Fix clip and offset issues of slimmingPaintInvalidation for SPv1

Fix clip, offset and filter issues of slimmingPaintInvalidation for SPv1

The filter fix temporarily uses the slow path before we have full
support of filter in paint property tree.

Test: run-webkit-tests --additional-driver-flag=--enable-blink-features=slimmingPaintInvalidation paint/invalidation
Ran 838 tests. 2 crashes, 99 failures. (55 less failures/crashes than
https://codereview.chromium.org/2334113002/).

BUG=646176
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2

Committed: https://crrev.com/0eae31ff1f741c5f1b913589c35993a23e7efff9
Cr-Commit-Position: refs/heads/master@{#418451}

[Xianzhu, 2016-09-13 17:40:30]

Description was changed from

==========
Fix clip and offset issues of slimmingPaintInvalidation for SPv1

Test: run-webkit-tests
--additional-driver-flag=--enable-blink-features=slimmingPaintInvalidation
paint/invalidation
Ran 838 tests. 2 crashes, 113 failures. (41 less failures/crashes than
https://codereview.chromium.org/2334113002/).

BUG=646176
==========

to

==========
Fix clip and offset issues of slimmingPaintInvalidation for SPv1

Test: run-webkit-tests
--additional-driver-flag=--enable-blink-features=slimmingPaintInvalidation
paint/invalidation
Ran 838 tests. 2 crashes, 113 failures. (41 less failures/crashes than
https://codereview.chromium.org/2334113002/).

BUG=646176
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2
==========

[Xianzhu, 2016-09-13 17:42:33]

wangxianzhu@chromium.org changed reviewers:
+ chrishtr@chromium.org

[Xianzhu, 2016-09-13 17:42:33]

https://codereview.chromium.org/2332353003/diff/1/third_party/WebKit/Source/c...
File third_party/WebKit/Source/core/paint/PaintInvalidator.cpp (right):

https://codereview.chromium.org/2332353003/diff/1/third_party/WebKit/Source/c...
third_party/WebKit/Source/core/paint/PaintInvalidator.cpp:210: 
This and the changes below are copied from
LayoutBoxMobelObject::invalidateTreeIfNeeded().

[Xianzhu, 2016-09-13 18:25:59]

Description was changed from

==========
Fix clip and offset issues of slimmingPaintInvalidation for SPv1

Test: run-webkit-tests
--additional-driver-flag=--enable-blink-features=slimmingPaintInvalidation
paint/invalidation
Ran 838 tests. 2 crashes, 113 failures. (41 less failures/crashes than
https://codereview.chromium.org/2334113002/).

BUG=646176
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2
==========

to

==========
Fix clip, offset and filter issues of slimmingPaintInvalidation for SPv1

The filter fix temporarily uses the slow path before we have full
support of filter in paint property tree.

Test: run-webkit-tests
--additional-driver-flag=--enable-blink-features=slimmingPaintInvalidation
paint/invalidation
Ran 838 tests. 2 crashes, 99 failures. (55 less failures/crashes than
https://codereview.chromium.org/2334113002/).

BUG=646176
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2
==========

[chrishtr, 2016-09-13 23:30:04]

lgtm

https://codereview.chromium.org/2332353003/diff/20001/third_party/WebKit/Sour...
File third_party/WebKit/Source/core/paint/PaintInvalidator.cpp (right):

https://codereview.chromium.org/2332353003/diff/20001/third_party/WebKit/Sour...
third_party/WebKit/Source/core/paint/PaintInvalidator.cpp:58:
result.moveBy(-containerPaintProperties->localBorderBoxProperties()->paintOffset);
This was a straight-up bug in slimmingPaintInvalidation? Why are you computing
the diff of the paint
offsets?

https://codereview.chromium.org/2332353003/diff/20001/third_party/WebKit/Sour...
third_party/WebKit/Source/core/paint/PaintInvalidator.cpp:222: ||
(style.transformOriginX() != Length(50, Percent) &&
style.transformOriginX().isPercentOrCalc())
There's no canonical way to detect non-default transform origin?

[chrishtr, 2016-09-13 23:30:20]

Sorry, accidental LGTM. Please see my comments.

