Introduce gfx::BufferFormat::YUYV_422

Introduce gfx::BufferFormat::YUYV_422

YUYV is only format supported by all linux stack such as kernel, mesa, minigbm,
and gpu decoder. So ChromeOS IA prefers to use YUYV for video decoding.

crrev.com/2648633005 depends on this CL.

BUG=683347
TEST=gl_unittests, ozone_unittests
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.android:android_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel

[dshwang, 2017-02-09 03:43:52]

Description was changed from

==========
Introduce gfx::BufferFormat::YUYV_422

YUYV is only format supported by all linux stack such as kernel, mesa, minigbm,
and gpu decoder. So ChromeOS IA prefers to use YUYV for video decoding.

crrev.com/2648633005 depends on this CL.

BUG=683347
TEST=gl_unittests, ozone_unittests
==========

to

==========
Introduce gfx::BufferFormat::YUYV_422

YUYV is only format supported by all linux stack such as kernel, mesa, minigbm,
and gpu decoder. So ChromeOS IA prefers to use YUYV for video decoding.

crrev.com/2648633005 depends on this CL.

BUG=683347
TEST=gl_unittests, ozone_unittests
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel
==========

[inactive_dshwang_plz_cc_intel, 2017-02-09 03:44:01]

The CQ bit was checked by dongseong.hwang@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2017-02-09 03:45:01]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[dshwang, 2017-02-09 03:47:27]

dongseong.hwang@intel.com changed reviewers:
+ bajones@chromium.org, ccameron@chromium.org, dcastagna@chromium.org,
reveman@chromium.org

[dshwang, 2017-02-09 03:47:27]

dcastagna@, as you reviewed in https://codereview.chromium.org/2648633005/, YUYV
introduction part is extracted to here.

bajones@: could you review gpu/?
ccameron@: could you review ui/?
reveman@: could you review component/exo?

[commit-bot: I haz the power, 2017-02-09 03:47:44]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2017-02-09 03:47:45]

Dry run: Try jobs failed on following builders:
  ios-device on master.tryserver.chromium.mac (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.mac/builders/ios-device/builds...)
  ios-device-xcode-clang on master.tryserver.chromium.mac (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.mac/builders/ios-device-xcode-...)
  ios-simulator-xcode-clang on master.tryserver.chromium.mac (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.mac/builders/ios-simulator-xco...)

[Daniele Castagna, 2017-02-09 03:55:15]

ccing marcheu

[inactive_dshwang_plz_cc_intel, 2017-02-09 03:55:30]

The CQ bit was checked by dongseong.hwang@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2017-02-09 03:56:22]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2017-02-09 06:15:05]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2017-02-09 06:15:06]

Dry run: This issue passed the CQ dry run.

[dshwang, 2017-02-10 23:33:10]

On 2017/02/09 03:55:15, Daniele Castagna wrote:
> ccing marcheu

Thanks for ccing.
marcheu@, could you review?

[dshwang, 2017-02-21 21:53:14]

Description was changed from

==========
Introduce gfx::BufferFormat::YUYV_422

YUYV is only format supported by all linux stack such as kernel, mesa, minigbm,
and gpu decoder. So ChromeOS IA prefers to use YUYV for video decoding.

crrev.com/2648633005 depends on this CL.

BUG=683347
TEST=gl_unittests, ozone_unittests
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel
==========

to

==========
Introduce gfx::BufferFormat::YUYV_422

YUYV is only format supported by all linux stack such as kernel, mesa, minigbm,
and gpu decoder. So ChromeOS IA prefers to use YUYV for video decoding.

crrev.com/2648633005 depends on this CL.

BUG=683347
TEST=gl_unittests, ozone_unittests
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel
==========

[dshwang, 2017-02-21 21:53:14]

dongseong.hwang@intel.com changed reviewers:
+ marcheu@google.com

[dshwang, 2017-02-21 22:04:07]

marcheu@, could you review?

