Switch main GPU thread to Default Message Pump

Switch main GPU thread to Default Message Pump

Since all windows has been moved from the main GPU thread
to a separate dedicated thread there is no need to run
UI message pump on the main thread.

This change switches the main GPU thread, on OS_WIN only,
to MessagePumpDefault. This should make MessagePumpForGpu
unnecessary. I am going to remove it in a separate change.

CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.win:win_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.android:android_optional_gpu_tests_rel

BUG=653181
Committed: https://crrev.com/5599ccdfa0d4167f13880db3560d8ad829157069
Cr-Commit-Position: refs/heads/master@{#430758}

[stanisc, 2016-10-19 20:54:55]

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

[commit-bot: I haz the power, 2016-10-19 20:55:37]

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, 2016-10-19 22:20:11]

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

[commit-bot: I haz the power, 2016-10-19 22:20:13]

Dry run: This issue passed the CQ dry run.

[stanisc, 2016-10-26 20:20:47]

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

[commit-bot: I haz the power, 2016-10-26 20:21:17]

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, 2016-10-26 21:11:14]

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

[commit-bot: I haz the power, 2016-10-26 21:11:15]

Dry run: This issue passed the CQ dry run.

[stanisc, 2016-10-26 21:19:02]

Description was changed from

==========
Experiment - Default MessagePump on main GPU thread

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

to

==========
Switch main GPU thread to Default Message Pump

Since all windows has been moved from the main GPU thread
to a separate dedicated thread there is no need to run
UI message pump on the main thread. 

This change switches the main GPU thread, on OS_WIN only,
to MessagePumpDefault. This should make MessagePumpForGpu
unnecessary. I am going to remove it in a separate change.

BUG=653181
==========

[stanisc, 2016-10-26 21:24:05]

stanisc@chromium.org changed reviewers:
+ jbauman@chromium.org, kbr@chromium.org, vmiura@chromium.org

[stanisc, 2016-10-26 21:24:05]

jbauman@, please review the change.

+kbr@, +vmiura@ because this might impact
https://bugs.chromium.org/p/chromium/issues/detail?id=609252

[Ken Russell (switch to Gerrit), 2016-10-26 22:22:05]

kbr@chromium.org changed reviewers:
+ geofflang@chromium.org

[Ken Russell (switch to Gerrit), 2016-10-26 22:22:06]

I'm not sure we want to make this change. Does ANGLE's OpenGL backend on Windows
still require the ability to create child windows in the GPU process?

[Geoff Lang, 2016-10-27 15:43:08]

On 2016/10/26 22:22:06, Ken Russell wrote:
> I'm not sure we want to make this change. Does ANGLE's OpenGL backend on
Windows
> still require the ability to create child windows in the GPU process?

ANGLE's GL backed creates a window so that it has a device context but never
depends on events from that window (uses the default window proc).  We don't
create a child window per surface anymore.  I think we could try out this change
and see if any of the Windows GPU FYI bots have issues.

[stanisc, 2016-10-27 23:00:36]

On 2016/10/27 15:43:08, Geoff Lang wrote:
> On 2016/10/26 22:22:06, Ken Russell wrote:
> > I'm not sure we want to make this change. Does ANGLE's OpenGL backend on
> Windows
> > still require the ability to create child windows in the GPU process?
> 
> ANGLE's GL backed creates a window so that it has a device context but never
> depends on events from that window (uses the default window proc).  We don't
> create a child window per surface anymore.  I think we could try out this
change
> and see if any of the Windows GPU FYI bots have issues.

Geoff, are there any unit tests for this that I could try running locally?

[Geoff Lang, 2016-10-28 14:50:04]

On 2016/10/27 23:00:36, stanisc wrote:
> On 2016/10/27 15:43:08, Geoff Lang wrote:
> > On 2016/10/26 22:22:06, Ken Russell wrote:
> > > I'm not sure we want to make this change. Does ANGLE's OpenGL backend on
> > Windows
> > > still require the ability to create child windows in the GPU process?
> > 
> > ANGLE's GL backed creates a window so that it has a device context but never
> > depends on events from that window (uses the default window proc).  We don't
> > create a child window per surface anymore.  I think we could try out this
> change
> > and see if any of the Windows GPU FYI bots have issues.
> 
> Geoff, are there any unit tests for this that I could try running locally?

I would just try running chrome with --use-angle=gl and verify it can initialize
the gpu process.  The FYI waterfall would be able to test additional
configurations once it lands.

I also added the optional GPU tests for this CL

[Geoff Lang, 2016-10-28 14:50:34]

Description was changed from

==========
Switch main GPU thread to Default Message Pump

Since all windows has been moved from the main GPU thread
to a separate dedicated thread there is no need to run
UI message pump on the main thread. 

This change switches the main GPU thread, on OS_WIN only,
to MessagePumpDefault. This should make MessagePumpForGpu
unnecessary. I am going to remove it in a separate change.

BUG=653181
==========

to

==========
Switch main GPU thread to Default Message Pump

Since all windows has been moved from the main GPU thread
to a separate dedicated thread there is no need to run
UI message pump on the main thread. 

This change switches the main GPU thread, on OS_WIN only,
to MessagePumpDefault. This should make MessagePumpForGpu
unnecessary. I am going to remove it in a separate change.

CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.win:win_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.android:android_optional_gpu_tests_rel

BUG=653181
==========

[stanisc, 2016-10-28 19:43:56]

On 2016/10/28 14:50:04, Geoff Lang wrote:
> On 2016/10/27 23:00:36, stanisc wrote:
> > On 2016/10/27 15:43:08, Geoff Lang wrote:
> > > On 2016/10/26 22:22:06, Ken Russell wrote:
> > > > I'm not sure we want to make this change. Does ANGLE's OpenGL backend on
> > > Windows
> > > > still require the ability to create child windows in the GPU process?
> > > 
> > > ANGLE's GL backed creates a window so that it has a device context but
never
> > > depends on events from that window (uses the default window proc).  We
don't
> > > create a child window per surface anymore.  I think we could try out this
> > change
> > > and see if any of the Windows GPU FYI bots have issues.
> > 
> > Geoff, are there any unit tests for this that I could try running locally?
> 
> I would just try running chrome with --use-angle=gl and verify it can
initialize
> the gpu process.  The FYI waterfall would be able to test additional
> configurations once it lands.
> 
> I also added the optional GPU tests for this CL

I've verified that chrome gpu works with --use-angle=gl. It creates a GL surface
rather than native window surface and there is no UI thread at all in the GPU
process in that case.

[Ken Russell (switch to Gerrit), 2016-10-28 20:33:32]

On 2016/10/28 19:43:56, stanisc wrote:
> On 2016/10/28 14:50:04, Geoff Lang wrote:
> > On 2016/10/27 23:00:36, stanisc wrote:
> > > On 2016/10/27 15:43:08, Geoff Lang wrote:
> > > > On 2016/10/26 22:22:06, Ken Russell wrote:
> > > > > I'm not sure we want to make this change. Does ANGLE's OpenGL backend
on
> > > > Windows
> > > > > still require the ability to create child windows in the GPU process?
> > > > 
> > > > ANGLE's GL backed creates a window so that it has a device context but
> never
> > > > depends on events from that window (uses the default window proc).  We
> don't
> > > > create a child window per surface anymore.  I think we could try out
this
> > > change
> > > > and see if any of the Windows GPU FYI bots have issues.
> > > 
> > > Geoff, are there any unit tests for this that I could try running locally?
> > 
> > I would just try running chrome with --use-angle=gl and verify it can
> initialize
> > the gpu process.  The FYI waterfall would be able to test additional
> > configurations once it lands.
> > 
> > I also added the optional GPU tests for this CL
> 
> I've verified that chrome gpu works with --use-angle=gl. It creates a GL
surface
> rather than native window surface and there is no UI thread at all in the GPU
> process in that case.

Excellent; thanks for verifying that. LGTM from my standpoint then.

[jbauman, 2016-11-07 20:30:32]

lgtm

[stanisc, 2016-11-08 21:41:21]

The CQ bit was checked by stanisc@chromium.org

[commit-bot: I haz the power, 2016-11-08 21:42:17]

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, 2016-11-08 23:34:15]

Description was changed from

==========
Switch main GPU thread to Default Message Pump

Since all windows has been moved from the main GPU thread
to a separate dedicated thread there is no need to run
UI message pump on the main thread. 

This change switches the main GPU thread, on OS_WIN only,
to MessagePumpDefault. This should make MessagePumpForGpu
unnecessary. I am going to remove it in a separate change.

CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.win:win_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.android:android_optional_gpu_tests_rel

BUG=653181
==========

to

==========
Switch main GPU thread to Default Message Pump

Since all windows has been moved from the main GPU thread
to a separate dedicated thread there is no need to run
UI message pump on the main thread. 

This change switches the main GPU thread, on OS_WIN only,
to MessagePumpDefault. This should make MessagePumpForGpu
unnecessary. I am going to remove it in a separate change.

CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.win:win_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.android:android_optional_gpu_tests_rel

BUG=653181
==========

[commit-bot: I haz the power, 2016-11-08 23:34:16]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2016-11-08 23:41:48]

Description was changed from

==========
Switch main GPU thread to Default Message Pump

Since all windows has been moved from the main GPU thread
to a separate dedicated thread there is no need to run
UI message pump on the main thread. 

This change switches the main GPU thread, on OS_WIN only,
to MessagePumpDefault. This should make MessagePumpForGpu
unnecessary. I am going to remove it in a separate change.

CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.win:win_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.android:android_optional_gpu_tests_rel

BUG=653181
==========

to

==========
Switch main GPU thread to Default Message Pump

Since all windows has been moved from the main GPU thread
to a separate dedicated thread there is no need to run
UI message pump on the main thread.

This change switches the main GPU thread, on OS_WIN only,
to MessagePumpDefault. This should make MessagePumpForGpu
unnecessary. I am going to remove it in a separate change.

CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.win:win_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.android:android_optional_gpu_tests_rel

BUG=653181

Committed: https://crrev.com/5599ccdfa0d4167f13880db3560d8ad829157069
Cr-Commit-Position: refs/heads/master@{#430758}
==========

[commit-bot: I haz the power, 2016-11-08 23:41:49]

Patchset 2 (id:??) landed as
https://crrev.com/5599ccdfa0d4167f13880db3560d8ad829157069
Cr-Commit-Position: refs/heads/master@{#430758}