[Xianzhu, 2016-09-13 23:47:15]

https://codereview.chromium.org/2332353003/diff/20001/third_party/WebKit/Sour...
File third_party/WebKit/Source/core/paint/PaintInvalidator.cpp (right):

https://codereview.chromium.org/2332353003/diff/20001/third_party/WebKit/Sour...
third_party/WebKit/Source/core/paint/PaintInvalidator.cpp:58:
result.moveBy(-containerPaintProperties->localBorderBoxProperties()->paintOffset);
On 2016/09/13 23:30:03, chrishtr wrote:
> This was a straight-up bug in slimmingPaintInvalidation? Why are you computing
> the diff of the paint
> offsets?

Yes, this was a bug, but didn't notice it because there are too many failures on
spv2. 

Now with it running on spv1 I discovered this from failures of a much smaller
number.

After Line 57 we get a rect in the space of container's PaintTreeStatus. The
rect still includes container's paint offset of the container which should be
subtracted from the rect to get a rect in container's space.

https://codereview.chromium.org/2332353003/diff/20001/third_party/WebKit/Sour...
third_party/WebKit/Source/core/paint/PaintInvalidator.cpp:222: ||
(style.transformOriginX() != Length(50, Percent) &&
style.transformOriginX().isPercentOrCalc())
On 2016/09/13 23:30:03, chrishtr wrote:
> There's no canonical way to detect non-default transform origin?

We only need to check non-default percentage transform origin here.

This code is copied from LayoutBoxModelObject.cpp and is a workaround of
crbug.com/533277 (see line 265 below) and will be removed when we implement
paint offset change detection.

[chrishtr, 2016-09-13 23:54:24]

The CQ bit was checked by chrishtr@chromium.org

[chrishtr, 2016-09-13 23:54:25]

lgtm

https://codereview.chromium.org/2332353003/diff/20001/third_party/WebKit/Sour...
File third_party/WebKit/Source/core/paint/PaintInvalidator.cpp (right):

https://codereview.chromium.org/2332353003/diff/20001/third_party/WebKit/Sour...
third_party/WebKit/Source/core/paint/PaintInvalidator.cpp:58:
result.moveBy(-containerPaintProperties->localBorderBoxProperties()->paintOffset);
On 2016/09/13 at 23:47:15, Xianzhu wrote:
> On 2016/09/13 23:30:03, chrishtr wrote:
> > This was a straight-up bug in slimmingPaintInvalidation? Why are you
computing
> > the diff of the paint
> > offsets?
> 
> Yes, this was a bug, but didn't notice it because there are too many failures
on spv2. 
> 
> Now with it running on spv1 I discovered this from failures of a much smaller
number.
> 
> After Line 57 we get a rect in the space of container's PaintTreeStatus. The
rect still includes container's paint offset of the container which should be
subtracted from the rect to get a rect in container's space.

Ah yes, makes sense.

[commit-bot: I haz the power, 2016-09-13 23:54:55]

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, 2016-09-14 01:40:24]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2016-09-14 01:42:37]

Description was changed from

==========
Fix clip, offset and filter issues of slimmingPaintInvalidation for SPv1

The filter fix temporarily uses the slow path before we have full
support of filter in paint property tree.

Test: run-webkit-tests
--additional-driver-flag=--enable-blink-features=slimmingPaintInvalidation
paint/invalidation
Ran 838 tests. 2 crashes, 99 failures. (55 less failures/crashes than
https://codereview.chromium.org/2334113002/).

BUG=646176
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2
==========

to

==========
Fix clip, offset and filter issues of slimmingPaintInvalidation for SPv1

The filter fix temporarily uses the slow path before we have full
support of filter in paint property tree.

Test: run-webkit-tests
--additional-driver-flag=--enable-blink-features=slimmingPaintInvalidation
paint/invalidation
Ran 838 tests. 2 crashes, 99 failures. (55 less failures/crashes than
https://codereview.chromium.org/2334113002/).

BUG=646176
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2

Committed: https://crrev.com/0eae31ff1f741c5f1b913589c35993a23e7efff9
Cr-Commit-Position: refs/heads/master@{#418451}
==========

[commit-bot: I haz the power, 2016-09-14 01:42:38]

Patchset 2 (id:??) landed as
https://crrev.com/0eae31ff1f741c5f1b913589c35993a23e7efff9
Cr-Commit-Position: refs/heads/master@{#418451}