aw: Ensure functor is called after binding init

aw: Ensure functor is called after binding init

Currently a crash from null bindings in functor can result
from this sequence:

* Compositor initialized
* BVR::OnDrawHardware sets hardware_enabled_, but renderer
  side has not been intialized so returns a null frame
* Detach or destroy webview, which will try to run functor
  for clean up since hardware_enabled_ is true.

The last step will race with GL binding init on gpu thread.
If functor runs first, then will result in crashes.

The small fix for now is make sure to not set
hardware_enabled_ until after receiving a hardware frame
from renderer side.

This crash is new in IPC-based sync compositing. The old
in-process path waits for renderer side to be up before
initializing browser side.

BUG=595811
Committed: https://crrev.com/a0f215e54b6ab4cf50e67ae7a2450e0327af6f7e
Cr-Commit-Position: refs/heads/master@{#383602}

[boliu, 2016-03-28 22:31:14]

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

[commit-bot: I haz the power, 2016-03-28 22:31:49]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1832423004/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1832423004/1

[boliu, 2016-03-28 22:32:05]

The CQ bit was unchecked by boliu@chromium.org

[boliu, 2016-03-28 22:34:11]

boliu@chromium.org changed reviewers:
+ hush@chromium.org

[boliu, 2016-03-28 22:34:12]

ptal

[hush (inactive), 2016-03-28 22:49:06]

On 2016/03/28 22:34:12, boliu wrote:
> ptal

lgtm

[hush (inactive), 2016-03-28 22:49:11]

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

[commit-bot: I haz the power, 2016-03-28 22:49:37]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1832423004/20001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1832423004/20001

[boliu, 2016-03-28 23:11:25]

Description was changed from

==========
aw: Ensure functor is called after binding init

BUG=595811
==========

to

==========
aw: Ensure functor is called after binding init

Currently a crash from null bindings in functor can result
from this sequence:

* Compositor initialized
* BVR::OnDrawHardware sets hardware_enabled_, but renderer
  side has not been intialized so returns a null frame
* Detach or destroy webview, which will try to run functor
  for clean up since hardware_enabled_ is true.

The last step will race with GL binding init on gpu thread.
If functor runs first, then will result in crashes.

The small fix for now is make sure to not set
hardware_enabled_ until after receiving a hardware frame
from renderer side.

This crash is new in IPC-based sync compositing. The old
in-process path waits for renderer side to be up before
initializing browser side.

BUG=595811
==========

[boliu, 2016-03-28 23:11:30]

The CQ bit was unchecked by boliu@chromium.org

[boliu, 2016-03-28 23:11:36]

The CQ bit was checked by boliu@chromium.org

[commit-bot: I haz the power, 2016-03-28 23:12:09]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1832423004/20001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1832423004/20001

[commit-bot: I haz the power, 2016-03-28 23:23:00]

Description was changed from

==========
aw: Ensure functor is called after binding init

Currently a crash from null bindings in functor can result
from this sequence:

* Compositor initialized
* BVR::OnDrawHardware sets hardware_enabled_, but renderer
  side has not been intialized so returns a null frame
* Detach or destroy webview, which will try to run functor
  for clean up since hardware_enabled_ is true.

The last step will race with GL binding init on gpu thread.
If functor runs first, then will result in crashes.

The small fix for now is make sure to not set
hardware_enabled_ until after receiving a hardware frame
from renderer side.

This crash is new in IPC-based sync compositing. The old
in-process path waits for renderer side to be up before
initializing browser side.

BUG=595811
==========

to

==========
aw: Ensure functor is called after binding init

Currently a crash from null bindings in functor can result
from this sequence:

* Compositor initialized
* BVR::OnDrawHardware sets hardware_enabled_, but renderer
  side has not been intialized so returns a null frame
* Detach or destroy webview, which will try to run functor
  for clean up since hardware_enabled_ is true.

The last step will race with GL binding init on gpu thread.
If functor runs first, then will result in crashes.

The small fix for now is make sure to not set
hardware_enabled_ until after receiving a hardware frame
from renderer side.

This crash is new in IPC-based sync compositing. The old
in-process path waits for renderer side to be up before
initializing browser side.

BUG=595811
==========

[commit-bot: I haz the power, 2016-03-28 23:23:01]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2016-03-28 23:24:36]

Description was changed from

==========
aw: Ensure functor is called after binding init

Currently a crash from null bindings in functor can result
from this sequence:

* Compositor initialized
* BVR::OnDrawHardware sets hardware_enabled_, but renderer
  side has not been intialized so returns a null frame
* Detach or destroy webview, which will try to run functor
  for clean up since hardware_enabled_ is true.

The last step will race with GL binding init on gpu thread.
If functor runs first, then will result in crashes.

The small fix for now is make sure to not set
hardware_enabled_ until after receiving a hardware frame
from renderer side.

This crash is new in IPC-based sync compositing. The old
in-process path waits for renderer side to be up before
initializing browser side.

BUG=595811
==========

to

==========
aw: Ensure functor is called after binding init

Currently a crash from null bindings in functor can result
from this sequence:

* Compositor initialized
* BVR::OnDrawHardware sets hardware_enabled_, but renderer
  side has not been intialized so returns a null frame
* Detach or destroy webview, which will try to run functor
  for clean up since hardware_enabled_ is true.

The last step will race with GL binding init on gpu thread.
If functor runs first, then will result in crashes.

The small fix for now is make sure to not set
hardware_enabled_ until after receiving a hardware frame
from renderer side.

This crash is new in IPC-based sync compositing. The old
in-process path waits for renderer side to be up before
initializing browser side.

BUG=595811

Committed: https://crrev.com/a0f215e54b6ab4cf50e67ae7a2450e0327af6f7e
Cr-Commit-Position: refs/heads/master@{#383602}
==========

[commit-bot: I haz the power, 2016-03-28 23:24:37]

Patchset 2 (id:??) landed as
https://crrev.com/a0f215e54b6ab4cf50e67ae7a2450e0327af6f7e
Cr-Commit-Position: refs/heads/master@{#383602}