Use Blink's computed filterPrimitiveSubregion() as CropRect

Use Blink's computed filterPrimitiveSubregion() as CropRect.

For filters which affect transparent black, we were falling back to Skia's clip rect to know which pixels to fill. This works for SVG, where we set the clip rect to the filter region. However, this is not possible for CSS filters on HTML content, where there may be multiple reference filters with different filter regions, so we don't actually set a clip rect at all.

The easiest fix is to set the crop rect unconditionally for all filters with a non-empty filter primitive subregion (aka all reference filters), which is what we do here. This also gets us closer to being able to removing the individual crop edges from Skia. This has minor pixel diffs in the affected tests.

BUG=599933
Committed: https://crrev.com/0abb8cd84553174c24442b173f63110198145e81
Cr-Commit-Position: refs/heads/master@{#385031}

[Stephen White, 2016-04-02 15:55:38]

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

[commit-bot: I haz the power, 2016-04-02 15:55:52]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1840303006/20001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1840303006/20001

[commit-bot: I haz the power, 2016-04-02 16:56:07]

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

[commit-bot: I haz the power, 2016-04-02 16:56:08]

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

[Stephen White, 2016-04-04 21:01:06]

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

[commit-bot: I haz the power, 2016-04-04 21:01:46]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1840303006/40001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1840303006/40001

[Stephen White, 2016-04-04 21:03:52]

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

[commit-bot: I haz the power, 2016-04-04 21:04:13]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1840303006/60001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1840303006/60001

[Stephen White, 2016-04-04 21:06:12]

Description was changed from

==========
Always use Blink's computed filterPrimitiveSubregion().

BUG=
==========

to

==========
Always use Blink's computed filterPrimitiveSubregion().

BUG=599933
==========

[Stephen White, 2016-04-04 21:06:12]

senorblanco@chromium.org changed reviewers:
+ fmalita@chromium.org

[Stephen White, 2016-04-04 21:12:09]

Description was changed from

==========
Always use Blink's computed filterPrimitiveSubregion().

BUG=599933
==========

to

==========
Use Blink's computed filterPrimitiveSubregion() as CropRect.

For filters which affect transparent black, we were falling back to Skia's clip
rect to know which pixels to fill. This works for SVG, where we set the clip
rect to the filter region. However, this is not possible for CSS filters on HTML
content, where there may be multiple reference filters with different filter
regions, so we don't actually set a clip rect at all.

The easiest fix is to set the crop rect unconditionally for all filters with a
non-empty filter primitive subregion (aka all reference filters), which is what
we do here. This has minor pixel diffs in the affected tests.

BUG=599933
==========

[Stephen White, 2016-04-04 21:12:47]

Description was changed from

==========
Use Blink's computed filterPrimitiveSubregion() as CropRect.

For filters which affect transparent black, we were falling back to Skia's clip
rect to know which pixels to fill. This works for SVG, where we set the clip
rect to the filter region. However, this is not possible for CSS filters on HTML
content, where there may be multiple reference filters with different filter
regions, so we don't actually set a clip rect at all.

The easiest fix is to set the crop rect unconditionally for all filters with a
non-empty filter primitive subregion (aka all reference filters), which is what
we do here. This has minor pixel diffs in the affected tests.

BUG=599933
==========

to

==========
Use Blink's computed filterPrimitiveSubregion() as CropRect.

For filters which affect transparent black, we were falling back to Skia's clip
rect to know which pixels to fill. This works for SVG, where we set the clip
rect to the filter region. However, this is not possible for CSS filters on HTML
content, where there may be multiple reference filters with different filter
regions, so we don't actually set a clip rect at all.

The easiest fix is to set the crop rect unconditionally for all filters with a
non-empty filter primitive subregion (aka all reference filters), which is what
we do here. This also gets us closer to being able to removing the individual
crop edges from Skia. This has minor pixel diffs in the affected tests.

BUG=599933
==========

[Stephen White, 2016-04-04 21:13:15]

Florin: PTAL. Thanks!

[f(malita), 2016-04-04 21:21:20]

LGTM

[commit-bot: I haz the power, 2016-04-04 22:22:00]

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

[commit-bot: I haz the power, 2016-04-04 22:22:01]

Dry run: This issue passed the CQ dry run.

[Stephen White, 2016-04-04 22:23:29]

The CQ bit was checked by senorblanco@chromium.org

[commit-bot: I haz the power, 2016-04-04 22:24:02]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1840303006/60001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1840303006/60001

[commit-bot: I haz the power, 2016-04-04 22:33:07]

Description was changed from

==========
Use Blink's computed filterPrimitiveSubregion() as CropRect.

For filters which affect transparent black, we were falling back to Skia's clip
rect to know which pixels to fill. This works for SVG, where we set the clip
rect to the filter region. However, this is not possible for CSS filters on HTML
content, where there may be multiple reference filters with different filter
regions, so we don't actually set a clip rect at all.

The easiest fix is to set the crop rect unconditionally for all filters with a
non-empty filter primitive subregion (aka all reference filters), which is what
we do here. This also gets us closer to being able to removing the individual
crop edges from Skia. This has minor pixel diffs in the affected tests.

BUG=599933
==========

to

==========
Use Blink's computed filterPrimitiveSubregion() as CropRect.

For filters which affect transparent black, we were falling back to Skia's clip
rect to know which pixels to fill. This works for SVG, where we set the clip
rect to the filter region. However, this is not possible for CSS filters on HTML
content, where there may be multiple reference filters with different filter
regions, so we don't actually set a clip rect at all.

The easiest fix is to set the crop rect unconditionally for all filters with a
non-empty filter primitive subregion (aka all reference filters), which is what
we do here. This also gets us closer to being able to removing the individual
crop edges from Skia. This has minor pixel diffs in the affected tests.

BUG=599933
==========

[commit-bot: I haz the power, 2016-04-04 22:33:08]

Committed patchset #4 (id:60001)

[commit-bot: I haz the power, 2016-04-04 22:34:35]

Description was changed from

==========
Use Blink's computed filterPrimitiveSubregion() as CropRect.

For filters which affect transparent black, we were falling back to Skia's clip
rect to know which pixels to fill. This works for SVG, where we set the clip
rect to the filter region. However, this is not possible for CSS filters on HTML
content, where there may be multiple reference filters with different filter
regions, so we don't actually set a clip rect at all.

The easiest fix is to set the crop rect unconditionally for all filters with a
non-empty filter primitive subregion (aka all reference filters), which is what
we do here. This also gets us closer to being able to removing the individual
crop edges from Skia. This has minor pixel diffs in the affected tests.

BUG=599933
==========

to

==========
Use Blink's computed filterPrimitiveSubregion() as CropRect.

For filters which affect transparent black, we were falling back to Skia's clip
rect to know which pixels to fill. This works for SVG, where we set the clip
rect to the filter region. However, this is not possible for CSS filters on HTML
content, where there may be multiple reference filters with different filter
regions, so we don't actually set a clip rect at all.

The easiest fix is to set the crop rect unconditionally for all filters with a
non-empty filter primitive subregion (aka all reference filters), which is what
we do here. This also gets us closer to being able to removing the individual
crop edges from Skia. This has minor pixel diffs in the affected tests.

BUG=599933

Committed: https://crrev.com/0abb8cd84553174c24442b173f63110198145e81
Cr-Commit-Position: refs/heads/master@{#385031}
==========

[commit-bot: I haz the power, 2016-04-04 22:34:36]

Patchset 4 (id:??) landed as
https://crrev.com/0abb8cd84553174c24442b173f63110198145e81
Cr-Commit-Position: refs/heads/master@{#385031}