SkPictureShader: scale down if width or height is larger than maxTextureSize

SkPictureShader: scale down if width or height is larger than maxTextureSize

Grdient renders black because of failure onCreateTexture
if one of width or height is larger than maxTextureSize
with --force-gpu-rasterization.

BUG=chromium:473166
Committed: https://skia.googlesource.com/skia/+/b9ed8845783ec3c59b2b185d26801f251adbf68d

[gen.kim, 2015-04-27 02:27:21]

gen.kim@samsung.com changed reviewers:
+ mtklein@google.com, reed@google.com

[sohanjg, 2015-04-27 10:07:52]

sohan.jyoti@samsung.com changed reviewers:
+ bsalomon@chromium.org

[mtklein, 2015-04-27 13:27:52]

mtklein@google.com changed reviewers:
+ fmalita@chromium.org
- mtklein@google.com

[gen.kim, 2015-04-28 02:07:03]

The CQ bit was checked by gen.kim@samsung.com to run a CQ dry run

[commit-bot: I haz the power, 2015-04-28 02:07:18]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1101513004/20001

[commit-bot: I haz the power, 2015-04-28 02:07:19]

Note for Reviewers:
The CQ is waiting for an approval.
If you believe that the CL is not ready yet, or if you would like to L-G-T-M
with comments then please uncheck the CQ checkbox.

Waiting for LGTM from valid reviewer(s) till 2015-04-28 08:07 UTC

[gen.kim, 2015-04-28 03:54:59]

The CQ bit was unchecked by gen.kim@samsung.com

[bsalomon, 2015-04-30 12:53:15]

bsalomon@google.com changed reviewers:
+ bsalomon@google.com

[bsalomon, 2015-04-30 12:53:15]

This seems ok to me, but Florin should look. It'd be great to address the fact
that the picture shader doesn't use the GPU to generate the source image but
instead always uses the sw rasterizer.

https://codereview.chromium.org/1101513004/diff/20001/src/core/SkPictureShade...
File src/core/SkPictureShader.cpp (right):

https://codereview.chromium.org/1101513004/diff/20001/src/core/SkPictureShade...
src/core/SkPictureShader.cpp:315: if (context)
small nit, we always use braces, even for one line ifs.

[f(malita), 2015-04-30 13:08:10]

LGTM

On 2015/04/30 12:53:15, bsalomon wrote:
> This seems ok to me, but Florin should look. It'd be great to address the fact
> that the picture shader doesn't use the GPU to generate the source image but
> instead always uses the sw rasterizer.

Agreed, especially as we're starting to use SkPictureShader more & more in
Blink. I recall long ago when we talked about it there was the issue of
variable-length texture cache key support. Can we make use of arbitrary keys
now?


> 
>
https://codereview.chromium.org/1101513004/diff/20001/src/core/SkPictureShade...
> File src/core/SkPictureShader.cpp (right):
> 
>
https://codereview.chromium.org/1101513004/diff/20001/src/core/SkPictureShade...
> src/core/SkPictureShader.cpp:315: if (context)
> small nit, we always use braces, even for one line ifs.

[bsalomon, 2015-04-30 13:38:08]

On 2015/04/30 13:08:10, f(malita) wrote:
> LGTM
> 
> On 2015/04/30 12:53:15, bsalomon wrote:
> > This seems ok to me, but Florin should look. It'd be great to address the
fact
> > that the picture shader doesn't use the GPU to generate the source image but
> > instead always uses the sw rasterizer.
> 
> Agreed, especially as we're starting to use SkPictureShader more & more in
> Blink. I recall long ago when we talked about it there was the issue of
> variable-length texture cache key support. Can we make use of arbitrary keys
> now?
> 

Hm, not sure why that'd be an issue here but keys can be variable length now. I
was thinking we'd create a SkSurface that matched the dst's type (cpu/gpu), draw
the picture, and snap an image.  Then image would be installed in the returned
shader. I'm not sure if we have an image shader yet, though.

[gen.kim, 2015-05-04 04:02:45]

https://codereview.chromium.org/1101513004/diff/20001/src/core/SkPictureShade...
File src/core/SkPictureShader.cpp (right):

https://codereview.chromium.org/1101513004/diff/20001/src/core/SkPictureShade...
src/core/SkPictureShader.cpp:315: if (context)
On 2015/04/30 12:53:15, bsalomon wrote:
> small nit, we always use braces, even for one line ifs.

Thanks, fixed

[gen.kim, 2015-05-04 05:32:52]

The CQ bit was checked by gen.kim@samsung.com

[gen.kim, 2015-05-04 05:32:52]

The patchset sent to the CQ was uploaded after l-g-t-m from fmalita@chromium.org
Link to the patchset: https://codereview.chromium.org/1101513004/#ps40001
(title: " ")

[commit-bot: I haz the power, 2015-05-04 05:32:56]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1101513004/40001

[commit-bot: I haz the power, 2015-05-04 05:36:37]

Committed patchset #3 (id:40001) as
https://skia.googlesource.com/skia/+/b9ed8845783ec3c59b2b185d26801f251adbf68d