Add gpu fast path for two point conical gradients

Add gpu fast path for two point conical gradients.

Also shader changes should help numerical issues on certain android gpus

BUG=179264
Committed: http://code.google.com/p/skia/source/detail?r=14316

[egdaniel, 2014-04-18 21:31:05]

Once I can get all my pre-patches in...here is the main conical gradient changes

[bsalomon, 2014-04-21 18:54:57]

https://codereview.chromium.org/238443006/diff/20001/gm/gradients_2pt_conical...
File gm/gradients_2pt_conical.cpp (right):

https://codereview.chromium.org/238443006/diff/20001/gm/gradients_2pt_conical...
gm/gradients_2pt_conical.cpp:340: class ConicalGradientsOutsideGM : public GM {
It seems like there is a lot of redundancy in these GMs. Would it be easy to
have a single class that took an array of grad makers and a GM name string?

https://codereview.chromium.org/238443006/diff/20001/src/effects/gradients/Sk...
File src/effects/gradients/SkTwoPointConicalGradient_gpu.cpp (right):

https://codereview.chromium.org/238443006/diff/20001/src/effects/gradients/Sk...
src/effects/gradients/SkTwoPointConicalGradient_gpu.cpp:20: enum ConicalType {
comment here... what are inside, outside, and edge

https://codereview.chromium.org/238443006/diff/20001/src/effects/gradients/Sk...
src/effects/gradients/SkTwoPointConicalGradient_gpu.cpp:179: // If the circles
are identical the factory will give us an empty shader.
maybe take this comment out since it isn't the primary reason for the loop any
longer.

https://codereview.chromium.org/238443006/diff/20001/src/effects/gradients/Sk...
src/effects/gradients/SkTwoPointConicalGradient_gpu.cpp:228: if (kVec3f_GrSLType
== coords[0].type()) {
There's a helper on GrGLShaderBuilder called ensureFSCoords2D. I'll leave up to
you to decide if it's worth using here and for the other two effects below.

https://codereview.chromium.org/238443006/diff/20001/src/effects/gradients/Sk...
src/effects/gradients/SkTwoPointConicalGradient_gpu.cpp:319: // If the focal
point is touch the edge of the circle it will
touch->touching

https://codereview.chromium.org/238443006/diff/20001/src/effects/gradients/Sk...
src/effects/gradients/SkTwoPointConicalGradient_gpu.cpp:448: // If the circles
are identical the factory will give us an empty shader.
same

https://codereview.chromium.org/238443006/diff/20001/src/effects/gradients/Sk...
src/effects/gradients/SkTwoPointConicalGradient_gpu.cpp:721:
builder->fsCodeAppendf("\t%s = vec4(0.0,0.0,0.0,0.0);\n", outputColor);
I think this initialization only needs to happen in the non-nested circle cases.

https://codereview.chromium.org/238443006/diff/20001/src/effects/gradients/Sk...
src/effects/gradients/SkTwoPointConicalGradient_gpu.cpp:1328:
set_matrix_edge_conical(shader, &matrix);
might be the browser extension but this looks double indented.

[egdaniel, 2014-04-22 14:46:51]

Compacted the GM file

Adjusted comments in the TestCreate loops to clarify what each one was for

Also had to adjust tolerance for falling into the Edge case since I was still
seeing some slight instability when just outside our tolerance range

[bsalomon, 2014-04-22 15:19:58]

two small comments, otherwise lgtm

https://codereview.chromium.org/238443006/diff/50001/gm/gradients_2pt_conical...
File gm/gradients_2pt_conical.cpp (right):

https://codereview.chromium.org/238443006/diff/50001/gm/gradients_2pt_conical...
gm/gradients_2pt_conical.cpp:293: { gGradMakersOutside,     4,    "outside"  },
Can 4, 6, and 7 be replaced by SK_ARRAY_COUNT?

https://codereview.chromium.org/238443006/diff/50001/src/effects/gradients/Sk...
File src/effects/gradients/SkTwoPointConicalGradient_gpu.cpp (right):

https://codereview.chromium.org/238443006/diff/50001/src/effects/gradients/Sk...
src/effects/gradients/SkTwoPointConicalGradient_gpu.cpp:1222:
builder->fsCodeAppendf("\tfloat temp = -1.0 * min(deter, %s - %s.w);\n",
tName.c_str(), params.c_str());
did you meant to leave this here?

[egdaniel, 2014-04-22 19:24:15]

The CQ bit was checked by egdaniel@google.com

[commit-bot: I haz the power, 2014-04-22 19:24:36]

CQ is trying da patch. Follow status at
https://skia-tree-status.appspot.com/cq/egdaniel@google.com/238443006/110001

[commit-bot: I haz the power, 2014-04-22 20:56:41]

Change committed as 14316