Determine mask UVs based on texture size

Determine mask UVs based on texture size

Currently, mask SINGLE_TEXTURE_MASK UVs are determined solely by the size
of the masked layer. This means that if the mask texture size does not match the
expected texture size, we will end up squishing the mask when it's applied.

This CL adds a new out param, |resource_uv_size| to GetContentsResourceId. This
param represents the UV coordinate size of the mask contents within its texture.

This factor is used to adjust the existing calculation, allowing for non-exact mask
texture sizes.


R=sunxd@chromium.org, enne@chromium.org
BUG=713872
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel;master.tryserver.chromium.android:android_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel

Review-Url: https://codereview.chromium.org/2828353003
Cr-Commit-Position: refs/heads/master@{#469464}
Committed: https://chromium.googlesource.com/chromium/src/+/664405a74320db4031a18694358f7d4fef327662

[ericrk, 2017-04-20 21:03:43]

Description was changed from

==========
Determine mask UVs based on tex coords

Currently, mask SINGLE_TEXTURE_MASK UVs are determined based on the size
of the masked layer. This means that if the mask layer size mismatches
the masked layer size, the mask is stretched to cover the masked layer.

This behavior is currently unused. The only uses of mask layers come
from Blink, and all uses in Blink appear to match the mask layer size to
the masked layer size.

Additionally, this behavior does not match MULTI_TEXTURE_MASKs, which
do not stretch to cover the masked layer.

As this behavior is unused and inconsistent, this change simplifies
masking logic to determine mask UVs based on mask texture size. This
has the added benefit of ensuring that masks can be used with texture
sizes that don't exactly match the mask layer size.

R=sunxd@chromium.org, enne@chromium.org
BUG=713872
==========

to

==========
Determine mask UVs based on tex coords

Currently, mask SINGLE_TEXTURE_MASK UVs are determined based on the size
of the masked layer. This means that if the mask layer size mismatches
the masked layer size, the mask is stretched to cover the masked layer.

This behavior is currently unused. The only uses of mask layers come
from Blink, and all uses in Blink appear to match the mask layer size to
the masked layer size.

Additionally, this behavior does not match MULTI_TEXTURE_MASKs, which
do not stretch to cover the masked layer.

As this behavior is unused and inconsistent, this change simplifies
masking logic to determine mask UVs based on mask texture size. This
has the added benefit of ensuring that masks can be used with texture
sizes that don't exactly match the mask layer size.

R=sunxd@chromium.org, enne@chromium.org
BUG=713872
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel
==========

[ericrk, 2017-04-20 21:04:53]

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

[commit-bot: I haz the power, 2017-04-20 21:05:37]

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

[ericrk, 2017-04-20 21:52:22]

This is the precursor to allowing CC tiles to re-use larger resources for
raster.

[ericrk, 2017-04-20 21:54:06]

Description was changed from

==========
Determine mask UVs based on tex coords

Currently, mask SINGLE_TEXTURE_MASK UVs are determined based on the size
of the masked layer. This means that if the mask layer size mismatches
the masked layer size, the mask is stretched to cover the masked layer.

This behavior is currently unused. The only uses of mask layers come
from Blink, and all uses in Blink appear to match the mask layer size to
the masked layer size.

Additionally, this behavior does not match MULTI_TEXTURE_MASKs, which
do not stretch to cover the masked layer.

As this behavior is unused and inconsistent, this change simplifies
masking logic to determine mask UVs based on mask texture size. This
has the added benefit of ensuring that masks can be used with texture
sizes that don't exactly match the mask layer size.

R=sunxd@chromium.org, enne@chromium.org
BUG=713872
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel
==========

to

==========
Determine mask UVs based on texture size

Currently, mask SINGLE_TEXTURE_MASK UVs are determined based on the size
of the masked layer. This means that if the mask layer size mismatches
the masked layer size, the mask is stretched to cover the masked layer.

This behavior is currently unused. The only uses of mask layers come
from Blink, and all uses in Blink appear to match the mask layer size to
the masked layer size.

Additionally, this behavior does not match MULTI_TEXTURE_MASKs, which
do not stretch to cover the masked layer.

As this behavior is unused and inconsistent, this change simplifies
masking logic to determine mask UVs based on mask texture size. This
has the added benefit of ensuring that masks can be used with texture
sizes that don't exactly match the mask layer size.

