Reland of Skip PaintLayer empty paint phases if it previously painted nothing

Reland of Skip PaintLayer empty paint phases if it previously painted nothing (patchset #2 id:140001 of https://codereview.chromium.org/1815293002/ )

The original CL was reverted because it caused crbug.com/595839.

The reason of the bug was that we also when we painted the root
background we also traverse descendant layers, which actually painted
nothing and mistakenly marked the descendant layers not having paintings
for descendant backgrounds, float and descendant outlines paint phases,
causing the normal painting to skip painting of the paint phases.

https://codereview.chromium.org/2294313002/ fixed the root background
painting issue, so we can reland this CL now.

Also added under-invalidation checking for the needsPaintPhaseXXX()
flags by forcing painting even if needsPaintPhaseXXX() is false, and
asserting no actual painting occurs.

Original issue's description:
> Revert of Skip PaintLayer empty paint phases if it previously painted nothing (patchset #12 id:220001 of https://codereview.chromium.org/1739893006/ )
>
> Reason for revert:
> BUG=595839
>
> Original issue's description:
> > Skip PaintLayer empty paint phases if it previously painted nothing
> >
> > This avoids later painting of empty phases if the layer is marked to need
> > a paint phase but the phase painted nothing in the following cases:
> > - the object marking the layer needing the paint phase is not visible
> >   (e.g. because it is clipped)
> > - the layer was marked needing the paint phase by some object that
> >   has been deleted or moved to another layer.
> >
> > Instead of directly clearing the needsPaintPhase flags, this CL adds
> > a new set of flags to indicate if the previous paint of the phases painted
> > nothing. This is because invalidations of the needsPaintPhase flags and the
> > new flags need different conditions. The new flags are cleared when scroll
> > offset, interest rect or clipping changes.
> >
> > Cluster telemetry runs show that this will improve non-cached painting
> > by 1.7% and fully-cached painting (with subsequence-caching disabled) by
> > 45%.
> >
> > BUG=574938
> >
> > Committed: https://crrev.com/8ad6e62af5301a7e8dacd12fc1ae7914258b9e7a
> > Cr-Commit-Position: refs/heads/master@{#380841}
>
> TBR=chrishtr@chromium.org
> # Not skipping CQ checks because original CL landed more than 1 days ago.
> BUG=574938
>
> Committed: https://crrev.com/e425e3a8c5204d2d75f34c885c33f077372d1ff8
> Cr-Commit-Position: refs/heads/master@{#382018}

TBR=chrishtr@chromium.org
# Not skipping CQ checks because original CL landed more than 1 days ago.
BUG=595839
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2

Committed: https://crrev.com/a6378700ce7244fa1da85321aeb31661cde0ff16
Cr-Commit-Position: refs/heads/master@{#416008}

[Xianzhu, 2016-08-26 22:17:07]

Created Reland of Skip PaintLayer empty paint phases if it previously painted
nothing

[Xianzhu, 2016-08-26 22:28:11]

Please ignore the published Email. Just created this for debugging, but
codereview published it automatically.

[Xianzhu, 2016-08-29 23:58:20]

Description was changed from

==========
Reland of Skip PaintLayer empty paint phases if it previously painted nothing
(patchset #2 id:140001 of https://codereview.chromium.org/1815293002/ )

Reason for revert:
Create a reland CL for debugging.