This prototype consists of 6 CLs.
cc: UYVY video is not premultiplied rgba
https://codereview.chromium.org/2684073006/
Introduce gfx::BufferFormat::YUYV_422 https://codereview.chromium.org/2689453002
Make OverlayCandidate use gfx::BufferFormat instead of cc::ResourceFormat
https://codereview.chromium.org/2683763003/
gpu: merge gpu/ipc/host/gpu_memory_buffer_support.cc to gpu/ipc/common/
https://codereview.chromium.org/2649553003/
chromeos: gpu video decoder on Intel decodes video in YUYV format, instead of
RGBA. https://codereview.chromium.org/2678343011/
ozone: prefer YUYV format for overlay.
https://codereview.chromium.org/2686903003/
cros: Enable YUYV GPU memory buffer video frames
https://codereview.chromium.org/2648633005/

Except for the last CL, it's not related to GBM mmap. It's rather related to
enable YUYV gpu video decoding. It directly saves power consumption about 7% on
exisiting chromebooke when video playback as
https://codereview.chromium.org/2678343011/ mentions. So I think we can proceed
reviewing except for the last CL. What do you think?

[dshwang, 2017-02-28 02:18:24]

dcastagna@, could we proceed this review?

"mmap(dma_buf) failure" and this CL is not related. Gpu video decoder will use
YUYV dma_buf, and Gpu video decoder never use mmap().

[Daniele Castagna, 2017-02-28 03:09:40]

On 2017/02/28 at 02:18:24, dongseong.hwang wrote:
> dcastagna@, could we proceed this review?
> 
> "mmap(dma_buf) failure" and this CL is not related. Gpu video decoder will use
YUYV dma_buf, and Gpu video decoder never use mmap().

Well, it'd be nice to figure out what it's happening with mmap before proceeding
with new features since content is much more important than videos and it
affects any webpage.

Also, can you please answer me to the other CL first
(https://codereview.chromium.org/2686903003#msg14)?

[dshwang, 2017-02-28 04:05:15]

On 2017/02/28 03:09:40, Daniele Castagna wrote:
> On 2017/02/28 at 02:18:24, dongseong.hwang wrote:
> > dcastagna@, could we proceed this review?
> > 
> > "mmap(dma_buf) failure" and this CL is not related. Gpu video decoder will
use
> YUYV dma_buf, and Gpu video decoder never use mmap().
> 
> Well, it'd be nice to figure out what it's happening with mmap before
proceeding
> with new features since content is much more important than videos and it
> affects any webpage.

Ok, thank you for patient explanation. anyway, we now reproduce it 100% by
https://chromium-review.googlesource.com/c/447556/
We will fix it very soon :)

> Also, can you please answer me to the other CL first
> (https://codereview.chromium.org/2686903003#msg14)?

Done.

[dshwang, 2017-06-01 22:35:16]

Description was changed from

==========
Introduce gfx::BufferFormat::YUYV_422

YUYV is only format supported by all linux stack such as kernel, mesa, minigbm,
and gpu decoder. So ChromeOS IA prefers to use YUYV for video decoding.

crrev.com/2648633005 depends on this CL.

BUG=683347
TEST=gl_unittests, ozone_unittests
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel
==========

to

==========
Introduce gfx::BufferFormat::YUYV_422

YUYV is only format supported by all linux stack such as kernel, mesa, minigbm,
and gpu decoder. So ChromeOS IA prefers to use YUYV for video decoding.

crrev.com/2648633005 depends on this CL.

BUG=683347
TEST=gl_unittests, ozone_unittests
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.android:android_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel
==========

[inactive_dshwang_plz_cc_intel, 2017-06-01 22:35:20]

The CQ bit was checked by dongseong.hwang@chromium.org to run a CQ dry run

[dshwang, 2017-06-01 22:36:03]

Hi Daniele, could you resume to review it again? Now dma_buf issue is fixed. :)

[commit-bot: I haz the power, 2017-06-01 22:36:33]

Dry run: CQ is trying da patch.

Follow status at:
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2017-06-02 01:53:15]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2017-06-02 01:53:17]

Dry run: Try jobs failed on following builders:
  linux_android_rel_ng on master.tryserver.chromium.android (JOB_FAILED,
https://build.chromium.org/p/tryserver.chromium.android/builders/linux_androi...)

[dshwang, 2017-06-03 00:00:31]

Description was changed from

==========
Introduce gfx::BufferFormat::YUYV_422

YUYV is only format supported by all linux stack such as kernel, mesa, minigbm,
and gpu decoder. So ChromeOS IA prefers to use YUYV for video decoding.

crrev.com/2648633005 depends on this CL.