R=sunxd@chromium.org, enne@chromium.org
BUG=713872
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel
==========

[commit-bot: I haz the power, 2017-04-20 22:23:02]

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

[commit-bot: I haz the power, 2017-04-20 22:23:04]

Dry run: Try jobs failed on following builders:
  android_compile_dbg on master.tryserver.chromium.android (JOB_FAILED,
https://build.chromium.org/p/tryserver.chromium.android/builders/android_comp...)

[enne (OOO), 2017-04-20 22:44:08]

lgtm

Thanks for all the investigation.

https://codereview.chromium.org/2828353003/diff/1/cc/trees/layer_tree_host_pi...
File cc/trees/layer_tree_host_pixeltest_masks.cc (right):

https://codereview.chromium.org/2828353003/diff/1/cc/trees/layer_tree_host_pi...
cc/trees/layer_tree_host_pixeltest_masks.cc:151:
TEST_P(LayerTreeHostMasksPixelTest, MaskOfLargerLayer) {
Thanks for the test!

[sunxd, 2017-04-21 16:08:07]

On 2017/04/20 22:44:08, enne wrote:
> lgtm
> 
> Thanks for all the investigation.
> 
>
https://codereview.chromium.org/2828353003/diff/1/cc/trees/layer_tree_host_pi...
> File cc/trees/layer_tree_host_pixeltest_masks.cc (right):
> 
>
https://codereview.chromium.org/2828353003/diff/1/cc/trees/layer_tree_host_pi...
> cc/trees/layer_tree_host_pixeltest_masks.cc:151:
> TEST_P(LayerTreeHostMasksPixelTest, MaskOfLargerLayer) {
> Thanks for the test!

lgtm. I think now the logic how we scale mask uv rect from coverage space to
normalized space is unified between single and multi texture masks.

[ericrk, 2017-04-21 20:39:00]

Description was changed from

==========
Determine mask UVs based on texture size

Currently, mask SINGLE_TEXTURE_MASK UVs are determined based on the size
of the masked layer. This means that if the mask layer size mismatches
the masked layer size, the mask is stretched to cover the masked layer.

This behavior is currently unused. The only uses of mask layers come
from Blink, and all uses in Blink appear to match the mask layer size to
the masked layer size.

Additionally, this behavior does not match MULTI_TEXTURE_MASKs, which
do not stretch to cover the masked layer.

As this behavior is unused and inconsistent, this change simplifies
masking logic to determine mask UVs based on mask texture size. This
has the added benefit of ensuring that masks can be used with texture
sizes that don't exactly match the mask layer size.

R=sunxd@chromium.org, enne@chromium.org
BUG=713872
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel
==========

to

==========
Determine mask UVs based on texture size

Currently, mask SINGLE_TEXTURE_MASK UVs are determined based on the size
of the masked layer. This means that if the mask layer size mismatches
the masked layer size, the mask is stretched to cover the masked layer.

This behavior is currently unused. The only uses of mask layers come
from Blink, and all uses in Blink appear to match the mask layer size to
the masked layer size.

Additionally, this behavior does not match MULTI_TEXTURE_MASKs, which
do not stretch to cover the masked layer.

As this behavior is unused and inconsistent, this change simplifies
masking logic to determine mask UVs based on mask texture size. This
has the added benefit of ensuring that masks can be used with texture
sizes that don't exactly match the mask layer size.

R=sunxd@chromium.org, enne@chromium.org
BUG=713872
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel;master.tryserver.chromium.android:android_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel
==========

[ericrk, 2017-04-21 20:58:22]

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

[commit-bot: I haz the power, 2017-04-21 20:58:40]

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, 2017-04-21 22:28:27]

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

[commit-bot: I haz the power, 2017-04-21 22:28:28]

Dry run: This issue passed the CQ dry run.

[ericrk, 2017-05-02 00:28:35]

Patchset #3 (id:40001) has been deleted

[ericrk, 2017-05-02 01:59:55]

Description was changed from

==========
Determine mask UVs based on texture size

Currently, mask SINGLE_TEXTURE_MASK UVs are determined based on the size
of the masked layer. This means that if the mask layer size mismatches
the masked layer size, the mask is stretched to cover the masked layer.

This behavior is currently unused. The only uses of mask layers come
from Blink, and all uses in Blink appear to match the mask layer size to
the masked layer size.

Additionally, this behavior does not match MULTI_TEXTURE_MASKs, which
do not stretch to cover the masked layer.

As this behavior is unused and inconsistent, this change simplifies
masking logic to determine mask UVs based on mask texture size. This
has the added benefit of ensuring that masks can be used with texture
sizes that don't exactly match the mask layer size.

