Ensure that the open gl context is current in the DXVA decoder while processing configuration chang…

Ensure that the open gl context is current in the DXVA decoder while processing configuration changes.

This should hopefully fix the DCHECK's which fire for a valid fence when the picture buffers are being torn down.

BUG=600323
Committed: https://crrev.com/0f3b2d37a14070c7d17b34053d3b345a844153fd
Cr-Commit-Position: refs/heads/master@{#385070}

[ananta, 2016-04-04 23:33:05]

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

[sandersd (OOO until July 31), 2016-04-04 23:35:46]

https://codereview.chromium.org/1861483004/diff/1/content/common/gpu/media/dx...
File content/common/gpu/media/dxva_video_decode_accelerator_win.cc (right):

https://codereview.chromium.org/1861483004/diff/1/content/common/gpu/media/dx...
content/common/gpu/media/dxva_video_decode_accelerator_win.cc:2835:
RETURN_AND_NOTIFY_ON_FAILURE(make_context_current_.Run(),
Should it be Invalidate() and/or Initialize() that do this?

[ananta, 2016-04-05 00:00:36]

https://codereview.chromium.org/1861483004/diff/1/content/common/gpu/media/dx...
File content/common/gpu/media/dxva_video_decode_accelerator_win.cc (right):

https://codereview.chromium.org/1861483004/diff/1/content/common/gpu/media/dx...
content/common/gpu/media/dxva_video_decode_accelerator_win.cc:2835:
RETURN_AND_NOTIFY_ON_FAILURE(make_context_current_.Run(),
On 2016/04/04 23:35:46, sandersd wrote:
> Should it be Invalidate() and/or Initialize() that do this?

In the general case Invalidate and Initialize get called in the context of the
IPC handler which makes the context current. I added a call to make the context
current in Invalidate as a best effort as it gets called during errors. We don't
need to do this in Initialize. 

For this case DismissStaleBuffers needs the context to be current. Making it
current in the top level call seems like the right way to go.

[jbauman, 2016-04-05 00:02:18]

https://codereview.chromium.org/1861483004/diff/1/content/common/gpu/media/dx...
File content/common/gpu/media/dxva_video_decode_accelerator_win.cc (right):

https://codereview.chromium.org/1861483004/diff/1/content/common/gpu/media/dx...
content/common/gpu/media/dxva_video_decode_accelerator_win.cc:2839:
DismissStaleBuffers(true);
DismissStaleBuffers is called from other places (e.g. directly as a callback),
so we should probably make it current there.

Also, we should probably do this for DeferredDismissStaleBuffer as well.

[ananta, 2016-04-05 00:22:10]

https://codereview.chromium.org/1861483004/diff/1/content/common/gpu/media/dx...
File content/common/gpu/media/dxva_video_decode_accelerator_win.cc (right):

https://codereview.chromium.org/1861483004/diff/1/content/common/gpu/media/dx...
content/common/gpu/media/dxva_video_decode_accelerator_win.cc:2839:
DismissStaleBuffers(true);
On 2016/04/05 00:02:18, jbauman wrote:
> DismissStaleBuffers is called from other places (e.g. directly as a callback),
> so we should probably make it current there.
> 
> Also, we should probably do this for DeferredDismissStaleBuffer as well. 

Done.

[sandersd (OOO until July 31), 2016-04-05 00:23:21]

Alright, lgtm.

[jbauman, 2016-04-05 00:32:32]

lgtm

[ananta, 2016-04-05 00:50:03]

The CQ bit was checked by ananta@chromium.org

[commit-bot: I haz the power, 2016-04-05 00:50:48]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1861483004/40001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1861483004/40001

[commit-bot: I haz the power, 2016-04-05 01:20:32]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2016-04-05 01:21:42]

Description was changed from

==========
Ensure that the open gl context is current in the DXVA decoder while processing
configuration changes.

This should hopefully fix the DCHECK's which fire for a valid fence when the
picture buffers are being torn down.

BUG=600323
==========

to

==========
Ensure that the open gl context is current in the DXVA decoder while processing
configuration changes.

This should hopefully fix the DCHECK's which fire for a valid fence when the
picture buffers are being torn down.

BUG=600323

Committed: https://crrev.com/0f3b2d37a14070c7d17b34053d3b345a844153fd
Cr-Commit-Position: refs/heads/master@{#385070}
==========

[commit-bot: I haz the power, 2016-04-05 01:21:43]

Patchset 3 (id:??) landed as
https://crrev.com/0f3b2d37a14070c7d17b34053d3b345a844153fd
Cr-Commit-Position: refs/heads/master@{#385070}