Original issue's description:
> Revert of Skip PaintLayer empty paint phases if it previously painted nothing
(patchset #12 id:220001 of https://codereview.chromium.org/1739893006/ )
> 
> Reason for revert:
> BUG=595839
> 
> Original issue's description:
> > Skip PaintLayer empty paint phases if it previously painted nothing
> >
> > This avoids later painting of empty phases if the layer is marked to need
> > a paint phase but the phase painted nothing in the following cases:
> > - the object marking the layer needing the paint phase is not visible
> >   (e.g. because it is clipped)
> > - the layer was marked needing the paint phase by some object that
> >   has been deleted or moved to another layer.
> >
> > Instead of directly clearing the needsPaintPhase flags, this CL adds
> > a new set of flags to indicate if the previous paint of the phases painted
> > nothing. This is because invalidations of the needsPaintPhase flags and the
> > new flags need different conditions. The new flags are cleared when scroll
> > offset, interest rect or clipping changes.
> >
> > Cluster telemetry runs show that this will improve non-cached painting
> > by 1.7% and fully-cached painting (with subsequence-caching disabled) by
> > 45%.
> >
> > BUG=574938
> >
> > Committed: https://crrev.com/8ad6e62af5301a7e8dacd12fc1ae7914258b9e7a
> > Cr-Commit-Position: refs/heads/master@{#380841}
> 
> TBR=chrishtr@chromium.org
> # Not skipping CQ checks because original CL landed more than 1 days ago.
> BUG=574938
> 
> Committed: https://crrev.com/e425e3a8c5204d2d75f34c885c33f077372d1ff8
> Cr-Commit-Position: refs/heads/master@{#382018}

TBR=chrishtr@chromium.org
# Not skipping CQ checks because original CL landed more than 1 days ago.
BUG=595839
==========

to

==========
Reland of Skip PaintLayer empty paint phases if it previously painted nothing
(patchset #2 id:140001 of https://codereview.chromium.org/1815293002/ )

Reason for revert:
Create a reland CL for debugging.

Original issue's description:
> Revert of Skip PaintLayer empty paint phases if it previously painted nothing
(patchset #12 id:220001 of https://codereview.chromium.org/1739893006/ )
> 
> Reason for revert:
> BUG=595839
> 
> Original issue's description:
> > Skip PaintLayer empty paint phases if it previously painted nothing
> >
> > This avoids later painting of empty phases if the layer is marked to need
> > a paint phase but the phase painted nothing in the following cases:
> > - the object marking the layer needing the paint phase is not visible
> >   (e.g. because it is clipped)
> > - the layer was marked needing the paint phase by some object that
> >   has been deleted or moved to another layer.
> >
> > Instead of directly clearing the needsPaintPhase flags, this CL adds
> > a new set of flags to indicate if the previous paint of the phases painted
> > nothing. This is because invalidations of the needsPaintPhase flags and the
> > new flags need different conditions. The new flags are cleared when scroll
> > offset, interest rect or clipping changes.
> >
> > Cluster telemetry runs show that this will improve non-cached painting
> > by 1.7% and fully-cached painting (with subsequence-caching disabled) by
> > 45%.
> >
> > BUG=574938
> >
> > Committed: https://crrev.com/8ad6e62af5301a7e8dacd12fc1ae7914258b9e7a
> > Cr-Commit-Position: refs/heads/master@{#380841}
> 
> TBR=chrishtr@chromium.org
> # Not skipping CQ checks because original CL landed more than 1 days ago.
> BUG=574938
> 
> Committed: https://crrev.com/e425e3a8c5204d2d75f34c885c33f077372d1ff8
> Cr-Commit-Position: refs/heads/master@{#382018}

TBR=chrishtr@chromium.org
# Not skipping CQ checks because original CL landed more than 1 days ago.
BUG=595839
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2
==========

[Xianzhu, 2016-08-29 23:58:25]

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

[commit-bot: I haz the power, 2016-08-30 00:31:01]

Dry run: 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, 2016-08-30 03:41:09]

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

[commit-bot: I haz the power, 2016-08-30 03:41:09]

Dry run: Try jobs failed on following builders:
  cast_shell_linux on master.tryserver.chromium.linux (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/cast_shell_linu...)

[Xianzhu, 2016-08-30 16:03:53]

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

[commit-bot: I haz the power, 2016-08-30 16:04:29]

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

[Xianzhu, 2016-08-30 16:05:10]

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

[commit-bot: I haz the power, 2016-08-30 16:05:24]

Dry run: 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, 2016-08-30 16:08:16]

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

[commit-bot: I haz the power, 2016-08-30 16:08:17]

Dry run: Try jobs failed on following builders:
  mac_chromium_compile_dbg_ng on master.tryserver.chromium.mac (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.mac/builders/mac_chromium_comp...)
  mac_chromium_rel_ng on master.tryserver.chromium.mac (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.mac/builders/mac_chromium_rel_...)