BUG=683347
TEST=gl_unittests, ozone_unittests
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.android:android_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel
==========

to

==========
Introduce gfx::BufferFormat::YUYV_422

YUYV is only format supported by all linux stack such as kernel, mesa, minigbm,
and gpu decoder. So ChromeOS IA prefers to use YUYV for video decoding.

crrev.com/2648633005 depends on this CL.

BUG=683347
TEST=gl_unittests, ozone_unittests
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.android:android_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel
==========

[dshwang, 2017-06-03 00:00:31]

dongseong.hwang@intel.com changed reviewers:
+ dnicoara@chromium.org

[dshwang, 2017-06-03 00:02:01]

On 2017/06/01 22:36:03, dshwang wrote:
> Hi Daniele, could you resume to review it again? Now dma_buf issue is fixed.
:)

dnicoara, could you review this YUYV introduction? I'll make VA video decoder
use YUYV instead of RGBA in crrev.com/2648633005

[Daniele Castagna, 2017-06-03 20:50:20]

Can UYVY be used instead? That is what we're using on Mac already. It'd be nice
to use the same format since we also have the code to convert to UYVY for SW
video decoders.

[dshwang, 2017-06-06 00:19:18]

On 2017/06/03 20:50:20, Daniele Castagna wrote:
> Can UYVY be used instead? That is what we're using on Mac already. It'd be
nice
> to use the same format since we also have the code to convert to UYVY for SW
> video decoders.

I agree. It's because mesa support YUYV for EGL Image, but not UYVY
mesa supports YUYV. refer to
src/third_party/mesa/src/mesa/drivers/dri/i965/intel_screen.c
minigbm supports YUYV and UYVY. refer to src/platform/minigbm/i915.c
linux supports YUYV and UYVY overlay. refer to
src/third_party/kernel/v4.4/drivers/gpu/drm/i915/intel_sprite.c
vaapi/intel-driver supports YUYV and UYVY decode. refer to src/i965_drv_video.c
in https://cgit.freedesktop.org/vaapi/intel-driver

YUYV support in mesa is added in 2013 and I'm not sure they are willing to
accept UYVY support.

Unfortunately, IOSurface supports only UYVY tho
https://developer.apple.com/documentation/corevideo/cvpixelformatdescription/...

[dshwang, 2017-06-08 23:38:01]

Discussed about UYVY support in mesa with Chad and Kristian. Mesa is willing to
accept UYUV support. Intel mesa engineers began implementing it.
On the other hands, I think NV12 is better choice while it's not overlayable in
IA yet. Could you review NV12 CL in https://codereview.chromium.org/2678343011/
?

[Daniele Castagna, 2017-06-13 19:20:43]

On 2017/06/08 at 23:38:01, dongseong.hwang wrote:
> Discussed about UYVY support in mesa with Chad and Kristian. Mesa is willing
to accept UYUV support. Intel mesa engineers began implementing it.
Great. This solution seems better.

> On the other hands, I think NV12 is better choice while it's not overlayable
in IA yet. Could you review NV12 CL in
https://codereview.chromium.org/2678343011/ ?
What do you mean by this? That NV12 is a better choice while we can't do HW
overlays? Will we ever be able to use HW overlays with NV12?

[dshwang, 2017-06-16 21:08:25]

On 2017/06/13 19:20:43, Daniele Castagna wrote:
> On 2017/06/08 at 23:38:01, dongseong.hwang wrote:
> > Discussed about UYVY support in mesa with Chad and Kristian. Mesa is willing
> to accept UYUV support. Intel mesa engineers began implementing it.
> Great. This solution seems better.

Johnson is working on UYUV support
https://lists.freedesktop.org/archives/mesa-dev/2017-June/159576.html

> > On the other hands, I think NV12 is better choice while it's not overlayable
> in IA yet. Could you review NV12 CL in
> https://codereview.chromium.org/2678343011/ ?
> What do you mean by this? That NV12 is a better choice while we can't do HW
> overlays? Will we ever be able to use HW overlays with NV12?

Gen9 Intel GPU hardware (i.e. Kabylake and Apollolake) supports NV12, but there
is missing bit in kernel.
Vidya is working in progress in upstream kernel.
https://patchwork.freedesktop.org/series/25377/
After it's landed, I'll backport it to cros kernel v4.4