services/ui: Split GpuServiceInternal into gpu vs. ws pieces.

services/ui: Split GpuServiceInternal into gpu vs. ws pieces.

Notable changes:

. MusGpuMemoryBufferManager is meant to be used by the gpu host
  (i.e. GpuServiceProxy). So move it into ws. It still has a
  temporary dependency on GpuServiceInternal. It will converted
  to communicate with it over mojom API instead.

. Convert GpuServiceProxy (which is the gpu-host, running in the
  window server) into GpuChannelHostFactory, and responsible for
  managing the gpu-channel for the host.

. Introduce a |privileged| parameter in GpuServiceInternal API.
  This privileged channel is created only for the host (and by
  the host).

. Move ownership of GpuServiceProxy into WindowServer. It makes it
  easier for GpuServiceProxy to notify when the gpu-channel has
  been established through a Delegate interface.

. DisplayCompositor instance is now created in the window server
  after the gpu-channel has been established. The gpu channel is
  passed through PlatformDisplay and FrameGenerator from
  WindowServer.

BUG=613366
Committed: https://crrev.com/212f35700eeb17abc2c5e2e1a54e08c70c13ed23
Cr-Commit-Position: refs/heads/master@{#414901}

[sadrul, 2016-08-25 20:05:44]

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

[commit-bot: I haz the power, 2016-08-25 20:07:05]

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, 2016-08-25 20:12:32]

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

[commit-bot: I haz the power, 2016-08-25 20:12:32]

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...)
  linux_chromium_asan_rel_ng on master.tryserver.chromium.linux (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/linux_chromium_...)
  linux_chromium_chromeos_ozone_rel_ng on master.tryserver.chromium.linux
(JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/linux_chromium_...)

[sadrul, 2016-08-25 22:59:10]

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

[commit-bot: I haz the power, 2016-08-25 22:59:56]

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, 2016-08-25 23:05:54]

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

[commit-bot: I haz the power, 2016-08-25 23:05:54]

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...)
  chromeos_amd64-generic_chromium_compile_only_ng on
master.tryserver.chromium.linux (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/chromeos_amd64-...)
  chromeos_daisy_chromium_compile_only_ng on master.tryserver.chromium.linux
(JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/chromeos_daisy_...)
  chromeos_x86-generic_chromium_compile_only_ng on
master.tryserver.chromium.linux (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/chromeos_x86-ge...)
  linux_chromium_asan_rel_ng on master.tryserver.chromium.linux (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/linux_chromium_...)
  linux_chromium_chromeos_ozone_rel_ng on master.tryserver.chromium.linux
(JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/linux_chromium_...)
  linux_chromium_chromeos_rel_ng on master.tryserver.chromium.linux (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/linux_chromium_...)
  linux_chromium_clobber_rel_ng on master.tryserver.chromium.linux (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/linux_chromium_...)
  linux_chromium_rel_ng on master.tryserver.chromium.linux (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/linux_chromium_...)

[sadrul, 2016-08-26 05:45:23]

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

[commit-bot: I haz the power, 2016-08-26 05:45:53]

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, 2016-08-26 05:50:03]

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

[commit-bot: I haz the power, 2016-08-26 05:50:05]

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...)
  linux_chromium_asan_rel_ng on master.tryserver.chromium.linux (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/linux_chromium_...)
  linux_chromium_chromeos_ozone_rel_ng on master.tryserver.chromium.linux
(JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/linux_chromium_...)
  linux_chromium_chromeos_rel_ng on master.tryserver.chromium.linux (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/linux_chromium_...)
  linux_chromium_rel_ng on master.tryserver.chromium.linux (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/linux_chromium_...)

[sadrul, 2016-08-26 05:58:51]

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

[commit-bot: I haz the power, 2016-08-26 05:59:29]

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, 2016-08-26 06:18:05]

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

[commit-bot: I haz the power, 2016-08-26 06:18:06]

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

[sadrul, 2016-08-26 15:18:54]

Description was changed from

==========
[wip] Split GpuServiceInternal into gpu vs. ws pieces.

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

to

==========
services/ui: Split GpuServiceInternal into gpu vs. ws pieces.