[Xianzhu, 2016-08-30 16:28:36]

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

[commit-bot: I haz the power, 2016-08-30 16:28:53]

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

[Xianzhu, 2016-08-30 18:15:09]

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

[commit-bot: I haz the power, 2016-08-30 18:16:18]

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

[Xianzhu, 2016-08-30 21:33:31]

Description was changed from

==========
Reland of Skip PaintLayer empty paint phases if it previously painted nothing
(patchset #2 id:140001 of https://codereview.chromium.org/1815293002/ )

Reason for revert:
Create a reland CL for debugging.

Original issue's description:
> Revert of Skip PaintLayer empty paint phases if it previously painted nothing
(patchset #12 id:220001 of https://codereview.chromium.org/1739893006/ )
> 
> Reason for revert:
> BUG=595839
> 
> Original issue's description:
> > Skip PaintLayer empty paint phases if it previously painted nothing
> >
> > This avoids later painting of empty phases if the layer is marked to need
> > a paint phase but the phase painted nothing in the following cases:
> > - the object marking the layer needing the paint phase is not visible
> >   (e.g. because it is clipped)
> > - the layer was marked needing the paint phase by some object that
> >   has been deleted or moved to another layer.
> >
> > Instead of directly clearing the needsPaintPhase flags, this CL adds
> > a new set of flags to indicate if the previous paint of the phases painted
> > nothing. This is because invalidations of the needsPaintPhase flags and the
> > new flags need different conditions. The new flags are cleared when scroll
> > offset, interest rect or clipping changes.
> >
> > Cluster telemetry runs show that this will improve non-cached painting
> > by 1.7% and fully-cached painting (with subsequence-caching disabled) by
> > 45%.
> >
> > BUG=574938
> >
> > Committed: https://crrev.com/8ad6e62af5301a7e8dacd12fc1ae7914258b9e7a
> > Cr-Commit-Position: refs/heads/master@{#380841}
> 
> TBR=chrishtr@chromium.org
> # Not skipping CQ checks because original CL landed more than 1 days ago.
> BUG=574938
> 
> Committed: https://crrev.com/e425e3a8c5204d2d75f34c885c33f077372d1ff8
> Cr-Commit-Position: refs/heads/master@{#382018}

TBR=chrishtr@chromium.org
# Not skipping CQ checks because original CL landed more than 1 days ago.
BUG=595839
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2
==========

to

==========
Reland of Skip PaintLayer empty paint phases if it previously painted nothing
(patchset #2 id:140001 of https://codereview.chromium.org/1815293002/ )

The original CL was reverted because it caused crbug.com/595839.

The reason of the bug was that we painted the background layer (for
fixed background of the viewport) with PaintLayerPaintingRootBackgroundOnly
and PaintLayerPaintingCompositingForegroundPhase flags which caused
unnecessary additional paint walk of the sub layers of the root layer,
which actually painted nothing and mistakenly marked the sub layers not
having paintings for descendant backgrounds, float and descendant
outlines paint phases, causing the normal painting (without the
PaintLayerPaintingRootBackgroundOnly flag) to skip painting of the
paint phases.

Because now we paint the root background in ViewPainter, we no longer
need to traverse the descendants to finish root background painting,
this CL removes the PaintLayerPaintingCompositingForegroundPhase flag
for background layer painting. Also added DCHECK to ensure no
PaintLayerPaintingRootBackgroundOnly flag ever passed to 
PaintLayerPainter::paintForegroundForFragments().

Also added under-invalidation checking for the needsPaintPhaseXXX()
flags by forcing painting even if needsPaintPhaseXXX() is false, and
asserting no actual painting occurs.

