Check for shader version mismatch

Check for shader version mismatch

Shading language version is queried from ANGLE when a shader is
translated and stored on the Shader object. When linking shader programs,
it is checked whether the version numbers of the shaders match, and if
they don't match, linking will fail. This is specified in OpenGL ES
Shading Language 3.00 spec sections 1.5 and 3.3.

Fixes WebGL 2 conformance test:
https://www.khronos.org/registry/webgl/sdk/tests/conformance2/glsl3/shader-linking.html

BUG=485060
TEST=WebGL conformance tests, gpu_unittests

Committed: https://crrev.com/e6bdcf3bd8efc99152a68b37d313867f28a187c2
Cr-Commit-Position: refs/heads/master@{#328776}

[oetuaho-nv, 2015-05-06 14:26:13]

oetuaho@nvidia.com changed reviewers:
+ jmadill@chromium.org, kbr@chromium.org, zmo@chromium.org

[oetuaho-nv, 2015-05-06 14:26:13]

Please review!

Further changes to support invariant qualifier in ESSL3 also depend on this.

[Zhenyao Mo, 2015-05-06 17:57:42]

Mostly looks good.

https://codereview.chromium.org/1126243002/diff/1/gpu/command_buffer/service/...
File gpu/command_buffer/service/program_manager.cc (right):

https://codereview.chromium.org/1126243002/diff/1/gpu/command_buffer/service/...
gpu/command_buffer/service/program_manager.cc:1048: return true;
This is based on the assumption that kMaxAttachedShaders == 2.

We should make the algorithm more robust without this hidden assumption here.

[oetuaho-nv, 2015-05-07 08:02:41]

https://codereview.chromium.org/1126243002/diff/1/gpu/command_buffer/service/...
File gpu/command_buffer/service/program_manager.cc (right):

https://codereview.chromium.org/1126243002/diff/1/gpu/command_buffer/service/...
gpu/command_buffer/service/program_manager.cc:1048: return true;
On 2015/05/06 17:57:42, Zhenyao Mo wrote:
> This is based on the assumption that kMaxAttachedShaders == 2.
> 
> We should make the algorithm more robust without this hidden assumption here.

I don't see how this would be based on that assumption. If any of the shaders
has a different version this function will report the mismatch.

[Jamie Madill, 2015-05-07 11:27:54]

Looks good here though I'm not an owner. Couldn't hurt to add a unit test for a
mismatched version. (Didn't see one, might have missed it)

https://codereview.chromium.org/1126243002/diff/1/gpu/command_buffer/service/...
File gpu/command_buffer/service/program_manager.cc (right):

https://codereview.chromium.org/1126243002/diff/1/gpu/command_buffer/service/...
gpu/command_buffer/service/program_manager.cc:1048: return true;
On 2015/05/07 08:02:41, oetuaho-nv wrote:
> On 2015/05/06 17:57:42, Zhenyao Mo wrote:
> > This is based on the assumption that kMaxAttachedShaders == 2.
> > 
> > We should make the algorithm more robust without this hidden assumption
here.
> 
> I don't see how this would be based on that assumption. If any of the shaders
> has a different version this function will report the mismatch.

Yes, I think it's sufficient to detect any mismatch between pairs of shaders.

[Zhenyao Mo, 2015-05-07 14:33:08]

LGTM

https://codereview.chromium.org/1126243002/diff/1/gpu/command_buffer/service/...
File gpu/command_buffer/service/program_manager.cc (right):

https://codereview.chromium.org/1126243002/diff/1/gpu/command_buffer/service/...
gpu/command_buffer/service/program_manager.cc:1048: return true;
On 2015/05/07 08:02:41, oetuaho-nv wrote:
> On 2015/05/06 17:57:42, Zhenyao Mo wrote:
> > This is based on the assumption that kMaxAttachedShaders == 2.
> > 
> > We should make the algorithm more robust without this hidden assumption
here.
> 
> I don't see how this would be based on that assumption. If any of the shaders
> has a different version this function will report the mismatch.

You are right.  I misread the code.

[oetuaho-nv, 2015-05-07 15:29:22]

The CQ bit was checked by oetuaho@nvidia.com

[commit-bot: I haz the power, 2015-05-07 15:30:11]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1126243002/1

[commit-bot: I haz the power, 2015-05-07 16:45:37]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2015-05-07 16:46:30]

Patchset 1 (id:??) landed as
https://crrev.com/e6bdcf3bd8efc99152a68b37d313867f28a187c2
Cr-Commit-Position: refs/heads/master@{#328776}