Enable glReadPixels from <video> to 2d context on Android.

Enable glReadPixels from <video> to 2d context on Android.

The ordering inside Skia's kARGB_8888_Config is platform-specific; on Android it
is RGBA but on non-Android it's BGRA.
(controlled by SkUserConfig and SK_[ARGB]32_SHIFT)

BUG=138930
Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=230229

[Ami GONE FROM CHROMIUM, 2013-10-21 23:46:12]

scherkus: PTAL
wjia: this fixes the borkedness you pointed out to me.

[scherkus (not reviewing), 2013-10-22 18:29:10]

lgtm w/ observation

https://codereview.chromium.org/31583004/diff/100001/content/renderer/media/r...
File content/renderer/media/renderer_gpu_video_accelerator_factories.cc (right):

https://codereview.chromium.org/31583004/diff/100001/content/renderer/media/r...
content/renderer/media/renderer_gpu_video_accelerator_factories.cc:260: #if
SK_B32_SHIFT == 0 && SK_G32_SHIFT == 8 && SK_R32_SHIFT == 16 && \
it seems like the cool kids only check for SK_B32_SHIFT being either 0 or 16
e.g., 
https://code.google.com/p/chromium/codesearch#chromium/src/cc/resources/platf...
https://code.google.com/p/chromium/codesearch#chromium/src/third_party/WebKit...

I guess you're being overly paranoid but there really isn't a reason to check
all four colour components, unless you value saving whitespace

[scherkus (not reviewing), 2013-10-22 18:29:46]

https://codereview.chromium.org/31583004/diff/100001/content/renderer/media/r...
File content/renderer/media/renderer_gpu_video_accelerator_factories.cc (right):

https://codereview.chromium.org/31583004/diff/100001/content/renderer/media/r...
content/renderer/media/renderer_gpu_video_accelerator_factories.cc:260: #if
SK_B32_SHIFT == 0 && SK_G32_SHIFT == 8 && SK_R32_SHIFT == 16 && \
On 2013/10/22 18:29:10, scherkus wrote:
> it seems like the cool kids only check for SK_B32_SHIFT being either 0 or 16
> e.g., 
>
https://code.google.com/p/chromium/codesearch#chromium/src/cc/resources/platf...
>
https://code.google.com/p/chromium/codesearch#chromium/src/third_party/WebKit...
> 
> I guess you're being overly paranoid but there really isn't a reason to check
> all four colour components, unless you value saving whitespace

... there really isn't a reason *NOT* to check ...

[Ami GONE FROM CHROMIUM, 2013-10-22 18:32:59]

Thanks for the review

https://codereview.chromium.org/31583004/diff/100001/content/renderer/media/r...
File content/renderer/media/renderer_gpu_video_accelerator_factories.cc (right):

https://codereview.chromium.org/31583004/diff/100001/content/renderer/media/r...
content/renderer/media/renderer_gpu_video_accelerator_factories.cc:260: #if
SK_B32_SHIFT == 0 && SK_G32_SHIFT == 8 && SK_R32_SHIFT == 16 && \
On 2013/10/22 18:29:10, scherkus wrote:
> it seems like the cool kids only check for SK_B32_SHIFT being either 0 or 16
> e.g., 
>
https://code.google.com/p/chromium/codesearch#chromium/src/cc/resources/platf...
>
https://code.google.com/p/chromium/codesearch#chromium/src/third_party/WebKit...
> 
> I guess you're being overly paranoid but there really isn't a reason to check
> all four colour components, unless you value saving whitespace

Yeah, the first version of this code was actually just
#if ANDROID
  use RGBA
#else
  use BGRA_EXT
#endif
but then realized that skia doesn't actually guarantee this, so decided to stay
safe.  When skia doco's that the above logic is correct then I'm fine reverting
to it, but until then planning to leave as-is.

(Skia's naming of ARGB_8888 has confused at least 5 experienced developers re:
this code, including me _twice_ over the last year or two, so opting for extra
explicitness)

[commit-bot: I haz the power, 2013-10-22 18:34:38]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/fischman@chromium.org/31583004/100001

[commit-bot: I haz the power, 2013-10-22 22:19:02]

Change committed as 230229