Original issue's description:
> Revert of Skip PaintLayer empty paint phases if it previously painted nothing
(patchset #12 id:220001 of https://codereview.chromium.org/1739893006/ )
> 
> Reason for revert:
> BUG=595839
> 
> Original issue's description:
> > Skip PaintLayer empty paint phases if it previously painted nothing
> >
> > This avoids later painting of empty phases if the layer is marked to need
> > a paint phase but the phase painted nothing in the following cases:
> > - the object marking the layer needing the paint phase is not visible
> >   (e.g. because it is clipped)
> > - the layer was marked needing the paint phase by some object that
> >   has been deleted or moved to another layer.
> >
> > Instead of directly clearing the needsPaintPhase flags, this CL adds
> > a new set of flags to indicate if the previous paint of the phases painted
> > nothing. This is because invalidations of the needsPaintPhase flags and the
> > new flags need different conditions. The new flags are cleared when scroll
> > offset, interest rect or clipping changes.
> >
> > Cluster telemetry runs show that this will improve non-cached painting
> > by 1.7% and fully-cached painting (with subsequence-caching disabled) by
> > 45%.
> >
> > BUG=574938
> >
> > Committed: https://crrev.com/8ad6e62af5301a7e8dacd12fc1ae7914258b9e7a
> > Cr-Commit-Position: refs/heads/master@{#380841}
> 
> TBR=chrishtr@chromium.org
> # Not skipping CQ checks because original CL landed more than 1 days ago.
> BUG=574938
> 
> Committed: https://crrev.com/e425e3a8c5204d2d75f34c885c33f077372d1ff8
> Cr-Commit-Position: refs/heads/master@{#382018}

TBR=chrishtr@chromium.org
# Not skipping CQ checks because original CL landed more than 1 days ago.
BUG=595839
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2
==========

[Xianzhu, 2016-08-30 21:36:59]

This is ready for review.

You can compare Patch Set 2 and the latest patch for the new changes made to fix
crbug.com/595839.

[Xianzhu, 2016-08-30 21:48:34]

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

[commit-bot: I haz the power, 2016-08-30 21:49:27]

Dry run: 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, 2016-08-30 23:23:36]

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

[commit-bot: I haz the power, 2016-08-30 23:23:37]

Dry run: This issue passed the CQ dry run.

[chrishtr, 2016-08-31 01:19:50]

Can the change to PaintLayerPaintingCompositingForegroundPhase be committed
independently of
the rest of this CL?

[Xianzhu, 2016-08-31 16:01:06]

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

[commit-bot: I haz the power, 2016-08-31 16:01:48]

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

[Xianzhu, 2016-08-31 16:04:44]

Description was changed from

==========
Reland of Skip PaintLayer empty paint phases if it previously painted nothing
(patchset #2 id:140001 of https://codereview.chromium.org/1815293002/ )

The original CL was reverted because it caused crbug.com/595839.

The reason of the bug was that we painted the background layer (for
fixed background of the viewport) with PaintLayerPaintingRootBackgroundOnly
and PaintLayerPaintingCompositingForegroundPhase flags which caused
unnecessary additional paint walk of the sub layers of the root layer,
which actually painted nothing and mistakenly marked the sub layers not
having paintings for descendant backgrounds, float and descendant
outlines paint phases, causing the normal painting (without the
PaintLayerPaintingRootBackgroundOnly flag) to skip painting of the
paint phases.

Because now we paint the root background in ViewPainter, we no longer
need to traverse the descendants to finish root background painting,
this CL removes the PaintLayerPaintingCompositingForegroundPhase flag
for background layer painting. Also added DCHECK to ensure no
PaintLayerPaintingRootBackgroundOnly flag ever passed to 
PaintLayerPainter::paintForegroundForFragments().

Also added under-invalidation checking for the needsPaintPhaseXXX()
flags by forcing painting even if needsPaintPhaseXXX() is false, and
asserting no actual painting occurs.