Notable changes:

. MusGpuMemoryBufferManager is meant to be used by the gpu host
  (i.e. GpuServiceProxy). So move it into ws. It still has a
  temporary dependency on GpuServiceInternal. It will converted
  to communicate with it over mojom API instead.

. Convert GpuServiceProxy (which is the gpu-host, running in the
  window server) into GpuChannelHostFactory, and responsible for
  managing the gpu-channel for the host.

. Introduce a |privileged| parameter in GpuServiceInternal API.
  This privileged channel is created only for the host (and by
  the host).

. Move ownership of GpuServiceProxy into WindowServer. It makes it
  easier for GpuServiceProxy to notify when the gpu-channel has
  been established through a Delegate interface.

. DisplayCompositor instance is now created in the window server
  after the gpu-channel has been established. The gpu channel is
  passed through PlatformDisplay and FrameGenerator from
  WindowServer.

BUG=613366
==========

[sadrul, 2016-08-26 15:33:48]

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

[commit-bot: I haz the power, 2016-08-26 15:34:11]

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

[sadrul, 2016-08-26 15:36:34]

sadrul@chromium.org changed reviewers:
+ fsamuel@chromium.org, piman@chromium.org, sky@chromium.org,
tsepez@chromium.org

[sadrul, 2016-08-26 15:36:35]

tsepez@ for mojom changes.

sky@ for ws changes, specifically, moving the ownership of GpuServiceProxy from
Service into WindowServer, and plumbing of the GpuChannelHost from
GpuServiceProxy to DisplayCompositor.

fsamuel@ for the gpu changes.

piman@ for DEPS additions to gpu from services/ui/ws/DEPS

[commit-bot: I haz the power, 2016-08-26 15:37:07]

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

[commit-bot: I haz the power, 2016-08-26 15:37:08]

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...)
  chromeos_x86-generic_chromium_compile_only_ng on
master.tryserver.chromium.linux (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/chromeos_x86-ge...)
  chromium_presubmit on master.tryserver.chromium.linux (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/chromium_presub...)
  linux_chromium_chromeos_rel_ng on master.tryserver.chromium.linux (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/linux_chromium_...)
  mac_chromium_rel_ng on master.tryserver.chromium.mac (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.mac/builders/mac_chromium_rel_...)

[sadrul, 2016-08-26 15:47:55]

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

[commit-bot: I haz the power, 2016-08-26 15:48:14]

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

[Fady Samuel, 2016-08-26 16:27:59]

lgtm + optional bikeshed colors.

https://codereview.chromium.org/2281583003/diff/180001/services/ui/gpu/gpu_se...
File services/ui/gpu/gpu_service_internal.cc (right):

https://codereview.chromium.org/2281583003/diff/180001/services/ui/gpu/gpu_se...
services/ui/gpu/gpu_service_internal.cc:244: bool privileged,
Maybe be more explicit on what this means? is_gpu_host?

https://codereview.chromium.org/2281583003/diff/180001/services/ui/gpu/gpu_se...
File services/ui/gpu/gpu_service_internal.h (right):

https://codereview.chromium.org/2281583003/diff/180001/services/ui/gpu/gpu_se...
services/ui/gpu/gpu_service_internal.h:42: // This runs in the GPU process, and
communicates with the window observer over
window observer? Can we not mention windows here since this shouldn't depend on
anything window related?

https://codereview.chromium.org/2281583003/diff/180001/services/ui/ws/gpu_ser...
File services/ui/ws/gpu_service_proxy.cc (right):

https://codereview.chromium.org/2281583003/diff/180001/services/ui/ws/gpu_ser...
services/ui/ws/gpu_service_proxy.cc:59: io_thread_.reset(new
base::Thread("GPUIOThread"));
A TODO might be useful to verify whether we need this extra thread.

https://codereview.chromium.org/2281583003/diff/180001/services/ui/ws/gpu_ser...
File services/ui/ws/gpu_service_proxy.h (right):

https://codereview.chromium.org/2281583003/diff/180001/services/ui/ws/gpu_ser...
services/ui/ws/gpu_service_proxy.h:33: class GpuServiceProxy : public
mojom::GpuService,
Huge optional bikeshed color: GpuServiceProxy=>GpuServiceHost or better yet:

GpuServiceInternal => Gpu

GpuService => GpuHost

GpuServiceProxy => GpuHost.