R=sunxd@chromium.org, enne@chromium.org
BUG=713872
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel;master.tryserver.chromium.android:android_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel
==========

to

==========
Determine mask UVs based on texture size

Currently, mask SINGLE_TEXTURE_MASK UVs are determined solely by the size
of the masked layer. This means that if the mask texture size does not match the
expected texture size, we will end up squishing the mask when it's applied.

This CL adds a new out param, |resource_uv_size| to GetContentsResourceId. This
param represents the UV coordinate size of the mask contents within its texture.

This factor is used to adjust the existing calculation, allowing for non-exact
mask
texture sizes.


R=sunxd@chromium.org, enne@chromium.org
BUG=713872
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel;master.tryserver.chromium.android:android_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel
==========

[ericrk, 2017-05-02 02:00:15]

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

[commit-bot: I haz the power, 2017-05-02 02:00:44]

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, 2017-05-02 02:22:14]

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

[commit-bot: I haz the power, 2017-05-02 02:22:15]

Dry run: Try jobs failed on following builders:
  android_n5x_swarming_rel on master.tryserver.chromium.android (JOB_FAILED,
https://build.chromium.org/p/tryserver.chromium.android/builders/android_n5x_...)
  linux_android_rel_ng on master.tryserver.chromium.android (JOB_FAILED,
https://build.chromium.org/p/tryserver.chromium.android/builders/linux_androi...)

[ericrk, 2017-05-02 17:52:25]

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

[commit-bot: I haz the power, 2017-05-02 17:52:47]

Dry run: CQ is trying da patch.

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

[ericrk, 2017-05-02 18:27:27]

Found some issues with compositor transforms in my last approach. I think
there's a way to get this working while maintaining the "don't stretch the mask
to the contents" approach, but the logic was getting pretty complicated.

I ended up moving to a more straightforward (and easy to verify) approach. In
this approach, we keep our previous behavior, but adjust the UV coords based on
a "desired texture size" / "actual texture size" transform, which compensates
for non-exact textures.

PTAL, thanks!

[commit-bot: I haz the power, 2017-05-02 18:54:06]

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

[commit-bot: I haz the power, 2017-05-02 18:54:07]

Dry run: Try jobs failed on following builders:
  win_chromium_x64_rel_ng on master.tryserver.chromium.win (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.win/builders/win_chromium_x64_...)

[ericrk, 2017-05-02 20:59:52]

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

[commit-bot: I haz the power, 2017-05-02 21:00:17]

Dry run: CQ is trying da patch.

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

[enne (OOO), 2017-05-02 21:31:38]

The new pixel test here is just verifying the behavior when the mask is a
different size than the layer it's masking, but in that case texture rect ==
resource size?

Is it possible to add a similar pixel test to verify that the output when
texture rect != resource size.  You have a unit test, but I'd feel more comfy in
terms of regression testing if there was a unit test for it too.  Maybe it could
create a resource at 2x the size so that it could be represented exactly in
floats and reuse the same expectations?

https://codereview.chromium.org/2828353003/diff/120001/cc/layers/picture_laye...
File cc/layers/picture_layer_impl.cc (right):

https://codereview.chromium.org/2828353003/diff/120001/cc/layers/picture_laye...
cc/layers/picture_layer_impl.cc:881: // coordinatesis are (1.0f, 1.0f). However,
if we are rasterizing to an
typo

https://codereview.chromium.org/2828353003/diff/120001/cc/layers/picture_laye...
cc/layers/picture_layer_impl.cc:885:
gfx::SizeF(iter->tiling()->tiling_data()->tiling_size());
Is this the same as bounds()?

https://codereview.chromium.org/2828353003/diff/120001/cc/trees/layer_tree_ho...
File cc/trees/layer_tree_host_pixeltest_masks.cc (right):

https://codereview.chromium.org/2828353003/diff/120001/cc/trees/layer_tree_ho...
cc/trees/layer_tree_host_pixeltest_masks.cc:159: gfx::Size mask_bounds(40, 40);
Feel free to take or leave this suggestion, but if you made the mask bounds 50,
50, would that make it stretch but line up on pixel boundaries, so all tests
ended with the same result?

[commit-bot: I haz the power, 2017-05-02 22:58:02]

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

[commit-bot: I haz the power, 2017-05-02 22:58:03]

Dry run: This issue passed the CQ dry run.