Original issue's description:
> Revert of Skip PaintLayer empty paint phases if it previously painted nothing
(patchset #12 id:220001 of https://codereview.chromium.org/1739893006/ )
> 
> Reason for revert:
> BUG=595839
> 
> Original issue's description:
> > Skip PaintLayer empty paint phases if it previously painted nothing
> >
> > This avoids later painting of empty phases if the layer is marked to need
> > a paint phase but the phase painted nothing in the following cases:
> > - the object marking the layer needing the paint phase is not visible
> >   (e.g. because it is clipped)
> > - the layer was marked needing the paint phase by some object that
> >   has been deleted or moved to another layer.
> >
> > Instead of directly clearing the needsPaintPhase flags, this CL adds
> > a new set of flags to indicate if the previous paint of the phases painted
> > nothing. This is because invalidations of the needsPaintPhase flags and the
> > new flags need different conditions. The new flags are cleared when scroll
> > offset, interest rect or clipping changes.
> >
> > Cluster telemetry runs show that this will improve non-cached painting
> > by 1.7% and fully-cached painting (with subsequence-caching disabled) by
> > 45%.
> >
> > BUG=574938
> >
> > Committed: https://crrev.com/8ad6e62af5301a7e8dacd12fc1ae7914258b9e7a
> > Cr-Commit-Position: refs/heads/master@{#380841}
> 
> TBR=chrishtr@chromium.org
> # Not skipping CQ checks because original CL landed more than 1 days ago.
> BUG=574938
> 
> Committed: https://crrev.com/e425e3a8c5204d2d75f34c885c33f077372d1ff8
> Cr-Commit-Position: refs/heads/master@{#382018}

TBR=chrishtr@chromium.org
# Not skipping CQ checks because original CL landed more than 1 days ago.
BUG=595839
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2
==========

to

==========
Reland of Skip PaintLayer empty paint phases if it previously painted nothing
(patchset #2 id:140001 of https://codereview.chromium.org/1815293002/ )

The original CL was reverted because it caused crbug.com/595839.

The reason of the bug was that we also when we painted the root
background we also traverse descendant layers, which actually painted
nothing and mistakenly marked the descendant layers not having paintings
for descendant backgrounds, float and descendant outlines paint phases,
causing the normal painting to skip painting of the paint phases.

https://codereview.chromium.org/2294313002/ fixed the root background
painting issue, so we can reland this CL now.

Also added under-invalidation checking for the needsPaintPhaseXXX()
flags by forcing painting even if needsPaintPhaseXXX() is false, and
asserting no actual painting occurs.

Original issue's description:
> Revert of Skip PaintLayer empty paint phases if it previously painted nothing
(patchset #12 id:220001 of https://codereview.chromium.org/1739893006/ )
> 
> Reason for revert:
> BUG=595839
> 
> Original issue's description:
> > Skip PaintLayer empty paint phases if it previously painted nothing
> >
> > This avoids later painting of empty phases if the layer is marked to need
> > a paint phase but the phase painted nothing in the following cases:
> > - the object marking the layer needing the paint phase is not visible
> >   (e.g. because it is clipped)
> > - the layer was marked needing the paint phase by some object that
> >   has been deleted or moved to another layer.
> >
> > Instead of directly clearing the needsPaintPhase flags, this CL adds
> > a new set of flags to indicate if the previous paint of the phases painted
> > nothing. This is because invalidations of the needsPaintPhase flags and the
> > new flags need different conditions. The new flags are cleared when scroll
> > offset, interest rect or clipping changes.
> >
> > Cluster telemetry runs show that this will improve non-cached painting
> > by 1.7% and fully-cached painting (with subsequence-caching disabled) by
> > 45%.
> >
> > BUG=574938
> >
> > Committed: https://crrev.com/8ad6e62af5301a7e8dacd12fc1ae7914258b9e7a
> > Cr-Commit-Position: refs/heads/master@{#380841}
> 
> TBR=chrishtr@chromium.org
> # Not skipping CQ checks because original CL landed more than 1 days ago.
> BUG=574938
> 
> Committed: https://crrev.com/e425e3a8c5204d2d75f34c885c33f077372d1ff8
> Cr-Commit-Position: refs/heads/master@{#382018}

TBR=chrishtr@chromium.org
# Not skipping CQ checks because original CL landed more than 1 days ago.
BUG=595839
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2
==========

[commit-bot: I haz the power, 2016-08-31 16:05:15]

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

[commit-bot: I haz the power, 2016-08-31 16:05:15]

Dry run: Try jobs failed on following builders:
  ios-device on master.tryserver.chromium.mac (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.mac/builders/ios-device/builds...)
  ios-simulator on master.tryserver.chromium.mac (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.mac/builders/ios-simulator/bui...)
  mac_chromium_compile_dbg_ng on master.tryserver.chromium.mac (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.mac/builders/mac_chromium_comp...)

