Ensure that pepper plugins honor the GPU H/W video decode blacklist and the disable-accelerated-vid…

Ensure that pepper plugins honor the GPU H/W video decode blacklist and the disable-accelerated-video-decode switch

Pepper plugins like Flash should honor the GPU blacklist. Not doing that causes them to
try and use buggy video drivers for video decoding which does not end well.

This is done by adding a flag in the accelerated_video_decode_enabled in the WebPreferences structure
This is initialized along with the other preferences in the GpuDataManagerImplPrivate::UpdateRendererWebPrefs
function. This is also passed to PPAPI plugins via the ppapi::Preferences structure.
The flag defaults to false. The plugin process checks this flag and if it is false does
not attempt to create the video decoder.

BUG=348154
R=bbudge@chromium.org, cpu@chromium.org, jam@chromium.org, jamesr@chromium.org, jbauman@chromium.org, palmer@chromium.org, bbudge, cpu, jbauman

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=255473

[ananta, 2014-03-06 00:40:55]

+jamesr for webkit\common OWNERS stamp

[bbudge, 2014-03-06 01:04:03]

ppapi parts LGTM with a comment.

https://codereview.chromium.org/188143003/diff/1/ppapi/proxy/ppb_video_decode...
File ppapi/proxy/ppb_video_decoder_proxy.cc (right):

https://codereview.chromium.org/188143003/diff/1/ppapi/proxy/ppb_video_decode...
ppapi/proxy/ppb_video_decoder_proxy.cc:200: if
(!dispatcher->preferences().is_accelerated_video_decode_enabled)
This seems out of place in the middle of grabbing the 3d context. Could this
follow the 'if (!dispatcher)' statement above?

[ananta, 2014-03-06 01:06:45]

https://codereview.chromium.org/188143003/diff/1/ppapi/proxy/ppb_video_decode...
File ppapi/proxy/ppb_video_decoder_proxy.cc (right):

https://codereview.chromium.org/188143003/diff/1/ppapi/proxy/ppb_video_decode...
ppapi/proxy/ppb_video_decoder_proxy.cc:200: if
(!dispatcher->preferences().is_accelerated_video_decode_enabled)
On 2014/03/06 01:04:04, bbudge wrote:
> This seems out of place in the middle of grabbing the 3d context. Could this
> follow the 'if (!dispatcher)' statement above?
Done.

[jamesr, 2014-03-06 04:20:46]

Hmm, this bit doesn't appear to  have anything to do with Blink.  Do we not have
a way to pass preferences directly to a pepper plugin?

[ananta, 2014-03-06 18:59:47]

On 2014/03/06 04:20:46, jamesr wrote:
> Hmm, this bit doesn't appear to  have anything to do with Blink.  Do we not
have
> a way to pass preferences directly to a pepper plugin?

This is the way we pass preferences to pepper plugins today. There are a bunch
of fields in the WebPreferences structure
which are used by pepper. Examples below:-
pepper_3d_enabled, flash_3d_enabled,flash_stage3d_baseline_enabled

These are sent by the renderer to the pepper plugin when it initializes via an
IPC message.

[cpu_(ooo_6.6-7.5), 2014-03-06 20:45:54]

jamesr, according to the pepperistas this is the way.

We are happy to invent a new way but we would like also something merge-able to
34.

[cpu_(ooo_6.6-7.5), 2014-03-06 20:47:21]

lgtm

[jamesr, 2014-03-06 20:48:41]

Can you prefix the flag with "pepper_" or something to indicate that it's not
relevant for blink?

Otherwise lgtm

[jbauman, 2014-03-06 20:51:50]

lgtm

[ananta, 2014-03-06 20:55:01]

On 2014/03/06 20:48:41, jamesr wrote:
> Can you prefix the flag with "pepper_" or something to indicate that it's not
> relevant for blink?
> 
> Otherwise lgtm

Done

[ananta, 2014-03-06 20:57:59]

+jam for content OWNERS

[jam, 2014-03-06 21:45:00]

lgtm

[ananta, 2014-03-07 00:01:52]

+piman for ppapi\proxy
+palmer for content\public\common\common_param_traits_macros.h

[bbudge, 2014-03-07 00:04:38]

On 2014/03/07 00:01:52, ananta wrote:
> +piman for ppapi\proxy
> +palmer for content\public\common\common_param_traits_macros.h

I'm an OWNER for ppapi.
You'll need a security review for ppapi_messages.h though.

[palmer, 2014-03-07 00:15:29]

IPC security LGTM

[ananta, 2014-03-07 00:18:36]

Committed patchset #4 manually as r255473 (presubmit successful).