[sunxd, 2017-05-03 15:14:13]

https://codereview.chromium.org/2828353003/diff/120001/cc/trees/layer_tree_ho...
File cc/trees/layer_tree_host_pixeltest_masks.cc (right):

https://codereview.chromium.org/2828353003/diff/120001/cc/trees/layer_tree_ho...
cc/trees/layer_tree_host_pixeltest_masks.cc:164:
mask->SetLayerMaskType(Layer::LayerMaskType::SINGLE_TEXTURE_MASK);
I made a recent change that makes pixeltest tests both SINGLE_TEXTURE_MASK and
MULTI_TEXTURE_MASK.

Can you change this line to:

mask->SetLayerMaskType(mask_type_);

[ericrk, 2017-05-04 19:23:53]

Updated. Thanks!

https://codereview.chromium.org/2828353003/diff/120001/cc/layers/picture_laye...
File cc/layers/picture_layer_impl.cc (right):

https://codereview.chromium.org/2828353003/diff/120001/cc/layers/picture_laye...
cc/layers/picture_layer_impl.cc:881: // coordinatesis are (1.0f, 1.0f). However,
if we are rasterizing to an
On 2017/05/02 21:31:37, enne wrote:
> typo

Done.

https://codereview.chromium.org/2828353003/diff/120001/cc/layers/picture_laye...
cc/layers/picture_layer_impl.cc:885:
gfx::SizeF(iter->tiling()->tiling_data()->tiling_size());
On 2017/05/02 21:31:37, enne wrote:
> Is this the same as bounds()?

Most of the time, yes. In certain cases, bounds appears to be updated without
re-tiling... not sure if these are error cases - will need to investigate
further. The value used here is reliably the right value (the value used to
compute tile size).

https://codereview.chromium.org/2828353003/diff/120001/cc/trees/layer_tree_ho...
File cc/trees/layer_tree_host_pixeltest_masks.cc (right):

https://codereview.chromium.org/2828353003/diff/120001/cc/trees/layer_tree_ho...
cc/trees/layer_tree_host_pixeltest_masks.cc:159: gfx::Size mask_bounds(40, 40);
On 2017/05/02 21:31:37, enne wrote:
> Feel free to take or leave this suggestion, but if you made the mask bounds
50,
> 50, would that make it stretch but line up on pixel boundaries, so all tests
> ended with the same result?

Per discussion, doesn't quite work. Added fuzzy comparison for BITMAP provider.

https://codereview.chromium.org/2828353003/diff/120001/cc/trees/layer_tree_ho...
cc/trees/layer_tree_host_pixeltest_masks.cc:164:
mask->SetLayerMaskType(Layer::LayerMaskType::SINGLE_TEXTURE_MASK);
On 2017/05/03 15:14:13, sunxd wrote:
> I made a recent change that makes pixeltest tests both SINGLE_TEXTURE_MASK and
> MULTI_TEXTURE_MASK.
> 
> Can you change this line to:
> 
> mask->SetLayerMaskType(mask_type_);

done. Thanks!

[ericrk, 2017-05-04 19:24:00]

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

[commit-bot: I haz the power, 2017-05-04 19:24:43]

Dry run: CQ is trying da patch.

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

[enne (OOO), 2017-05-04 19:40:48]

lgtm, thanks for the extra test :)

[sunxd, 2017-05-04 19:43:33]

On 2017/05/04 19:40:48, enne wrote:
> lgtm, thanks for the extra test :)

lgtm % Nit.

[sunxd, 2017-05-04 19:43:49]

https://codereview.chromium.org/2828353003/diff/140001/cc/trees/layer_tree_ho...
File cc/trees/layer_tree_host_pixeltest_masks.cc (right):

https://codereview.chromium.org/2828353003/diff/140001/cc/trees/layer_tree_ho...
cc/trees/layer_tree_host_pixeltest_masks.cc:205:
mask->SetLayerMaskType(Layer::LayerMaskType::SINGLE_TEXTURE_MASK);
Nit: Could you also change this to mask_type_? Thanks!

[ericrk, 2017-05-04 19:51:38]

https://codereview.chromium.org/2828353003/diff/140001/cc/trees/layer_tree_ho...
File cc/trees/layer_tree_host_pixeltest_masks.cc (right):

https://codereview.chromium.org/2828353003/diff/140001/cc/trees/layer_tree_ho...
cc/trees/layer_tree_host_pixeltest_masks.cc:205:
mask->SetLayerMaskType(Layer::LayerMaskType::SINGLE_TEXTURE_MASK);
On 2017/05/04 19:43:49, sunxd wrote:
> Nit: Could you also change this to mask_type_? Thanks!