[Xianzhu, 2016-08-31 16:11:42]

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

[commit-bot: I haz the power, 2016-08-31 16:12:24]

Dry run: 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, 2016-08-31 19:05:47]

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

[commit-bot: I haz the power, 2016-08-31 19:05:48]

Dry run: This issue passed the CQ dry run.

[Xianzhu, 2016-09-01 16:44:20]

On 2016/08/31 01:19:50, chrishtr wrote:
> Can the change to PaintLayerPaintingCompositingForegroundPhase be committed
> independently of
> the rest of this CL?

The CL fixing root background painting is landed:
https://codereview.chromium.org/2294313002/, then this CL is just a relanding of
https://codereview.chromium.org/1739893006/.

[Xianzhu, 2016-09-01 16:44:26]

The CQ bit was checked by wangxianzhu@chromium.org

[commit-bot: I haz the power, 2016-09-01 16:44:42]

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

[chrishtr, 2016-09-01 16:44:44]

lgtm

[commit-bot: I haz the power, 2016-09-01 18:51:54]

Description was changed from

==========
Reland of Skip PaintLayer empty paint phases if it previously painted nothing
(patchset #2 id:140001 of https://codereview.chromium.org/1815293002/ )

The original CL was reverted because it caused crbug.com/595839.

The reason of the bug was that we also when we painted the root
background we also traverse descendant layers, which actually painted
nothing and mistakenly marked the descendant layers not having paintings
for descendant backgrounds, float and descendant outlines paint phases,
causing the normal painting to skip painting of the paint phases.

https://codereview.chromium.org/2294313002/ fixed the root background
painting issue, so we can reland this CL now.

Also added under-invalidation checking for the needsPaintPhaseXXX()
flags by forcing painting even if needsPaintPhaseXXX() is false, and
asserting no actual painting occurs.

Original issue's description:
> Revert of Skip PaintLayer empty paint phases if it previously painted nothing
(patchset #12 id:220001 of https://codereview.chromium.org/1739893006/ )
> 
> Reason for revert:
> BUG=595839
> 
> Original issue's description:
> > Skip PaintLayer empty paint phases if it previously painted nothing
> >
> > This avoids later painting of empty phases if the layer is marked to need
> > a paint phase but the phase painted nothing in the following cases:
> > - the object marking the layer needing the paint phase is not visible
> >   (e.g. because it is clipped)
> > - the layer was marked needing the paint phase by some object that
> >   has been deleted or moved to another layer.
> >
> > Instead of directly clearing the needsPaintPhase flags, this CL adds
> > a new set of flags to indicate if the previous paint of the phases painted
> > nothing. This is because invalidations of the needsPaintPhase flags and the
> > new flags need different conditions. The new flags are cleared when scroll
> > offset, interest rect or clipping changes.
> >
> > Cluster telemetry runs show that this will improve non-cached painting
> > by 1.7% and fully-cached painting (with subsequence-caching disabled) by
> > 45%.
> >
> > BUG=574938
> >
> > Committed: https://crrev.com/8ad6e62af5301a7e8dacd12fc1ae7914258b9e7a
> > Cr-Commit-Position: refs/heads/master@{#380841}
> 
> TBR=chrishtr@chromium.org
> # Not skipping CQ checks because original CL landed more than 1 days ago.
> BUG=574938
> 
> Committed: https://crrev.com/e425e3a8c5204d2d75f34c885c33f077372d1ff8
> Cr-Commit-Position: refs/heads/master@{#382018}

TBR=chrishtr@chromium.org
# Not skipping CQ checks because original CL landed more than 1 days ago.
BUG=595839
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2
==========

to

==========
Reland of Skip PaintLayer empty paint phases if it previously painted nothing
(patchset #2 id:140001 of https://codereview.chromium.org/1815293002/ )

The original CL was reverted because it caused crbug.com/595839.