https://codereview.chromium.org/2281583003/diff/180001/services/ui/ws/gpu_ser...
services/ui/ws/gpu_service_proxy.h:36: class Delegate {
Move this Delegate to a separate file?

https://codereview.chromium.org/2281583003/diff/180001/services/ui/ws/mus_gpu...
File services/ui/ws/mus_gpu_memory_buffer_manager.h (left):

https://codereview.chromium.org/2281583003/diff/180001/services/ui/ws/mus_gpu...
services/ui/ws/mus_gpu_memory_buffer_manager.h:20: class
MusGpuMemoryBufferManager : public gpu::GpuMemoryBufferManager {
Optional bikeshed color: remove the word Mus here?

[sky, 2016-08-26 16:46:30]

I think it's a bit weird to have the ws namespace in a couple of files, but I'm
ok with it. LGTM

https://codereview.chromium.org/2281583003/diff/180001/services/ui/ws/gpu_ser...
File services/ui/ws/gpu_service_proxy.h (right):

https://codereview.chromium.org/2281583003/diff/180001/services/ui/ws/gpu_ser...
services/ui/ws/gpu_service_proxy.h:36: class Delegate {
On 2016/08/26 16:27:59, Fady Samuel wrote:
> Move this Delegate to a separate file?

+1

https://codereview.chromium.org/2281583003/diff/180001/services/ui/ws/platfor...
File services/ui/ws/platform_display.h (right):

https://codereview.chromium.org/2281583003/diff/180001/services/ui/ws/platfor...
services/ui/ws/platform_display.h:99: // Notifies the PlatformDisplay that a
connection to gpu has been established.
'gpu' -> 'the gpu'

[sadrul, 2016-08-26 17:20:34]

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

[commit-bot: I haz the power, 2016-08-26 17:20:57]

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

[sadrul, 2016-08-26 17:23:16]

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

[commit-bot: I haz the power, 2016-08-26 17:23:47]

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

[sadrul, 2016-08-26 17:23:50]

https://codereview.chromium.org/2281583003/diff/180001/services/ui/gpu/gpu_se...
File services/ui/gpu/gpu_service_internal.cc (right):

https://codereview.chromium.org/2281583003/diff/180001/services/ui/gpu/gpu_se...
services/ui/gpu/gpu_service_internal.cc:244: bool privileged,
On 2016/08/26 16:27:59, Fady Samuel wrote:
> Maybe be more explicit on what this means? is_gpu_host?

Done.

https://codereview.chromium.org/2281583003/diff/180001/services/ui/gpu/gpu_se...
File services/ui/gpu/gpu_service_internal.h (right):

https://codereview.chromium.org/2281583003/diff/180001/services/ui/gpu/gpu_se...
services/ui/gpu/gpu_service_internal.h:42: // This runs in the GPU process, and
communicates with the window observer over
On 2016/08/26 16:27:59, Fady Samuel wrote:
> window observer? Can we not mention windows here since this shouldn't depend
on
> anything window related?

Meant window server. Fixed.

https://codereview.chromium.org/2281583003/diff/180001/services/ui/ws/gpu_ser...
File services/ui/ws/gpu_service_proxy.cc (right):

https://codereview.chromium.org/2281583003/diff/180001/services/ui/ws/gpu_ser...
services/ui/ws/gpu_service_proxy.cc:59: io_thread_.reset(new
base::Thread("GPUIOThread"));
On 2016/08/26 16:27:59, Fady Samuel wrote:
> A TODO might be useful to verify whether we need this extra thread.

GpuChannelHost creates a sync ipc-channel, and that needs an explicitly separate
IO thread
(https://cs.chromium.org/chromium/src/ipc/ipc_sync_channel.cc?sq=package:chrom...).

https://codereview.chromium.org/2281583003/diff/180001/services/ui/ws/gpu_ser...
File services/ui/ws/gpu_service_proxy.h (right):

https://codereview.chromium.org/2281583003/diff/180001/services/ui/ws/gpu_ser...
services/ui/ws/gpu_service_proxy.h:33: class GpuServiceProxy : public
mojom::GpuService,
On 2016/08/26 16:27:59, Fady Samuel wrote:
> Huge optional bikeshed color: GpuServiceProxy=>GpuServiceHost or better yet:
> 
> GpuServiceInternal => Gpu
> 
> GpuService => GpuHost
> 
> GpuServiceProxy => GpuHost.