I don't think this test makes sense for tiled masks - see the early-out at line
189

[ericrk, 2017-05-04 19:55:06]

https://codereview.chromium.org/2828353003/diff/140001/cc/trees/layer_tree_ho...
File cc/trees/layer_tree_host_pixeltest_masks.cc (right):

https://codereview.chromium.org/2828353003/diff/140001/cc/trees/layer_tree_ho...
cc/trees/layer_tree_host_pixeltest_masks.cc:205:
mask->SetLayerMaskType(Layer::LayerMaskType::SINGLE_TEXTURE_MASK);
On 2017/05/04 19:51:38, ericrk wrote:
> On 2017/05/04 19:43:49, sunxd wrote:
> > Nit: Could you also change this to mask_type_? Thanks!
> 
> I don't think this test makes sense for tiled masks - see the early-out at
line
> 189

I guess it can't hurt though... ok, added

[ericrk, 2017-05-04 19:55:15]

The CQ bit was checked by ericrk@chromium.org

[ericrk, 2017-05-04 19:55:15]

The patchset sent to the CQ was uploaded after l-g-t-m from sunxd@chromium.org,
enne@chromium.org
Link to the patchset: https://codereview.chromium.org/2828353003/#ps160001
(title: "nit")

[commit-bot: I haz the power, 2017-05-04 19:55:50]

CQ is trying da patch.

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

[sunxd, 2017-05-04 20:31:55]

On 2017/05/04 19:55:06, ericrk wrote:
>
https://codereview.chromium.org/2828353003/diff/140001/cc/trees/layer_tree_ho...
> File cc/trees/layer_tree_host_pixeltest_masks.cc (right):
> 
>
https://codereview.chromium.org/2828353003/diff/140001/cc/trees/layer_tree_ho...
> cc/trees/layer_tree_host_pixeltest_masks.cc:205:
> mask->SetLayerMaskType(Layer::LayerMaskType::SINGLE_TEXTURE_MASK);
> On 2017/05/04 19:51:38, ericrk wrote:
> > On 2017/05/04 19:43:49, sunxd wrote:
> > > Nit: Could you also change this to mask_type_? Thanks!
> > 
> > I don't think this test makes sense for tiled masks - see the early-out at
> line
> > 189
> 
> I guess it can't hurt though... ok, added

Thanks for the change! Just want to make the test formats unified.

[commit-bot: I haz the power, 2017-05-04 20:58:40]

CQ is committing da patch.

Bot data: {"patchset_id": 160001, "attempt_start_ts": 1493927715372660,
"parent_rev": "1e4814ea23696171d50c1368ac2c37891db5e77a", "commit_rev":
"664405a74320db4031a18694358f7d4fef327662"}

[commit-bot: I haz the power, 2017-05-04 20:58:51]

Description was changed from

==========
Determine mask UVs based on texture size

Currently, mask SINGLE_TEXTURE_MASK UVs are determined solely by the size
of the masked layer. This means that if the mask texture size does not match the
expected texture size, we will end up squishing the mask when it's applied.

This CL adds a new out param, |resource_uv_size| to GetContentsResourceId. This
param represents the UV coordinate size of the mask contents within its texture.

This factor is used to adjust the existing calculation, allowing for non-exact
mask
texture sizes.


R=sunxd@chromium.org, enne@chromium.org
BUG=713872
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel;master.tryserver.chromium.android:android_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel
==========

to

==========
Determine mask UVs based on texture size

Currently, mask SINGLE_TEXTURE_MASK UVs are determined solely by the size
of the masked layer. This means that if the mask texture size does not match the
expected texture size, we will end up squishing the mask when it's applied.

This CL adds a new out param, |resource_uv_size| to GetContentsResourceId. This
param represents the UV coordinate size of the mask contents within its texture.

This factor is used to adjust the existing calculation, allowing for non-exact
mask
texture sizes.


R=sunxd@chromium.org, enne@chromium.org
BUG=713872
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel;master.tryserver.chromium.android:android_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel

Review-Url: https://codereview.chromium.org/2828353003
Cr-Commit-Position: refs/heads/master@{#469464}
Committed:
https://chromium.googlesource.com/chromium/src/+/664405a74320db4031a18694358f...
==========

[commit-bot: I haz the power, 2017-05-04 20:58:53]

Committed patchset #8 (id:160001) as
https://chromium.googlesource.com/chromium/src/+/664405a74320db4031a18694358f...