The reason of the bug was that we also when we painted the root
background we also traverse descendant layers, which actually painted
nothing and mistakenly marked the descendant layers not having paintings
for descendant backgrounds, float and descendant outlines paint phases,
causing the normal painting to skip painting of the paint phases.

https://codereview.chromium.org/2294313002/ fixed the root background
painting issue, so we can reland this CL now.

Also added under-invalidation checking for the needsPaintPhaseXXX()
flags by forcing painting even if needsPaintPhaseXXX() is false, and
asserting no actual painting occurs.

Original issue's description:
> Revert of Skip PaintLayer empty paint phases if it previously painted nothing
(patchset #12 id:220001 of https://codereview.chromium.org/1739893006/ )
> 
> Reason for revert:
> BUG=595839
> 
> Original issue's description:
> > Skip PaintLayer empty paint phases if it previously painted nothing
> >
> > This avoids later painting of empty phases if the layer is marked to need
> > a paint phase but the phase painted nothing in the following cases:
> > - the object marking the layer needing the paint phase is not visible
> >   (e.g. because it is clipped)
> > - the layer was marked needing the paint phase by some object that
> >   has been deleted or moved to another layer.
> >
> > Instead of directly clearing the needsPaintPhase flags, this CL adds
> > a new set of flags to indicate if the previous paint of the phases painted
> > nothing. This is because invalidations of the needsPaintPhase flags and the
> > new flags need different conditions. The new flags are cleared when scroll
> > offset, interest rect or clipping changes.
> >
> > Cluster telemetry runs show that this will improve non-cached painting
> > by 1.7% and fully-cached painting (with subsequence-caching disabled) by
> > 45%.
> >
> > BUG=574938
> >
> > Committed: https://crrev.com/8ad6e62af5301a7e8dacd12fc1ae7914258b9e7a
> > Cr-Commit-Position: refs/heads/master@{#380841}
> 
> TBR=chrishtr@chromium.org
> # Not skipping CQ checks because original CL landed more than 1 days ago.
> BUG=574938
> 
> Committed: https://crrev.com/e425e3a8c5204d2d75f34c885c33f077372d1ff8
> Cr-Commit-Position: refs/heads/master@{#382018}

TBR=chrishtr@chromium.org
# Not skipping CQ checks because original CL landed more than 1 days ago.
BUG=595839
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2
==========

[commit-bot: I haz the power, 2016-09-01 18:51:55]

Committed patchset #9 (id:270001)

[commit-bot: I haz the power, 2016-09-01 18:53:59]

Description was changed from

==========
Reland of Skip PaintLayer empty paint phases if it previously painted nothing
(patchset #2 id:140001 of https://codereview.chromium.org/1815293002/ )

The original CL was reverted because it caused crbug.com/595839.

The reason of the bug was that we also when we painted the root
background we also traverse descendant layers, which actually painted
nothing and mistakenly marked the descendant layers not having paintings
for descendant backgrounds, float and descendant outlines paint phases,
causing the normal painting to skip painting of the paint phases.

https://codereview.chromium.org/2294313002/ fixed the root background
painting issue, so we can reland this CL now.

Also added under-invalidation checking for the needsPaintPhaseXXX()
flags by forcing painting even if needsPaintPhaseXXX() is false, and
asserting no actual painting occurs.

