Enable ANGLE workaround for Intel MacOS abs bug

Enable ANGLE workaround for Intel MacOS abs bug

The result of abs(i) in vertex shader is wrong in Intel Mac. The
workaround replaces abs(i) with i * sign(i).

BUG=642227
TEST=deqp/functional/gles3/shadercommonfunction.html
CQ_INCLUDE_TRYBOTS=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

Committed: https://crrev.com/1786885d8e8fad3108a6073c78c2e72dbc886dd5
Cr-Commit-Position: refs/heads/master@{#415995}

[qiankun, 2016-08-30 10:29:59]

Description was changed from

==========
Enable ANGLE workaround for Intel MacOS abs bug

The result of abs(i) in vertex shader is wrong in Intel Mac. The
workaround replaces abs(i) with i * sign(i).

BUG=642227
TEST=deqp/functional/gles3/shadercommonfunction.html
==========

to

==========
Enable ANGLE workaround for Intel MacOS abs bug

The result of abs(i) in vertex shader is wrong in Intel Mac. The
workaround replaces abs(i) with i * sign(i).

BUG=642227
TEST=deqp/functional/gles3/shadercommonfunction.html
CQ_INCLUDE_TRYBOTS=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
==========

[qiankun, 2016-08-30 11:09:35]

qiankun.miao@intel.com changed reviewers:
+ jmadill@chromium.org, kbr@chromium.org, yunchao.he@intel.com, zmo@chromium.org

[qiankun, 2016-08-30 11:09:35]

PTAL. 
This CL depends on https://chromium-review.googlesource.com/#/c/377628/.

[Corentin Wallez, 2016-08-30 13:29:34]

cwallez@chromium.org changed reviewers:
+ cwallez@chromium.org

[Corentin Wallez, 2016-08-30 13:30:08]

On 2016/08/30 at 11:09:35, qiankun.miao wrote:
> PTAL. 
> This CL depends on https://chromium-review.googlesource.com/#/c/377628/.

LGTM, once the ANGLE CL is landed and rolled in. (not OWNER though)

[Yang Gu, 2016-08-31 03:04:31]

It seems to me we have this Chromium change only because Chromium has the
workaround mechanism. This has some problems:
1. Chromium will have extra dependency with ANGLE code. In this example, without
ANGLE's change, Chromium can't be built.
2. ANGLE can be used without Chromium. For example, it can be used to run dEQP
on Windows, or be integrated in FireFox. Thus if we need these ANGLE specific
workarounds, we have to implement them in other projects.

Can we port Chromium's workaround mechanism to ANGLE?

[Zhenyao Mo, 2016-08-31 03:07:28]

On 2016/08/31 03:04:31, Yang Gu wrote:
> It seems to me we have this Chromium change only because Chromium has the
> workaround mechanism. This has some problems:
> 1. Chromium will have extra dependency with ANGLE code. In this example,
without
> ANGLE's change, Chromium can't be built.
> 2. ANGLE can be used without Chromium. For example, it can be used to run dEQP
> on Windows, or be integrated in FireFox. Thus if we need these ANGLE specific
> workarounds, we have to implement them in other projects.
> 
> Can we port Chromium's workaround mechanism to ANGLE?

Yes, that is the plan.

[Ken Russell (switch to Gerrit), 2016-08-31 18:51:16]

LGTM. Mo already answered your question about the strange split between Chromium
and ANGLE in the application of driver bug workarounds.

[qiankun, 2016-09-01 06:06:35]

shadercommonfunction.html passed in the bot. 

zmo: owner review for gpu/command_buffer/.

[Zhenyao Mo, 2016-09-01 16:58:07]

lgtm

[Zhenyao Mo, 2016-09-01 16:58:11]

The CQ bit was checked by zmo@chromium.org

[Zhenyao Mo, 2016-09-01 16:58:11]

The patchset sent to the CQ was uploaded after l-g-t-m from
cwallez@chromium.org, kbr@chromium.org
Link to the patchset: https://codereview.chromium.org/2297603002/#ps100001
(title: "update version")

[commit-bot: I haz the power, 2016-09-01 16:58:27]

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, 2016-09-01 18:04:18]

Committed patchset #6 (id:100001)

[commit-bot: I haz the power, 2016-09-01 18:06:18]

Description was changed from

==========
Enable ANGLE workaround for Intel MacOS abs bug

The result of abs(i) in vertex shader is wrong in Intel Mac. The
workaround replaces abs(i) with i * sign(i).

BUG=642227
TEST=deqp/functional/gles3/shadercommonfunction.html
CQ_INCLUDE_TRYBOTS=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

==========
Enable ANGLE workaround for Intel MacOS abs bug

The result of abs(i) in vertex shader is wrong in Intel Mac. The
workaround replaces abs(i) with i * sign(i).

BUG=642227
TEST=deqp/functional/gles3/shadercommonfunction.html
CQ_INCLUDE_TRYBOTS=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

Committed: https://crrev.com/1786885d8e8fad3108a6073c78c2e72dbc886dd5
Cr-Commit-Position: refs/heads/master@{#415995}
==========

[commit-bot: I haz the power, 2016-09-01 18:06:19]

Patchset 6 (id:??) landed as
https://crrev.com/1786885d8e8fad3108a6073c78c2e72dbc886dd5
Cr-Commit-Position: refs/heads/master@{#415995}

[Jamie Madill, 2016-09-02 14:19:25]

https://codereview.chromium.org/2297603002/diff/100001/gpu/command_buffer/ser...
File gpu/command_buffer/service/gles2_cmd_decoder.cc (right):

https://codereview.chromium.org/2297603002/diff/100001/gpu/command_buffer/ser...
gpu/command_buffer/service/gles2_cmd_decoder.cc:3730: driver_bug_workarounds |=
SH_EMULATE_ABS_INT_FUNCTION;
not sure how this compiled, but it's breaking the ANGLE trybots because the
ANGLE-side CL hasn't landed.

[Jamie Madill, 2016-09-02 14:23:18]

On 2016/09/02 14:19:25, Jamie Madill wrote:
>
https://codereview.chromium.org/2297603002/diff/100001/gpu/command_buffer/ser...
> File gpu/command_buffer/service/gles2_cmd_decoder.cc (right):
> 
>
https://codereview.chromium.org/2297603002/diff/100001/gpu/command_buffer/ser...
> gpu/command_buffer/service/gles2_cmd_decoder.cc:3730: driver_bug_workarounds
|=
> SH_EMULATE_ABS_INT_FUNCTION;
> not sure how this compiled, but it's breaking the ANGLE trybots because the
> ANGLE-side CL hasn't landed.

Sorry, spoke too soon. Was a result of the ANGLE trybots not rebasing a patch.