GpuServiceInternal could be named Gpu. I don't have a strong preference either
way. I slightly prefer GpuServiceInternal because that makes it more obvious to
me that it's an internal mus API that isn't exposed to clients.

re GpuService: I think it'd be useful to have separate names for the
communication API for client->window-server, and the API for gpu->window-server.
The first one (client->window) is currently called mojom.GpuService, and the
latter will be called mojom.GpuHost. GpuServiceProxy will implement this
mojom.GpuHost interface, so we could call it GpuHost at that time.

I am going to keep the current names for now in this CL. We can do no-op renames
separately.

(pic:
https://docs.google.com/presentation/d/1ieNtQNpWmVMkMqA6FrHC1bUPg2mkZ_zGVDKPk...)

https://codereview.chromium.org/2281583003/diff/180001/services/ui/ws/gpu_ser...
services/ui/ws/gpu_service_proxy.h:36: class Delegate {
On 2016/08/26 16:27:59, Fady Samuel wrote:
> Move this Delegate to a separate file?

Done.

https://codereview.chromium.org/2281583003/diff/180001/services/ui/ws/mus_gpu...
File services/ui/ws/mus_gpu_memory_buffer_manager.h (left):

https://codereview.chromium.org/2281583003/diff/180001/services/ui/ws/mus_gpu...
services/ui/ws/mus_gpu_memory_buffer_manager.h:20: class
MusGpuMemoryBufferManager : public gpu::GpuMemoryBufferManager {
On 2016/08/26 16:27:59, Fady Samuel wrote:
> Optional bikeshed color: remove the word Mus here?

I plan to refactor+share this with the chrome code for crbug.com/637923. OK to
keep the current name until then?

https://codereview.chromium.org/2281583003/diff/180001/services/ui/ws/platfor...
File services/ui/ws/platform_display.h (right):

https://codereview.chromium.org/2281583003/diff/180001/services/ui/ws/platfor...
services/ui/ws/platform_display.h:99: // Notifies the PlatformDisplay that a
connection to gpu has been established.
On 2016/08/26 16:46:30, sky wrote:
> 'gpu' -> 'the gpu'

Done.

[commit-bot: I haz the power, 2016-08-26 18:18:58]

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

[commit-bot: I haz the power, 2016-08-26 18:18:59]

Dry run: Try jobs failed on following builders:
  mac_chromium_rel_ng on master.tryserver.chromium.mac (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.mac/builders/mac_chromium_rel_...)

[Tom Sepez, 2016-08-26 18:51:37]

lgtm

[sadrul, 2016-08-26 20:08:09]

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

[commit-bot: I haz the power, 2016-08-26 20:08:56]

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

[no sievers, 2016-08-26 20:10:02]

gpu deps lgtm

[commit-bot: I haz the power, 2016-08-26 21:14:03]

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

[commit-bot: I haz the power, 2016-08-26 21:14:05]

Dry run: This issue passed the CQ dry run.

[sadrul, 2016-08-27 00:13:59]

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

[commit-bot: I haz the power, 2016-08-27 00:14:39]

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

[sadrul, 2016-08-27 00:21:23]

The CQ bit was checked by sadrul@chromium.org

[sadrul, 2016-08-27 00:21:24]

The patchset sent to the CQ was uploaded after l-g-t-m from
fsamuel@chromium.org, sky@chromium.org, tsepez@chromium.org,
sievers@chromium.org
Link to the patchset: https://codereview.chromium.org/2281583003/#ps260001
(title: "tot merge")

[commit-bot: I haz the power, 2016-08-27 00:22:25]

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, 2016-08-27 10:48:09]

Description was changed from

==========
services/ui: Split GpuServiceInternal into gpu vs. ws pieces.

Notable changes:

. MusGpuMemoryBufferManager is meant to be used by the gpu host
  (i.e. GpuServiceProxy). So move it into ws. It still has a
  temporary dependency on GpuServiceInternal. It will converted
  to communicate with it over mojom API instead.