Original issue's description:
> Revert of Skip PaintLayer empty paint phases if it previously painted nothing
(patchset #12 id:220001 of https://codereview.chromium.org/1739893006/ )
> 
> Reason for revert:
> BUG=595839
> 
> Original issue's description:
> > Skip PaintLayer empty paint phases if it previously painted nothing
> >
> > This avoids later painting of empty phases if the layer is marked to need
> > a paint phase but the phase painted nothing in the following cases:
> > - the object marking the layer needing the paint phase is not visible
> >   (e.g. because it is clipped)
> > - the layer was marked needing the paint phase by some object that
> >   has been deleted or moved to another layer.
> >
> > Instead of directly clearing the needsPaintPhase flags, this CL adds
> > a new set of flags to indicate if the previous paint of the phases painted
> > nothing. This is because invalidations of the needsPaintPhase flags and the
> > new flags need different conditions. The new flags are cleared when scroll
> > offset, interest rect or clipping changes.
> >
> > Cluster telemetry runs show that this will improve non-cached painting
> > by 1.7% and fully-cached painting (with subsequence-caching disabled) by
> > 45%.
> >
> > BUG=574938
> >
> > Committed: https://crrev.com/8ad6e62af5301a7e8dacd12fc1ae7914258b9e7a
> > Cr-Commit-Position: refs/heads/master@{#380841}
> 
> TBR=chrishtr@chromium.org
> # Not skipping CQ checks because original CL landed more than 1 days ago.
> BUG=574938
> 
> Committed: https://crrev.com/e425e3a8c5204d2d75f34c885c33f077372d1ff8
> Cr-Commit-Position: refs/heads/master@{#382018}

TBR=chrishtr@chromium.org
# Not skipping CQ checks because original CL landed more than 1 days ago.
BUG=595839
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2
==========

to

==========
Reland of Skip PaintLayer empty paint phases if it previously painted nothing
(patchset #2 id:140001 of https://codereview.chromium.org/1815293002/ )

The original CL was reverted because it caused crbug.com/595839.

The reason of the bug was that we also when we painted the root
background we also traverse descendant layers, which actually painted
nothing and mistakenly marked the descendant layers not having paintings
for descendant backgrounds, float and descendant outlines paint phases,
causing the normal painting to skip painting of the paint phases.

https://codereview.chromium.org/2294313002/ fixed the root background
painting issue, so we can reland this CL now.

Also added under-invalidation checking for the needsPaintPhaseXXX()
flags by forcing painting even if needsPaintPhaseXXX() is false, and
asserting no actual painting occurs.

Original issue's description:
> Revert of Skip PaintLayer empty paint phases if it previously painted nothing
(patchset #12 id:220001 of https://codereview.chromium.org/1739893006/ )
>
> Reason for revert:
> BUG=595839
>
> Original issue's description:
> > Skip PaintLayer empty paint phases if it previously painted nothing
> >
> > This avoids later painting of empty phases if the layer is marked to need
> > a paint phase but the phase painted nothing in the following cases:
> > - the object marking the layer needing the paint phase is not visible
> >   (e.g. because it is clipped)
> > - the layer was marked needing the paint phase by some object that
> >   has been deleted or moved to another layer.
> >
> > Instead of directly clearing the needsPaintPhase flags, this CL adds
> > a new set of flags to indicate if the previous paint of the phases painted
> > nothing. This is because invalidations of the needsPaintPhase flags and the
> > new flags need different conditions. The new flags are cleared when scroll
> > offset, interest rect or clipping changes.
> >
> > Cluster telemetry runs show that this will improve non-cached painting
> > by 1.7% and fully-cached painting (with subsequence-caching disabled) by
> > 45%.
> >
> > BUG=574938
> >
> > Committed: https://crrev.com/8ad6e62af5301a7e8dacd12fc1ae7914258b9e7a
> > Cr-Commit-Position: refs/heads/master@{#380841}
>
> TBR=chrishtr@chromium.org
> # Not skipping CQ checks because original CL landed more than 1 days ago.
> BUG=574938
>
> Committed: https://crrev.com/e425e3a8c5204d2d75f34c885c33f077372d1ff8
> Cr-Commit-Position: refs/heads/master@{#382018}

TBR=chrishtr@chromium.org
# Not skipping CQ checks because original CL landed more than 1 days ago.
BUG=595839
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2

Committed: https://crrev.com/a6378700ce7244fa1da85321aeb31661cde0ff16
Cr-Commit-Position: refs/heads/master@{#416008}
==========

[commit-bot: I haz the power, 2016-09-01 18:54:01]

Patchset 9 (id:??) landed as
https://crrev.com/a6378700ce7244fa1da85321aeb31661cde0ff16
Cr-Commit-Position: refs/heads/master@{#416008}