Invalidate non-composited subtree on needsPaintInvalidationLayer

Invalidate non-composited subtree on needsPaintInvalidationLayer

This is important for slimming paint because invalidation of layer
doesn't mean invalidation of all objects on the layer.

Changed the layer tree walk to object tree walk. We may have the
same performance concerns as we had when we changed
invalidatePaintIncludingNonCompositingDescendants() from layer tree
walk to object tree walk but we haven't observed performance
degradation for that change. I believe this change will be similar.

BUG=491993
TEST=fast/repaint/add-transform-repaint-descendants.html

Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=196534

[Xianzhu, 2015-06-04 00:46:31]

wangxianzhu@chromium.org changed reviewers:
+ chrishtr@chromium.org, jchaffraix@chromium.org, trchen@chromium.org

[trchen, 2015-06-04 00:59:36]

https://codereview.chromium.org/1164713010/diff/1/Source/core/frame/FrameView...
File Source/core/frame/FrameView.cpp (right):

https://codereview.chromium.org/1164713010/diff/1/Source/core/frame/FrameView...
Source/core/frame/FrameView.cpp:1348:
layoutObject->setShouldDoFullPaintInvalidationIncludingNonCompositingDescendants();
I'm not certain if this is needed. The fixed-position elements are now painted
using transform code path. Non-compositing descendants should have static paint
offset.

I think we only need to invalidate layoutObject's display item, while
invalidating the visual overflow rect of the whole RenderLayer.

[Xianzhu, 2015-06-04 01:49:56]

https://codereview.chromium.org/1164713010/diff/1/Source/core/frame/FrameView...
File Source/core/frame/FrameView.cpp (right):

https://codereview.chromium.org/1164713010/diff/1/Source/core/frame/FrameView...
Source/core/frame/FrameView.cpp:1348:
layoutObject->setShouldDoFullPaintInvalidationIncludingNonCompositingDescendants();
On 2015/06/04 00:59:36, trchen wrote:
> I'm not certain if this is needed. The fixed-position elements are now painted
> using transform code path. Non-compositing descendants should have static
paint
> offset.
> 
> I think we only need to invalidate layoutObject's display item, while
> invalidating the visual overflow rect of the whole RenderLayer.

For now there seems simple way to implement this. At this place, we haven't
calculated visual overflow rect yet, so we can only set paint invalidation
flags. I think this will be a good optimization opportunity during phase 2.

[Xianzhu, 2015-06-04 02:02:05]

On 2015/06/04 01:49:56, Xianzhu wrote:

> For now there seems simple way to implement this. At this place, we haven't
> calculated visual overflow rect yet, so we can only set paint invalidation
> flags. I think this will be a good optimization opportunity during phase 2.

s/seems simple way/seems *no* simple way/ :)

[chrishtr, 2015-06-04 17:41:09]

lgtm

https://codereview.chromium.org/1164713010/diff/1/Source/core/frame/FrameView...
File Source/core/frame/FrameView.cpp (right):

https://codereview.chromium.org/1164713010/diff/1/Source/core/frame/FrameView...
Source/core/frame/FrameView.cpp:1348:
layoutObject->setShouldDoFullPaintInvalidationIncludingNonCompositingDescendants();
On 2015/06/04 at 01:49:56, Xianzhu wrote:
> On 2015/06/04 00:59:36, trchen wrote:
> > I'm not certain if this is needed. The fixed-position elements are now
painted
> > using transform code path. Non-compositing descendants should have static
paint
> > offset.
> > 
> > I think we only need to invalidate layoutObject's display item, while
> > invalidating the visual overflow rect of the whole RenderLayer.
> 
> For now there seems simple way to implement this. At this place, we haven't
calculated visual overflow rect yet, so we can only set paint invalidation
flags. I think this will be a good optimization opportunity during phase 2.

Please write this down in a bug so we don't forget.

[Xianzhu, 2015-06-04 18:04:39]

https://codereview.chromium.org/1164713010/diff/1/Source/core/frame/FrameView...
File Source/core/frame/FrameView.cpp (right):

https://codereview.chromium.org/1164713010/diff/1/Source/core/frame/FrameView...
Source/core/frame/FrameView.cpp:1348:
layoutObject->setShouldDoFullPaintInvalidationIncludingNonCompositingDescendants();
On 2015/06/04 17:41:09, chrishtr wrote:
> On 2015/06/04 at 01:49:56, Xianzhu wrote:
> > On 2015/06/04 00:59:36, trchen wrote:
> > > I'm not certain if this is needed. The fixed-position elements are now
> painted
> > > using transform code path. Non-compositing descendants should have static
> paint
> > > offset.
> > > 
> > > I think we only need to invalidate layoutObject's display item, while
> > > invalidating the visual overflow rect of the whole RenderLayer.
> > 
> > For now there seems simple way to implement this. At this place, we haven't
> calculated visual overflow rect yet, so we can only set paint invalidation
> flags. I think this will be a good optimization opportunity during phase 2.
> 
> Please write this down in a bug so we don't forget.

crbug.com/451090#c17

[Xianzhu, 2015-06-04 18:04:45]

The CQ bit was checked by wangxianzhu@chromium.org

[commit-bot: I haz the power, 2015-06-04 18:05:00]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1164713010/1

[commit-bot: I haz the power, 2015-06-04 20:00:30]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2015-06-04 20:00:31]

Try jobs failed on following builders:
  linux_blink_rel on tryserver.blink (JOB_FAILED,
http://build.chromium.org/p/tryserver.blink/builders/linux_blink_rel/builds/6...)

[Xianzhu, 2015-06-04 20:31:09]

The CQ bit was checked by wangxianzhu@chromium.org

[Xianzhu, 2015-06-04 20:31:09]

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/1164713010/#ps20001
(title: "Update expectations")

[commit-bot: I haz the power, 2015-06-04 20:32:04]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1164713010/20001

[commit-bot: I haz the power, 2015-06-04 22:26:27]

Committed patchset #2 (id:20001) as
https://src.chromium.org/viewvc/blink?view=rev&revision=196534