Remove NameMap from shader translator and shader/program managers

Remove NameMap from shader translator and shader/program managers

NameMap, which after the previous patch is no longer used by the shader/program managers, is removed, and the tests updated.

Tests are also added for GetOriginalNameFromHashedName, though these don't actually hit the original error case in the bug report. That's a difficult integration issue and is tested instead by the WebGL conformance tests.

Usages of GetOriginalNameFromHashedName are replaced with more specific methods calls in Program::{GetTransformFeedbackVaryings,GetUniformBlocks}. This avoids querying the unnecessary extra four info maps.

BUG=716018
TEST=https://github.com/KhronosGroup/WebGL/pull/2388
TEST=https://github.com/KhronosGroup/WebGL/pull/2389
TEST=ShaderManagerTest.DoCompile
CQ_INCLUDE_TRYBOTS=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/2852923004
Cr-Commit-Position: refs/heads/master@{#469398}
Committed: https://chromium.googlesource.com/chromium/src/+/78c79d2aff281ddebbe220a1f41214d3c071f16b

[Kai Ninomiya, 2017-05-01 20:56:37]

Description was changed from

==========
Fix loading of uniform block names from shader cache

Previously, uniform block names were unmapped using the name_map, but
name_map is not restored upon load from the shader cache so it doesn't
work correctly.

BUG=716018
==========

to

==========
Fix loading of uniform block names from shader cache

Previously, uniform block names were unmapped using the name_map, but
name_map is not restored upon load from the shader cache so it doesn't
work correctly.

BUG=716018
CQ_INCLUDE_TRYBOTS=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
==========

[Kai Ninomiya, 2017-05-01 20:58:06]

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

[commit-bot: I haz the power, 2017-05-01 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, 2017-05-01 21:08:33]

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

[commit-bot: I haz the power, 2017-05-01 21:08:34]

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...)
  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_...)

[Kai Ninomiya, 2017-05-01 21:39:47]

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

[commit-bot: I haz the power, 2017-05-01 21:41:11]

Dry run: CQ is trying da patch.

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

[Kai Ninomiya, 2017-05-01 22:00:46]

Description was changed from

==========
Fix loading of uniform block names from shader cache

Previously, uniform block names were unmapped using the name_map, but
name_map is not restored upon load from the shader cache so it doesn't
work correctly.

BUG=716018
CQ_INCLUDE_TRYBOTS=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

==========
Fix loading of uniform block names from shader cache

Previously, uniform block names were unmapped using the name_map, but
name_map is not restored upon load from the shader cache so it doesn't
work correctly.

BUG=716018
TEST=https://github.com/KhronosGroup/WebGL/pull/2388
CQ_INCLUDE_TRYBOTS=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
==========

[Kai Ninomiya, 2017-05-01 22:43:33]

kainino@chromium.org changed reviewers:
+ kbr@chromium.org

[Kai Ninomiya, 2017-05-01 22:43:34]

kbr: PTAL, although I think this may need some more work to remove the
fragile-looking path for transform feedback varyings.

[Ken Russell (switch to Gerrit), 2017-05-01 23:02:30]

kbr@chromium.org changed reviewers:
+ jbauman@chromium.org

[Ken Russell (switch to Gerrit), 2017-05-01 23:02:30]

This looks good overall but could you see whether a unit test in
src/gpu/command_buffer/service/gles2_cmd_decoder_unittest_programs.cc is
feasible?

jbauman@'s an owner so will need to review this too.

https://codereview.chromium.org/2852923004/diff/20001/gpu/command_buffer/serv...
File gpu/command_buffer/service/program_manager.cc (right):

https://codereview.chromium.org/2852923004/diff/20001/gpu/command_buffer/serv...
gpu/command_buffer/service/program_manager.cc:1549: // used instead.
Please refer to crbug.com/716018 at the end of this comment. Thanks.

[commit-bot: I haz the power, 2017-05-01 23:17:11]

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

[commit-bot: I haz the power, 2017-05-01 23:17:12]

Dry run: This issue passed the CQ dry run.

[Kai Ninomiya, 2017-05-02 00:06:08]

On 2017/05/01 23:02:30, Ken Russell wrote:
> This looks good overall but could you see whether a unit test in
> src/gpu/command_buffer/service/gles2_cmd_decoder_unittest_programs.cc is
> feasible?

