Add DIRECT_COMPOSITION GpuMemoryBuffer type.

Add DIRECT_COMPOSITION GpuMemoryBuffer type.

This is intended only for use as an onscreen swapchain, so it can't be
mapped in the renderer. It can be backed either by an IDXGISwapChain (if
asynchronous) or an IDCompositionSurface (if synchronous).

Currently this type can't be drawn into, but in the future the backing
surface will be able to be made current on a 3D context so it can be
written to. It will also be able to be used as an overlay.

BUG=726597
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

[jbauman, 2017-06-10 00:33:37]

Description was changed from

==========
Add DIRECT_COMPOSITION GpuMemoryBuffer type.

This is intended only for use as an onscreen swapchain, so it can't be
mapped in the renderer. It can be backed either by an IDXGISwapChain (if
asynchronous) or an IDCompositionSurface (if synchronous).

Currently it can't be drawn into, but in the future its backing surface
will be able to be made current on a 3D context so it can be written to.
It will also be able to be used as an overlay.

BUG=
==========

to

==========
Add DIRECT_COMPOSITION GpuMemoryBuffer type.

This is intended only for use as an onscreen swapchain, so it can't be
mapped in the renderer. It can be backed either by an IDXGISwapChain (if
asynchronous) or an IDCompositionSurface (if synchronous).

Currently it can't be drawn into, but in the future its backing surface
will be able to be made current on a 3D context so it can be written to.
It will also be able to be used as an overlay.

BUG=
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
==========

[jbauman, 2017-06-10 00:33:46]

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

[commit-bot: I haz the power, 2017-06-10 00:34:05]

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, 2017-06-10 00:41:38]

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

[commit-bot: I haz the power, 2017-06-10 00:41:39]

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

[jbauman, 2017-06-10 00:55:27]

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

[commit-bot: I haz the power, 2017-06-10 00:55:44]

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, 2017-06-10 01:03:34]

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

[commit-bot: I haz the power, 2017-06-10 01:03:35]

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

[jbauman, 2017-06-10 01:21:31]

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

[commit-bot: I haz the power, 2017-06-10 01:21:48]

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, 2017-06-10 01:31:10]

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

[commit-bot: I haz the power, 2017-06-10 01:31:11]

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

[jbauman, 2017-06-12 21:14:51]

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

[commit-bot: I haz the power, 2017-06-12 21:15:34]

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, 2017-06-12 21:30:01]

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

[commit-bot: I haz the power, 2017-06-12 21:30:02]

Dry run: Try jobs failed on following builders:
  linux_chromium_headless_rel on master.tryserver.chromium.linux (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/linux_chromium_...)

[jbauman, 2017-06-12 21:51:43]

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

[commit-bot: I haz the power, 2017-06-12 21:52:10]

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, 2017-06-12 22:05:20]

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

[commit-bot: I haz the power, 2017-06-12 22:05:21]

Dry run: Try jobs failed on following builders:
  cast_shell_linux on master.tryserver.chromium.linux (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/cast_shell_linu...)

[jbauman, 2017-06-13 00:30:39]

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

[commit-bot: I haz the power, 2017-06-13 00:30:57]

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, 2017-06-13 00:44:39]

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

[commit-bot: I haz the power, 2017-06-13 00:44:39]

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

[jbauman, 2017-06-13 02:06:00]

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

[commit-bot: I haz the power, 2017-06-13 02:06:15]

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, 2017-06-13 02:43:48]

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

[commit-bot: I haz the power, 2017-06-13 02:43:49]

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

[jbauman, 2017-06-13 21:01:56]

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

[commit-bot: I haz the power, 2017-06-13 21:02:34]

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, 2017-06-13 22:02:18]

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

[commit-bot: I haz the power, 2017-06-13 22:02:19]

Dry run: Try jobs failed on following builders:
  linux_chromium_chromeos_ozone_rel_ng on master.tryserver.chromium.linux
(JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/linux_chromium_...)

[jbauman, 2017-06-14 00:16:36]

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

[commit-bot: I haz the power, 2017-06-14 00:19:16]

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, 2017-06-14 02:15:35]

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

[commit-bot: I haz the power, 2017-06-14 02:15:36]

Dry run: This issue passed the CQ dry run.

[jbauman, 2017-06-14 21:03:47]

Description was changed from

==========
Add DIRECT_COMPOSITION GpuMemoryBuffer type.

This is intended only for use as an onscreen swapchain, so it can't be
mapped in the renderer. It can be backed either by an IDXGISwapChain (if
asynchronous) or an IDCompositionSurface (if synchronous).

Currently it can't be drawn into, but in the future its backing surface
will be able to be made current on a 3D context so it can be written to.
It will also be able to be used as an overlay.

BUG=
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

==========
Add DIRECT_COMPOSITION GpuMemoryBuffer type.

This is intended only for use as an onscreen swapchain, so it can't be
mapped in the renderer. It can be backed either by an IDXGISwapChain (if
asynchronous) or an IDCompositionSurface (if synchronous).