. Convert GpuServiceProxy (which is the gpu-host, running in the
  window server) into GpuChannelHostFactory, and responsible for
  managing the gpu-channel for the host.

. Introduce a |privileged| parameter in GpuServiceInternal API.
  This privileged channel is created only for the host (and by
  the host).

. Move ownership of GpuServiceProxy into WindowServer. It makes it
  easier for GpuServiceProxy to notify when the gpu-channel has
  been established through a Delegate interface.

. DisplayCompositor instance is now created in the window server
  after the gpu-channel has been established. The gpu channel is
  passed through PlatformDisplay and FrameGenerator from
  WindowServer.

BUG=613366
==========

to

==========
services/ui: Split GpuServiceInternal into gpu vs. ws pieces.

Notable changes:

. MusGpuMemoryBufferManager is meant to be used by the gpu host
  (i.e. GpuServiceProxy). So move it into ws. It still has a
  temporary dependency on GpuServiceInternal. It will converted
  to communicate with it over mojom API instead.

. Convert GpuServiceProxy (which is the gpu-host, running in the
  window server) into GpuChannelHostFactory, and responsible for
  managing the gpu-channel for the host.

. Introduce a |privileged| parameter in GpuServiceInternal API.
  This privileged channel is created only for the host (and by
  the host).

. Move ownership of GpuServiceProxy into WindowServer. It makes it
  easier for GpuServiceProxy to notify when the gpu-channel has
  been established through a Delegate interface.

. DisplayCompositor instance is now created in the window server
  after the gpu-channel has been established. The gpu channel is
  passed through PlatformDisplay and FrameGenerator from
  WindowServer.

BUG=613366
==========

[commit-bot: I haz the power, 2016-08-27 10:48:10]

Committed patchset #14 (id:260001)

[commit-bot: I haz the power, 2016-08-27 10:49:57]

Description was changed from

==========
services/ui: Split GpuServiceInternal into gpu vs. ws pieces.

Notable changes:

. MusGpuMemoryBufferManager is meant to be used by the gpu host
  (i.e. GpuServiceProxy). So move it into ws. It still has a
  temporary dependency on GpuServiceInternal. It will converted
  to communicate with it over mojom API instead.

. Convert GpuServiceProxy (which is the gpu-host, running in the
  window server) into GpuChannelHostFactory, and responsible for
  managing the gpu-channel for the host.

. Introduce a |privileged| parameter in GpuServiceInternal API.
  This privileged channel is created only for the host (and by
  the host).

. Move ownership of GpuServiceProxy into WindowServer. It makes it
  easier for GpuServiceProxy to notify when the gpu-channel has
  been established through a Delegate interface.

. DisplayCompositor instance is now created in the window server
  after the gpu-channel has been established. The gpu channel is
  passed through PlatformDisplay and FrameGenerator from
  WindowServer.

BUG=613366
==========

to

==========
services/ui: Split GpuServiceInternal into gpu vs. ws pieces.

Notable changes:

. MusGpuMemoryBufferManager is meant to be used by the gpu host
  (i.e. GpuServiceProxy). So move it into ws. It still has a
  temporary dependency on GpuServiceInternal. It will converted
  to communicate with it over mojom API instead.

. Convert GpuServiceProxy (which is the gpu-host, running in the
  window server) into GpuChannelHostFactory, and responsible for
  managing the gpu-channel for the host.

. Introduce a |privileged| parameter in GpuServiceInternal API.
  This privileged channel is created only for the host (and by
  the host).

. Move ownership of GpuServiceProxy into WindowServer. It makes it
  easier for GpuServiceProxy to notify when the gpu-channel has
  been established through a Delegate interface.

. DisplayCompositor instance is now created in the window server
  after the gpu-channel has been established. The gpu channel is
  passed through PlatformDisplay and FrameGenerator from
  WindowServer.

BUG=613366

Committed: https://crrev.com/212f35700eeb17abc2c5e2e1a54e08c70c13ed23
Cr-Commit-Position: refs/heads/master@{#414901}
==========

[commit-bot: I haz the power, 2016-08-27 10:49:58]

Patchset 14 (id:??) landed as
https://crrev.com/212f35700eeb17abc2c5e2e1a54e08c70c13ed23
Cr-Commit-Position: refs/heads/master@{#414901}