I don't know whether this is feasible as the bug itself is dependent on the
shader cache loading. Instead of a command unittest, maybe we would need a test
for the name_map load from cache, or something.

[Kai Ninomiya, 2017-05-02 00:06:13]

https://codereview.chromium.org/2852923004/diff/20001/gpu/command_buffer/serv...
File gpu/command_buffer/service/program_manager.cc (right):

https://codereview.chromium.org/2852923004/diff/20001/gpu/command_buffer/serv...
gpu/command_buffer/service/program_manager.cc:1549: // used instead.
On 2017/05/01 23:02:30, Ken Russell wrote:
> Please refer to crbug.com/716018 at the end of this comment. Thanks.

Acknowledged.

[Kai Ninomiya, 2017-05-02 00:36:14]

jbauman: actually, you can defer on the review. I'm investigating further into
the issues here and I think there's also a deeper problem (race between shader
cache load and compile?)

[Kai Ninomiya, 2017-05-02 01:35:09]

On 2017/05/02 00:36:14, Kai Ninomiya wrote:
> jbauman: actually, you can defer on the review. I'm investigating further into
> the issues here and I think there's also a deeper problem (race between shader
> cache load and compile?)

Okay, not that bad. Notes for self:

Looks like the path to Shader::GetOriginalNameFromHashedName via
getUniformBlockIndex simply doesn't hit shader->DoCompile(), so the shader is in
a needs-compile state [1]. It looked like translator->Translate() was being
called more than it should be, but I guess it's supposed to be called in the
cache-hit case [2].

[1] getShaderParameter calls DoCompile which is why it hides the problem. But
getAttribLocation, getUniformLocation just don't _need_ DoCompile() to have been
called, AFAICT, so they work (maybe accidentally?)

[2] I need to double check that this is what's supposed to happen, and that
we're not extraneously translating.

[Kai Ninomiya, 2017-05-02 18:12:09]

Description was changed from

==========
Fix loading of uniform block names from shader cache

Previously, uniform block names were unmapped using the name_map, but
name_map is not restored upon load from the shader cache so it doesn't
work correctly.

BUG=716018
TEST=https://github.com/KhronosGroup/WebGL/pull/2388
CQ_INCLUDE_TRYBOTS=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

==========
Fix loading of uniform block/transform feedback varying names from shader cache

Previously, uniform block names were unmapped using the name_map, but
name_map is not restored upon load from the shader cache so it doesn't
work correctly.

BUG=716018
TEST=https://github.com/KhronosGroup/WebGL/pull/2388
TEST=https://github.com/KhronosGroup/WebGL/pull/2389
CQ_INCLUDE_TRYBOTS=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
==========

[Kai Ninomiya, 2017-05-02 18:12:28]

Description was changed from

==========
Fix loading of uniform block/transform feedback varying names from shader cache

Previously, uniform block names were unmapped using the name_map, but
name_map is not restored upon load from the shader cache so it doesn't
work correctly.

BUG=716018
TEST=https://github.com/KhronosGroup/WebGL/pull/2388
TEST=https://github.com/KhronosGroup/WebGL/pull/2389
CQ_INCLUDE_TRYBOTS=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

==========
Fix loading of uniform block/transform feedback varying names from shader cache

Previously, uniform block names were unmapped using the name_map, but
name_map is not restored upon load from the shader cache so it doesn't
work correctly.

// TODO: fix for transform feedback varyings
COMMIT=false

BUG=716018
TEST=https://github.com/KhronosGroup/WebGL/pull/2388
TEST=https://github.com/KhronosGroup/WebGL/pull/2389
CQ_INCLUDE_TRYBOTS=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
==========

[Kai Ninomiya, 2017-05-02 23:32:05]

Description was changed from

==========
Fix loading of uniform block/transform feedback varying names from shader cache

Previously, uniform block names were unmapped using the name_map, but
name_map is not restored upon load from the shader cache so it doesn't
work correctly.

