Add switch to use ipc command buffer in Android WebView

Add switch to use ipc command buffer in Android WebView

Still default to use in-process command buffer. Add switch to make
testing easier.

With ipc command buffer, no need to initialize bindings early in start
up. This is ok because ipc command buffer is guaranteed to start before
the in-process one on render thread is created.

Video still runs with context on render thread. Child compositor, WebGL
and accelerated canvas will run with ipc command buffer when switch
is present.

BUG=509702
Committed: https://crrev.com/1463e1e5261571689fca740aafb1bf156f722f9d
Cr-Commit-Position: refs/heads/master@{#339999}

[boliu, 2015-07-22 21:48:16]

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

[boliu, 2015-07-22 21:48:16]

ptal

got ~4 weeks left to turn this on by default :D

[no sievers, 2015-07-22 22:17:48]

https://codereview.chromium.org/1253553002/diff/1/content/browser/android/in_...
File content/browser/android/in_process/synchronous_compositor_factory_impl.cc
(right):

https://codereview.chromium.org/1253553002/diff/1/content/browser/android/in_...
content/browser/android/in_process/synchronous_compositor_factory_impl.cc:95:
bool lose_context_when_out_of_memory = true;
really? do you even handle lost contexts? :)

https://codereview.chromium.org/1253553002/diff/1/content/browser/android/in_...
content/browser/android/in_process/synchronous_compositor_factory_impl.cc:223:
RENDER_WORKER_CONTEXT);
you mean RENDER_COMPOSITOR_CONTEXT?

https://codereview.chromium.org/1253553002/diff/1/content/browser/android/in_...
content/browser/android/in_process/synchronous_compositor_factory_impl.cc:248:
DCHECK(use_ipc_command_buffer_);
DCHECK(!use_ipc_command_buffer)?

https://codereview.chromium.org/1253553002/diff/1/content/browser/android/in_...
content/browser/android/in_process/synchronous_compositor_factory_impl.cc:256:
DCHECK(use_ipc_command_buffer_);
same here? since this code was here before

[boliu, 2015-07-22 22:28:58]

https://codereview.chromium.org/1253553002/diff/1/content/browser/android/in_...
File content/browser/android/in_process/synchronous_compositor_factory_impl.cc
(right):

https://codereview.chromium.org/1253553002/diff/1/content/browser/android/in_...
content/browser/android/in_process/synchronous_compositor_factory_impl.cc:95:
bool lose_context_when_out_of_memory = true;
On 2015/07/22 22:17:47, sievers wrote:
> really? do you even handle lost contexts? :)

No.. but does it really matter? This has been true since forever.

https://codereview.chromium.org/1253553002/diff/1/content/browser/android/in_...
content/browser/android/in_process/synchronous_compositor_factory_impl.cc:223:
RENDER_WORKER_CONTEXT);
On 2015/07/22 22:17:47, sievers wrote:
> you mean RENDER_COMPOSITOR_CONTEXT?

Actually both. This is called for both the compositor and worker context. Pass
it in instead.

https://codereview.chromium.org/1253553002/diff/1/content/browser/android/in_...
content/browser/android/in_process/synchronous_compositor_factory_impl.cc:248:
DCHECK(use_ipc_command_buffer_);
On 2015/07/22 22:17:48, sievers wrote:
> DCHECK(!use_ipc_command_buffer)?

/facepalm

Fixed all 3

[no sievers, 2015-07-22 22:31:03]

lgtm

[no sievers, 2015-07-22 22:31:45]

On 2015/07/22 22:31:03, sievers wrote:
> lgtm

oh and... yay! hopefully we can remove the switch and make it the default :)

[boliu, 2015-07-22 22:33:17]

On 2015/07/22 22:31:45, sievers wrote:
> On 2015/07/22 22:31:03, sievers wrote:
> > lgtm
> 
> oh and... yay! hopefully we can remove the switch and make it the default :)

Default hopefully very soon. I still have some things to check, eg memory and
webgl tests etc.

But not removing da switch until it launches fully. I've been burned enough
times on release branches to learn...

[boliu, 2015-07-22 22:39:36]

The CQ bit was checked by boliu@chromium.org

[commit-bot: I haz the power, 2015-07-22 22:40:38]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1253553002/20001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1253553002/20001

[commit-bot: I haz the power, 2015-07-23 00:18:10]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2015-07-23 00:19:36]

Patchset 2 (id:??) landed as
https://crrev.com/1463e1e5261571689fca740aafb1bf156f722f9d
Cr-Commit-Position: refs/heads/master@{#339999}