Make GrGLProgram be available to GrGLProgramDataManager

Make GrGLProgram be available to GrGLProgramDataManager

GL APIs such as glProgramPathFragmentInputGenNV need the explicit program id in
order to set the program resources. For GrGLProgramDataManager to call the APIs,
the program id needs to be available to the GrGLProgramDataManager.

One GrGLProgramDataManager is used to explicitly manage the resources of one
GrGLProgram. Refactor the classes to reflect this: make GrGLProgram create and
own GrGLProgramDataManager. This way the program id can be made available through
GrGLProgram backpointer.

In order to do this, remove the population of the program link results
information for uniforms in GrGLProgramDataManager during shader generation.

Instead, accumulate the info to the uniform list of GrGLShaderBuilder. After the
generation, compilation and link is successful, this info can be used to
construct both GrGLProgram and GrGLProgramDataManager.

Changes names of functions to create UniformHandles and convert them to list
indices. This tries to highlight that they're different lists, even though at
the moment they match index-wise. E.g the contract is that handles can be
created based on the shader builder state, and the object can later be set with
the handle and the manager that was created with the builder.

This is needed to support NV_path_rendering for GLES.

Committed: https://skia.googlesource.com/skia/+/dddc18a6bba67ad43a65f4c244b95f76cefef617

[Kimmo Kinnunen, 2014-08-01 11:24:16]

How does this sound? In order to set the fragment shader variables without
vertex shader, we need a ref to the program id. This tries to provide a way for
the ProgramDataManager (previously UniformManager) to obtain that ref via
ProgramDataManager constructor.

[bsalomon, 2014-08-01 13:46:45]

lgtm w/ a question

https://codereview.chromium.org/426553011/diff/20001/src/gpu/gl/GrGLProgramDa...
File src/gpu/gl/GrGLProgramDataManager.h (right):

https://codereview.chromium.org/426553011/diff/20001/src/gpu/gl/GrGLProgramDa...
src/gpu/gl/GrGLProgramDataManager.h:49: int toProgramDataIndex() const {
SkASSERT(isValid()); return ~fValue; }
I'm not sure we still need this munging at all. IIRC, originally this was just a
typedef'ed int we transformed just so we'd catch any places where we we're not
treating this as an opaque value. Now that the value is private, it may not make
sense.

[Kimmo Kinnunen, 2014-08-04 05:43:27]

https://codereview.chromium.org/426553011/diff/20001/src/gpu/gl/GrGLProgramDa...
File src/gpu/gl/GrGLProgramDataManager.h (right):

https://codereview.chromium.org/426553011/diff/20001/src/gpu/gl/GrGLProgramDa...
src/gpu/gl/GrGLProgramDataManager.h:49: int toProgramDataIndex() const {
SkASSERT(isValid()); return ~fValue; }
On 2014/08/01 13:46:45, bsalomon wrote:
> I'm not sure we still need this munging at all. IIRC, originally this was just
a
> typedef'ed int we transformed just so we'd catch any places where we we're not
> treating this as an opaque value. Now that the value is private, it may not
make
> sense.

Done.

[Kimmo Kinnunen, 2014-08-04 05:47:18]

The CQ bit was checked by kkinnunen@nvidia.com

[commit-bot: I haz the power, 2014-08-04 05:47:44]

CQ is trying da patch. Follow status at
 https://skia-tree-status.appspot.com/cq/kkinnunen@nvidia.com/426553011/40001

[commit-bot: I haz the power, 2014-08-04 06:00:24]

FYI, CQ is re-trying this CL (attempt #1).
The failing builders are:
  Build-Win-VS2013-x86-Debug-Trybot on tryserver.skia
(http://108.170.220.76:10117/builders/Build-Win-VS2013-x86-Debug-Trybot/builds...)

[Kimmo Kinnunen, 2014-08-04 06:08:37]

The CQ bit was checked by kkinnunen@nvidia.com

[commit-bot: I haz the power, 2014-08-04 06:08:54]

CQ is trying da patch. Follow status at
 https://skia-tree-status.appspot.com/cq/kkinnunen@nvidia.com/426553011/60001

[commit-bot: I haz the power, 2014-08-04 06:19:50]

Change committed as dddc18a6bba67ad43a65f4c244b95f76cefef617