// TODO: fix for transform feedback varyings
COMMIT=false

BUG=716018
TEST=https://github.com/KhronosGroup/WebGL/pull/2388
TEST=https://github.com/KhronosGroup/WebGL/pull/2389
CQ_INCLUDE_TRYBOTS=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

==========
Fix loading of uniform block/transform feedback varying names from shader cache

The main fix here is to correct the behavior of GetOriginalNameFromHashedName.
Previously, uniform block names were unmapped using "name_map", but
name_map was not restored upon load. Now, instead of name_map, the five
info maps (attrib, uniform, varying, interface block, output variable)
are queried serially in GetOriginalNameFromHashedName. Tests are added for
GetOriginalNameFromHashedName, though these don't actually hit the
original error case. That's a difficult integration issue and is tested instead
by the WebGL conformance tests linked below.

In addition to that fix, usages of GetOriginalNameFromHashedName are replaced
with more specific methods calls in
Program::{GetTransformFeedbackVaryings,GetUniformBlocks}. This avoids querying
the unnecessary extra four info maps.

BUG=716018
TEST=https://github.com/KhronosGroup/WebGL/pull/2388
TEST=https://github.com/KhronosGroup/WebGL/pull/2389
TEST=ShaderManagerTest.DoCompile
CQ_INCLUDE_TRYBOTS=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
==========

[Kai Ninomiya, 2017-05-02 23:33:26]

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

[commit-bot: I haz the power, 2017-05-02 23:34:00]

Dry run: CQ is trying da patch.

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

[Kai Ninomiya, 2017-05-03 00:08:06]

Description was changed from

==========
Fix loading of uniform block/transform feedback varying names from shader cache

The main fix here is to correct the behavior of GetOriginalNameFromHashedName.
Previously, uniform block names were unmapped using "name_map", but
name_map was not restored upon load. Now, instead of name_map, the five
info maps (attrib, uniform, varying, interface block, output variable)
are queried serially in GetOriginalNameFromHashedName. Tests are added for
GetOriginalNameFromHashedName, though these don't actually hit the
original error case. That's a difficult integration issue and is tested instead
by the WebGL conformance tests linked below.

In addition to that fix, usages of GetOriginalNameFromHashedName are replaced
with more specific methods calls in
Program::{GetTransformFeedbackVaryings,GetUniformBlocks}. This avoids querying
the unnecessary extra four info maps.

BUG=716018
TEST=https://github.com/KhronosGroup/WebGL/pull/2388
TEST=https://github.com/KhronosGroup/WebGL/pull/2389
TEST=ShaderManagerTest.DoCompile
CQ_INCLUDE_TRYBOTS=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

==========
Remove NameMap from shader translator and shader/program managers

NameMap, which after the previous patch is no longer used by the shader/program
managers, is removed, and the tests updated.

Tests are also added for GetOriginalNameFromHashedName, though these don't
actually hit the original error case in the bug report. That's a difficult
integration issue and is tested instead by the WebGL conformance tests.

Usages of GetOriginalNameFromHashedName are replaced with more specific methods
calls in Program::{GetTransformFeedbackVaryings,GetUniformBlocks}. This avoids
querying the unnecessary extra four info maps.

BUG=716018
TEST=https://github.com/KhronosGroup/WebGL/pull/2388
TEST=https://github.com/KhronosGroup/WebGL/pull/2389
TEST=ShaderManagerTest.DoCompile
CQ_INCLUDE_TRYBOTS=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
==========

[Kai Ninomiya, 2017-05-03 00:08:15]

The CQ bit was unchecked by kainino@chromium.org

[Kai Ninomiya, 2017-05-03 00:12:37]

kbr, jbauman: PTAL, this is ready for review again. I've split this CL in half;
first half is the fix, and this half is really just a refactor.

[Kai Ninomiya, 2017-05-03 18:13:24]

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

[commit-bot: I haz the power, 2017-05-03 18:14:03]

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, 2017-05-03 19:12:21]

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

[commit-bot: I haz the power, 2017-05-03 19:12:22]

Dry run: This issue passed the CQ dry run.

[Ken Russell (switch to Gerrit), 2017-05-03 22:52:49]

CC'ing zmo@ as he may have more comments having worked with the name map before.

