Add TransformFeedback related APIs to command buffer: PART I.

Add TransformFeedback related APIs to command buffer: PART I.

glGenTransformFeedbacks
glDeleteTransformFeedbacks
glIsTransformFeedback
glBindTransformFeedback
glBeginTransformFeedback
glEndTransformFeedback
glPauseTransformFeedback
glResumeTransformFeedback

The missing ones are:
glTransformFeedbackVaryings
glGetTransformFeedbackVarying

Also this CL fixed a bug that's introduced in a previous CL for glIs* handling.  Basically the client ID needs to be mapped to the service ID before passing down to the driver. The test is generalized to mask this bug, so this CL fixes both the code and the test behavior.

BUG=429051
TEST=gpu_unittests
R=bajones@chromium.org
NOTRY=true

Committed: https://crrev.com/d01af9d9c924bcca8e965a89eb21e6299b2eb3e0
Cr-Commit-Position: refs/heads/master@{#308004}

[bajones, 2014-12-11 23:13:39]

lgtm

[Zhenyao Mo, 2014-12-11 23:34:15]

The CQ bit was checked by zmo@chromium.org

[commit-bot: I haz the power, 2014-12-11 23:34:44]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/795243002/40001

[Zhenyao Mo, 2014-12-11 23:53:25]

The CQ bit was unchecked by zmo@chromium.org

[Zhenyao Mo, 2014-12-11 23:54:25]

The CQ bit was checked by zmo@chromium.org

[commit-bot: I haz the power, 2014-12-11 23:56:28]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/795243002/40001

[Zhenyao Mo, 2014-12-12 00:11:17]

The CQ bit was unchecked by zmo@chromium.org

[Zhenyao Mo, 2014-12-12 00:11:42]

The CQ bit was checked by zmo@chromium.org

[commit-bot: I haz the power, 2014-12-12 00:14:05]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/795243002/40001

[commit-bot: I haz the power, 2014-12-12 00:16:09]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2014-12-12 00:17:01]

Patchset 3 (id:??) landed as
https://crrev.com/d01af9d9c924bcca8e965a89eb21e6299b2eb3e0
Cr-Commit-Position: refs/heads/master@{#308004}

[piman, 2015-01-09 07:16:46]

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

[piman, 2015-01-09 07:16:47]

https://codereview.chromium.org/795243002/diff/40001/gpu/command_buffer/servi...
File gpu/command_buffer/service/gles2_cmd_decoder_autogen.h (right):

https://codereview.chromium.org/795243002/diff/40001/gpu/command_buffer/servi...
gpu/command_buffer/service/gles2_cmd_decoder_autogen.h:122:
group_->GetTransformFeedbackServiceId(transformfeedback, &transformfeedback);
Here is the same as the other CLs. It ignores the case where the client id is
not in the map (GL_INVALID_OPERATION should be raised).

[Zhenyao Mo, 2015-01-09 17:27:51]

https://codereview.chromium.org/795243002/diff/40001/gpu/command_buffer/servi...
File gpu/command_buffer/service/gles2_cmd_decoder_autogen.h (right):

https://codereview.chromium.org/795243002/diff/40001/gpu/command_buffer/servi...
gpu/command_buffer/service/gles2_cmd_decoder_autogen.h:122:
group_->GetTransformFeedbackServiceId(transformfeedback, &transformfeedback);
On 2015/01/09 07:16:47, piman (Very slow to review) wrote:
> Here is the same as the other CLs. It ignores the case where the client id is
> not in the map (GL_INVALID_OPERATION should be raised).

My understanding is that our client IDs and service IDs are in different range,
so if the client ID is invalid and we fail to map to a valid service ID here,
the client ID will be passed down to ANGLE, where a GL error will be generated. 
That's why I didn't do any error handling here.

If my understanding is incorrect, then yes, we need to check the return of this
Get*Id function and generate an error on false.

[Zhenyao Mo, 2015-01-09 17:44:41]

https://codereview.chromium.org/795243002/diff/40001/gpu/command_buffer/servi...
File gpu/command_buffer/service/gles2_cmd_decoder_autogen.h (right):

https://codereview.chromium.org/795243002/diff/40001/gpu/command_buffer/servi...
gpu/command_buffer/service/gles2_cmd_decoder_autogen.h:122:
group_->GetTransformFeedbackServiceId(transformfeedback, &transformfeedback);
On 2015/01/09 17:27:51, Zhenyao Mo wrote:
> On 2015/01/09 07:16:47, piman (Very slow to review) wrote:
> > Here is the same as the other CLs. It ignores the case where the client id
is
> > not in the map (GL_INVALID_OPERATION should be raised).
> 
> My understanding is that our client IDs and service IDs are in different
range,
> so if the client ID is invalid and we fail to map to a valid service ID here,
> the client ID will be passed down to ANGLE, where a GL error will be
generated. 
> That's why I didn't do any error handling here.
> 
> If my understanding is incorrect, then yes, we need to check the return of
this
> Get*Id function and generate an error on false.

Actually you are right. I just double checked the client id generators, and
there is no guarantee that a client ID and a service ID may not collide. I got
the wrong impression that client IDs are in a different range.  Will add the
error check in a separate CL as it affects several other commands also.