects, which was updated to use
> viewRect() in http://crrev.com/1262063002.
> 
> I've updated ViewPainter to call PaintLayer::boundingBoxForCompositing so that
> this logic is reused.

Thanks, lgtm.

[skobes, 2015-12-29 01:01:09]

The CQ bit was checked by skobes@chromium.org

[commit-bot: I haz the power, 2015-12-29 01:02:48]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1472083006/160001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1472083006/160001

[commit-bot: I haz the power, 2015-12-29 01:09:22]

Committed patchset #3 (id:160001)

[commit-bot: I haz the power, 2015-12-29 01:10:06]

Description was changed from

==========
Root layer scrolling: set correct size for document's main GraphicsLayer.

The main GraphicsLayer of a composited scroller should be the size of the
scroller's frame.  Report this size in PaintLayer::boundingBoxForCompositing
and update ViewPainter to use the same rect when painting the background fill.

This fixes placement of scrollbar layers in RTL documents, which were confused
by a bogus m_offsetFromLayoutObject because the rect returned by
LayoutView::documentRect is relative to the scroll origin.

This patch makes compositing/rtl/rtl-overflow-invalidation.html pass (discovered
through flag-specific expectations!)

Sadly we regress several repaint-tracking tests due to the new invalidation in
FrameView::viewportSizeChanged.  These will need flag-specific baselines (not
yet implemented) at least until http://crbug.com/568847 is addressed.

BUG=542432
==========

to

==========
Root layer scrolling: set correct size for document's main GraphicsLayer.

The main GraphicsLayer of a composited scroller should be the size of the
scroller's frame.  Report this size in PaintLayer::boundingBoxForCompositing
and update ViewPainter to use the same rect when painting the background fill.

This fixes placement of scrollbar layers in RTL documents, which were confused
by a bogus m_offsetFromLayoutObject because the rect returned by
LayoutView::documentRect is relative to the scroll origin.

This patch makes compositing/rtl/rtl-overflow-invalidation.html pass (discovered
through flag-specific expectations!)

Sadly we regress several repaint-tracking tests due to the new invalidation in
FrameView::viewportSizeChanged.  These will need flag-specific baselines (not
yet implemented) at least until http://crbug.com/568847 is addressed.

BUG=542432

Committed: https://crrev.com/027a020589e4b35c14a40193e4b953733e70f410
Cr-Commit-Position: refs/heads/master@{#367034}
==========

[commit-bot: I haz the power, 2015-12-29 01:10:07]

Patchset 3 (id:??) landed as
https://crrev.com/027a020589e4b35c14a40193e4b953733e70f410
Cr-Commit-Position: refs/heads/master@{#367034}