Overall this looks good. It is good to get rid of this redundant data structure
which isn't being correctly maintained when reloading shader binaries from the
cache. A few questions.

https://codereview.chromium.org/2852923004/diff/80001/gpu/command_buffer/serv...
File gpu/command_buffer/service/program_manager.cc (right):

https://codereview.chromium.org/2852923004/diff/80001/gpu/command_buffer/serv...
gpu/command_buffer/service/program_manager.cc:2268: names[ii] =
interface_block->name + array_index_str;
This change is well tested with the existing conformance tests, correct?

https://codereview.chromium.org/2852923004/diff/80001/gpu/command_buffer/serv...
gpu/command_buffer/service/program_manager.cc:2401: names[ii] = varying->name;
Same question here.

https://codereview.chromium.org/2852923004/diff/80001/gpu/command_buffer/serv...
File gpu/command_buffer/service/shader_manager_unittest.cc (right):

https://codereview.chromium.org/2852923004/diff/80001/gpu/command_buffer/serv...
gpu/command_buffer/service/shader_manager_unittest.cc:248: // TODO(kainino):
Check interface block infos got copied.
What's the plan to address this TODO?

[Kai Ninomiya, 2017-05-03 23:02:34]

https://codereview.chromium.org/2852923004/diff/80001/gpu/command_buffer/serv...
File gpu/command_buffer/service/program_manager.cc (right):

https://codereview.chromium.org/2852923004/diff/80001/gpu/command_buffer/serv...
gpu/command_buffer/service/program_manager.cc:2268: names[ii] =
interface_block->name + array_index_str;
On 2017/05/03 22:52:48, Ken Russell wrote:
> This change is well tested with the existing conformance tests, correct?

That the interface block name returned will be the correct one? Yes.

https://codereview.chromium.org/2852923004/diff/80001/gpu/command_buffer/serv...
gpu/command_buffer/service/program_manager.cc:2401: names[ii] = varying->name;
On 2017/05/03 22:52:48, Ken Russell wrote:
> Same question here.

Ditto.

https://codereview.chromium.org/2852923004/diff/80001/gpu/command_buffer/serv...
File gpu/command_buffer/service/shader_manager_unittest.cc (right):

https://codereview.chromium.org/2852923004/diff/80001/gpu/command_buffer/serv...
gpu/command_buffer/service/shader_manager_unittest.cc:248: // TODO(kainino):
Check interface block infos got copied.
On 2017/05/03 22:52:48, Ken Russell wrote:
> What's the plan to address this TODO?

I didn't have much of a plan except "not now because it would add too much
complexity to this CL" - there's no helper code to make this work. I don't mind
queueing it up after this CL, should hopefully not take too long.

[Ken Russell (switch to Gerrit), 2017-05-03 23:10:02]

Thanks for the updates. LGTM but I'm not a directory OWNER.

https://codereview.chromium.org/2852923004/diff/80001/gpu/command_buffer/serv...
File gpu/command_buffer/service/shader_manager_unittest.cc (right):

https://codereview.chromium.org/2852923004/diff/80001/gpu/command_buffer/serv...
gpu/command_buffer/service/shader_manager_unittest.cc:248: // TODO(kainino):
Check interface block infos got copied.
On 2017/05/03 23:02:34, Kai Ninomiya wrote:
> On 2017/05/03 22:52:48, Ken Russell wrote:
> > What's the plan to address this TODO?
> 
> I didn't have much of a plan except "not now because it would add too much
> complexity to this CL" - there's no helper code to make this work. I don't
mind
> queueing it up after this CL, should hopefully not take too long.

Could you please do this in a follow-on CL? It will help ensure that this
doesn't regress. Thanks much.

[Zhenyao Mo, 2017-05-03 23:24:11]

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

[Zhenyao Mo, 2017-05-03 23:24:12]

Thanks a lot for cleaning this up.

Now we no longer use the general name map, we should remove that map from ANGLE
side also (including its public interface).  Please do that as a followup of
this CL.

https://codereview.chromium.org/2852923004/diff/80001/gpu/command_buffer/serv...
File gpu/command_buffer/service/program_manager.cc (right):

https://codereview.chromium.org/2852923004/diff/80001/gpu/command_buffer/serv...
gpu/command_buffer/service/program_manager.cc:2398: // TODO(zmo): optimize the
name mapping lookup.
You can probably remove this comment because I don't think we will do any
further optimization on this.