Currently this type can't be drawn into, but in the future the backing
surface will be able to be made current on a 3D context so it can be
written to. It will also be able to be used as an overlay.

BUG=726597
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
==========

[jbauman, 2017-06-14 21:03:56]

jbauman@chromium.org changed reviewers:
+ reveman@chromium.org

[reveman, 2017-06-14 21:35:59]

https://codereview.chromium.org/2930143004/diff/160001/gpu/ipc/service/gpu_me...
File gpu/ipc/service/gpu_memory_buffer_factory_win.h (right):

https://codereview.chromium.org/2930143004/diff/160001/gpu/ipc/service/gpu_me...
gpu/ipc/service/gpu_memory_buffer_factory_win.h:56: usage_map_;
The factory needs to be stateless to handle GPU process crashes correctly. Can
ownership be passed using a handle instead of of keeping track of allocated
buffers using this map?

https://codereview.chromium.org/2930143004/diff/160001/ui/gfx/buffer_types.h
File ui/gfx/buffer_types.h (right):

https://codereview.chromium.org/2930143004/diff/160001/ui/gfx/buffer_types.h#...
ui/gfx/buffer_types.h:46: SCANOUT_ASYNC,
Why do we need to add this here? How flips are synchronized doesn't seem like
something that buffer usage should be used for..

https://codereview.chromium.org/2930143004/diff/160001/ui/gl/gl_image.h
File ui/gl/gl_image.h (right):

https://codereview.chromium.org/2930143004/diff/160001/ui/gl/gl_image.h#newco...
ui/gl/gl_image.h:97: enum class Type { NONE, MEMORY, IOSURFACE, DXGI_IMAGE,
DIRECT_COMPOSITION };
This type of downcasting is an indication that're not able to create the proper
API and abstraction for what we're trying to do. I think we should strive
towards being able to remove this. Is this really needed in the direct
composition case?

https://codereview.chromium.org/2930143004/diff/160001/ui/gl/gl_image.h#newco...
ui/gl/gl_image.h:100: virtual scoped_refptr<GLSurface> GetSurface();
How is this suppose to be used? Can we do something similar to
ScheduleOverlayPlane instead?

[jbauman, 2017-06-14 22:18:17]

https://codereview.chromium.org/2930143004/diff/160001/gpu/ipc/service/gpu_me...
File gpu/ipc/service/gpu_memory_buffer_factory_win.h (right):

https://codereview.chromium.org/2930143004/diff/160001/gpu/ipc/service/gpu_me...
gpu/ipc/service/gpu_memory_buffer_factory_win.h:56: usage_map_;
On 2017/06/14 21:35:58, reveman wrote:
> The factory needs to be stateless to handle GPU process crashes correctly. Can
> ownership be passed using a handle instead of of keeping track of allocated
> buffers using this map?

I could put the buffer usage in the handle so it could be recreated in a
different process. There's no way to access DirectComposition surfaces outside
the process that created them, so I can't actually own them with a handle.

https://codereview.chromium.org/2930143004/diff/160001/ui/gfx/buffer_types.h
File ui/gfx/buffer_types.h (right):

https://codereview.chromium.org/2930143004/diff/160001/ui/gfx/buffer_types.h#...
ui/gfx/buffer_types.h:46: SCANOUT_ASYNC,
On 2017/06/14 21:35:59, reveman wrote:
> Why do we need to add this here? How flips are synchronized doesn't seem like
> something that buffer usage should be used for..

I wasn't sure how to specify the type of buffer needed (IDCompositionSurface or
IDXGISwapChain) to GpuMemoryBufferManager::CreateGpuMemoryBuffer. Would you want
me to add an extra argument to that?

https://codereview.chromium.org/2930143004/diff/160001/ui/gl/gl_image.h
File ui/gl/gl_image.h (right):

https://codereview.chromium.org/2930143004/diff/160001/ui/gl/gl_image.h#newco...
ui/gl/gl_image.h:97: enum class Type { NONE, MEMORY, IOSURFACE, DXGI_IMAGE,
DIRECT_COMPOSITION };
On 2017/06/14 21:35:59, reveman wrote:
> This type of downcasting is an indication that're not able to create the
proper
> API and abstraction for what we're trying to do. I think we should strive
> towards being able to remove this. Is this really needed in the direct
> composition case?

I guess I could have windows-only GetSwapChain() and GetDCompositionSurface()
methods that would be implemented only by the DirectCompositionChildSurfaceWin,
if you want.

https://codereview.chromium.org/2930143004/diff/160001/ui/gl/gl_image.h#newco...
ui/gl/gl_image.h:100: virtual scoped_refptr<GLSurface> GetSurface();
On 2017/06/14 21:35:59, reveman wrote:
> How is this suppose to be used? Can we do something similar to
> ScheduleOverlayPlane instead?

This is actually going to be used to draw into the texture. The
DirectComposition surfaces can't really be used as FBOs, so we'll need to have a
new GL command that makes a GLImage current on the context. This will retrieve
the GLSurface that will actually be made current.