Provide functionality in the dxva decoder on Windows to use the angle created device object if poss…

Provide functionality in the dxva decoder on Windows to use the angle created device object if possible.

This uses the angle provided extensions to query the device object. This is used by default now if DX11
is enabled and for D3D9 when the disable-d3d11 flag is passed in.

The idea behind using angle's device is that it will improve playback performance by eliminating copies
across devices.

This is based on the work which sandersd initially started.

BUG=337606
Committed: https://crrev.com/1cef88d98d7777b826cb076b1e20f9874bb3cf57
Cr-Commit-Position: refs/heads/master@{#355477}

[ananta, 2015-10-10 02:08:23]

ananta@chromium.org changed reviewers:
+ dalecurtis@chromium.org, sandersd@chromium.org

[sandersd (OOO until July 31), 2015-10-12 18:44:03]

When I tried this, I wasn't able to use the D3D11 device for decoding. I assumed
that the problem was how Angle created it, but perhaps I misunderstood. How
widely has this been tested?

https://codereview.chromium.org/1394403004/diff/40001/content/common/gpu/medi...
File content/common/gpu/media/dxva_video_decode_accelerator.cc (right):

https://codereview.chromium.org/1394403004/diff/40001/content/common/gpu/medi...
content/common/gpu/media/dxva_video_decode_accelerator.cc:312: "Failed to obtain
eglQueryDeviceAttribEXT()",
The same message three times isn't too helpful for debugging, can you make the
messages unique?

[ananta, 2015-10-12 22:18:15]

I tested this on Windows 8 and Windows 10. On Windows 7 the angle device works
if we disable dx11.

https://codereview.chromium.org/1394403004/diff/40001/content/common/gpu/medi...
File content/common/gpu/media/dxva_video_decode_accelerator.cc (right):

https://codereview.chromium.org/1394403004/diff/40001/content/common/gpu/medi...
content/common/gpu/media/dxva_video_decode_accelerator.cc:312: "Failed to obtain
eglQueryDeviceAttribEXT()",
On 2015/10/12 18:44:02, sandersd wrote:
> The same message three times isn't too helpful for debugging, can you make the
> messages unique?

Done.

[sandersd (OOO until July 31), 2015-10-12 22:27:06]

LGTM.

[ananta, 2015-10-12 23:56:26]

On 2015/10/12 22:27:06, sandersd wrote:
> LGTM.

Thanks. Dale is on vacation till Oct 16th. Would it be reasonable to TBR this as
I am on vacation from tomorrow through
Oct 27th. I will look at the progress of this in Canary and will revert if this
causes fires.

[ananta, 2015-10-12 23:56:27]

On 2015/10/12 22:27:06, sandersd wrote:
> LGTM.

Thanks. Dale is on vacation till Oct 16th. Would it be reasonable to TBR this as
I am on vacation from tomorrow through
Oct 27th. I will look at the progress of this in Canary and will revert if this
causes fires.

[sandersd (OOO until July 31), 2015-10-13 00:24:47]

On 2015/10/12 23:56:27, ananta wrote:
> On 2015/10/12 22:27:06, sandersd wrote:
> > LGTM.
> 
> Thanks. Dale is on vacation till Oct 16th. Would it be reasonable to TBR this
as
> I am on vacation from tomorrow through
> Oct 27th. I will look at the progress of this in Canary and will revert if
this
> causes fires.

I don't think you need Dale's stamp for this one, but I am worried about
possible compatibility issues with this once it hits dev. Perhaps it would be
better to wait until you get back?

[ananta, 2015-10-13 00:48:08]

On 2015/10/13 00:24:47, sandersd wrote:
> On 2015/10/12 23:56:27, ananta wrote:
> > On 2015/10/12 22:27:06, sandersd wrote:
> > > LGTM.
> > 
> > Thanks. Dale is on vacation till Oct 16th. Would it be reasonable to TBR
this
> as
> > I am on vacation from tomorrow through
> > Oct 27th. I will look at the progress of this in Canary and will revert if
> this
> > causes fires.
> 
> I don't think you need Dale's stamp for this one, but I am worried about
> possible compatibility issues with this once it hits dev. Perhaps it would be
> better to wait until you get back?

Was wondering if we could see how this plays out in Canary before we take a call
on that. I will ask
QA to monitor it and let us know if there are issues. We could revert based on
that.

[sandersd (OOO until July 31), 2015-10-13 01:27:27]

On 2015/10/13 00:48:08, ananta wrote:
> On 2015/10/13 00:24:47, sandersd wrote:
> > On 2015/10/12 23:56:27, ananta wrote:
> > > On 2015/10/12 22:27:06, sandersd wrote:
> > > > LGTM.
> > > 
> > > Thanks. Dale is on vacation till Oct 16th. Would it be reasonable to TBR
> this
> > as
> > > I am on vacation from tomorrow through
> > > Oct 27th. I will look at the progress of this in Canary and will revert if
> > this
> > > causes fires.
> > 
> > I don't think you need Dale's stamp for this one, but I am worried about
> > possible compatibility issues with this once it hits dev. Perhaps it would
be
> > better to wait until you get back?
> 
> Was wondering if we could see how this plays out in Canary before we take a
call
> on that. I will ask
> QA to monitor it and let us know if there are issues. We could revert based on
> that.

I don't see any reason to rush, but I can take on the responsibility here if
that's necessary.

[DaleCurtis, 2015-10-22 00:40:51]

lgtm

[ananta, 2015-10-22 00:49:42]

The CQ bit was checked by ananta@chromium.org

[commit-bot: I haz the power, 2015-10-22 00:50:32]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1394403004/20002
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1394403004/20002

[commit-bot: I haz the power, 2015-10-22 02:18:12]

Committed patchset #4 (id:20002)

[commit-bot: I haz the power, 2015-10-22 02:19:06]

Patchset 4 (id:??) landed as
https://crrev.com/1cef88d98d7777b826cb076b1e20f9874bb3cf57
Cr-Commit-Position: refs/heads/master@{#355477}