[Zhenyao Mo, 2017-05-03 23:24:21]

On 2017/05/03 23:24:12, Zhenyao Mo wrote:
> Thanks a lot for cleaning this up.
> 
> Now we no longer use the general name map, we should remove that map from
ANGLE
> side also (including its public interface).  Please do that as a followup of
> this CL.
> 
>
https://codereview.chromium.org/2852923004/diff/80001/gpu/command_buffer/serv...
> File gpu/command_buffer/service/program_manager.cc (right):
> 
>
https://codereview.chromium.org/2852923004/diff/80001/gpu/command_buffer/serv...
> gpu/command_buffer/service/program_manager.cc:2398: // TODO(zmo): optimize the
> name mapping lookup.
> You can probably remove this comment because I don't think we will do any
> further optimization on this.

LGTM

[Kai Ninomiya, 2017-05-04 00:20:21]

The CQ bit was checked by kainino@chromium.org

[Kai Ninomiya, 2017-05-04 00:20:22]

The patchset sent to the CQ was uploaded after l-g-t-m from kbr@chromium.org,
zmo@chromium.org
Link to the patchset: https://codereview.chromium.org/2852923004/#ps100001
(title: "remove some TODO(zmo)s")

[commit-bot: I haz the power, 2017-05-04 00:21:07]

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, 2017-05-04 00:27:57]

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

[commit-bot: I haz the power, 2017-05-04 00:27:58]

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_...)
  win_optional_gpu_tests_rel on master.tryserver.chromium.win (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.win/builders/win_optional_gpu_...)

[Kai Ninomiya, 2017-05-04 17:36:43]

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

[commit-bot: I haz the power, 2017-05-04 17:37:19]

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, 2017-05-04 18:34:26]

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

[commit-bot: I haz the power, 2017-05-04 18:34:27]

Dry run: This issue passed the CQ dry run.

[Kai Ninomiya, 2017-05-04 18:38:33]

The CQ bit was checked by kainino@chromium.org

[commit-bot: I haz the power, 2017-05-04 18:39:11]

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, 2017-05-04 18:45:35]

CQ is committing da patch.

Bot data: {"patchset_id": 100001, "attempt_start_ts": 1493923113949620,
"parent_rev": "eb8892ff597ac955e4eebe688a57ca0b81b02bd4", "commit_rev":
"78c79d2aff281ddebbe220a1f41214d3c071f16b"}

[commit-bot: I haz the power, 2017-05-04 18:45:50]

Description was changed from

==========
Remove NameMap from shader translator and shader/program managers

NameMap, which after the previous patch is no longer used by the shader/program
managers, is removed, and the tests updated.

Tests are also added for GetOriginalNameFromHashedName, though these don't
actually hit the original error case in the bug report. That's a difficult
integration issue and is tested instead by the WebGL conformance tests.

Usages of GetOriginalNameFromHashedName are replaced with more specific methods
calls in Program::{GetTransformFeedbackVaryings,GetUniformBlocks}. This avoids
querying the unnecessary extra four info maps.

BUG=716018
TEST=https://github.com/KhronosGroup/WebGL/pull/2388
TEST=https://github.com/KhronosGroup/WebGL/pull/2389
TEST=ShaderManagerTest.DoCompile
CQ_INCLUDE_TRYBOTS=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

==========
Remove NameMap from shader translator and shader/program managers

NameMap, which after the previous patch is no longer used by the shader/program
managers, is removed, and the tests updated.

Tests are also added for GetOriginalNameFromHashedName, though these don't
actually hit the original error case in the bug report. That's a difficult
integration issue and is tested instead by the WebGL conformance tests.

Usages of GetOriginalNameFromHashedName are replaced with more specific methods
calls in Program::{GetTransformFeedbackVaryings,GetUniformBlocks}. This avoids
querying the unnecessary extra four info maps.

BUG=716018
TEST=https://github.com/KhronosGroup/WebGL/pull/2388
TEST=https://github.com/KhronosGroup/WebGL/pull/2389
TEST=ShaderManagerTest.DoCompile
CQ_INCLUDE_TRYBOTS=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/2852923004
Cr-Commit-Position: refs/heads/master@{#469398}
Committed:
https://chromium.googlesource.com/chromium/src/+/78c79d2aff281ddebbe220a1f412...
==========

[commit-bot: I haz the power, 2017-05-04 18:45:51]

Committed patchset #6 (id:100001) as
https://chromium.googlesource.com/chromium/src/+/78c79d2aff281ddebbe220a1f412...