V4L2VDA: Fix format initialization order.

V4L2VDA: Fix format initialization order.

The correct order of initialization is to set up both input and output
formats before starting streaming, as the device may have limitations on
supported output formats. Supported output formats may depend on the input
format, so we set input format first and then query available output
formats for it.
Moreover, since format cannot be changed after buffers are requested
for a queue, we need to de facto set both formats up before allocation.

BUG=430638
TEST=video playback on affected platforms

Committed: https://crrev.com/c5c56aed4d336726d6059dcb0a40101459d3703a
Cr-Commit-Position: refs/heads/master@{#310259}

[Pawel Osciak, 2015-01-06 05:16:25]

posciak@chromium.org changed reviewers:
+ wuchengli@chromium.org

[Pawel Osciak, 2015-01-06 05:16:30]

ptal

[wuchengli, 2015-01-06 08:45:16]

https://codereview.chromium.org/836093002/diff/1/content/common/gpu/media/v4l...
File content/common/gpu/media/v4l2_video_decode_accelerator.cc (right):

https://codereview.chromium.org/836093002/diff/1/content/common/gpu/media/v4l...
content/common/gpu/media/v4l2_video_decode_accelerator.cc:284: if
(!CreateInputBuffers())
Why this is moved here?

https://codereview.chromium.org/836093002/diff/1/content/common/gpu/media/v4l...
content/common/gpu/media/v4l2_video_decode_accelerator.cc:1077: if
(dqbuf.m.planes[0].bytesused == 0) {
Is this related to format initialization order?

https://codereview.chromium.org/836093002/diff/1/content/common/gpu/media/v4l...
content/common/gpu/media/v4l2_video_decode_accelerator.cc:1706: __u32
input_format_fourcc =
indent

[Pawel Osciak, 2015-01-06 13:23:06]

https://codereview.chromium.org/836093002/diff/1/content/common/gpu/media/v4l...
File content/common/gpu/media/v4l2_video_decode_accelerator.cc (right):

https://codereview.chromium.org/836093002/diff/1/content/common/gpu/media/v4l...
content/common/gpu/media/v4l2_video_decode_accelerator.cc:284: if
(!CreateInputBuffers())
On 2015/01/06 08:45:16, wuchengli wrote:
> Why this is moved here?

I wanted to finish all the initialization before creating buffers. Not really a
big difference, but why not.

https://codereview.chromium.org/836093002/diff/1/content/common/gpu/media/v4l...
content/common/gpu/media/v4l2_video_decode_accelerator.cc:1077: if
(dqbuf.m.planes[0].bytesused == 0) {
On 2015/01/06 08:45:16, wuchengli wrote:
> Is this related to format initialization order?

No, this should've gone to the other enum formats CL. I had the CLs stacked for
testing, sorry about that. Removing.

https://codereview.chromium.org/836093002/diff/1/content/common/gpu/media/v4l...
content/common/gpu/media/v4l2_video_decode_accelerator.cc:1706: __u32
input_format_fourcc =
On 2015/01/06 08:45:16, wuchengli wrote:
> indent

Sorry where? Perhaps you missed the "__" before "u32"...

[Pawel Osciak, 2015-01-06 13:23:19]

posciak@chromium.org changed reviewers:
+ xhwang@chromium.org

[Pawel Osciak, 2015-01-06 13:23:49]

xhwang@: ptal. Thanks.

[Pawel Osciak, 2015-01-06 13:24:29]

On 2015/01/06 13:23:49, Pawel Osciak wrote:
> xhwang@: ptal. Thanks.

Uh oh, you are not an OWNER, sorry about that.

[Pawel Osciak, 2015-01-06 13:24:43]

posciak@chromium.org changed reviewers:
+ scherkus@chromium.org
- xhwang@chromium.org

[Pawel Osciak, 2015-01-06 13:25:00]

scherkus@: ptal. Thanks.

[wuchengli, 2015-01-06 13:55:26]

LGTM

[scherkus (not reviewing), 2015-01-06 22:22:53]

lgtm

[Pawel Osciak, 2015-01-07 08:54:30]

The CQ bit was checked by posciak@chromium.org

[commit-bot: I haz the power, 2015-01-07 08:55:28]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/836093002/20001

[commit-bot: I haz the power, 2015-01-07 09:37:43]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2015-01-07 09:38:39]

Patchset 2 (id:??) landed as
https://crrev.com/c5c56aed4d336726d6059dcb0a40101459d3703a
Cr-Commit-Position: refs/heads/master@{#310259}