Add threaded compositing mode to field trial

Add threaded compositing mode to field trial

This splits the force-compositing-mode field trial into 3 groups:
- disabled
- force-compositing-mode
- force-compositing-mode + threaded-compositing


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=147907

[jbates, 2012-07-19 01:39:42]

ilya: ptal
jam: OWNERS

[jam, 2012-07-19 16:31:45]

content lgtm

[Ilya Sherman, 2012-07-20 05:36:04]

LGTM

https://chromiumcodereview.appspot.com/10811002/diff/1/chrome/browser/gpu_uti...
File chrome/browser/gpu_util.cc (right):

https://chromiumcodereview.appspot.com/10811002/diff/1/chrome/browser/gpu_uti...
chrome/browser/gpu_util.cc:209: const base::FieldTrial::Probability kDivisor =
99;
Optional nit: Perhaps set this to 3 rather than to 99?  I think 1/3 is clearer
than 33/99, since it's not a percentage either way.

https://chromiumcodereview.appspot.com/10811002/diff/1/content/renderer/rende...
File content/renderer/render_thread_impl.cc (right):

https://chromiumcodereview.appspot.com/10811002/diff/1/content/renderer/rende...
content/renderer/render_thread_impl.cc:509: bool enable = (is_thread_trial ||
(has_enable && !has_disable)) && !is_guest;
Is it correct that the field trial should override the "disable" flag?  I would
have expected the flag to take precedence.

Actually, it looks like this is currently enforced within gpu_util when setting
up the trial -- maybe there should be a DCHECK here to double-check that
invariant?

(Same applies to the change in render_widget_host_impl.cc)

https://chromiumcodereview.appspot.com/10811002/diff/1/content/renderer/rende...
content/renderer/render_thread_impl.cc:515:
WebKit::WebCompositor::initialize(NULL);
nit: As long as you're editing nearby code, please add curly braces to the else
stmt, since the if-stmt needs them.

[jbates, 2012-07-23 17:47:00]

https://chromiumcodereview.appspot.com/10811002/diff/1/chrome/browser/gpu_uti...
File chrome/browser/gpu_util.cc (right):

https://chromiumcodereview.appspot.com/10811002/diff/1/chrome/browser/gpu_uti...
chrome/browser/gpu_util.cc:209: const base::FieldTrial::Probability kDivisor =
99;
On 2012/07/20 05:36:05, Ilya Sherman wrote:
> Optional nit: Perhaps set this to 3 rather than to 99?  I think 1/3 is clearer
> than 33/99, since it's not a percentage either way.

Done.

https://chromiumcodereview.appspot.com/10811002/diff/1/content/renderer/rende...
File content/renderer/render_thread_impl.cc (right):

https://chromiumcodereview.appspot.com/10811002/diff/1/content/renderer/rende...
content/renderer/render_thread_impl.cc:509: bool enable = (is_thread_trial ||
(has_enable && !has_disable)) && !is_guest;
On 2012/07/20 05:36:05, Ilya Sherman wrote:
> Is it correct that the field trial should override the "disable" flag?  I
would
> have expected the flag to take precedence.
> 
> Actually, it looks like this is currently enforced within gpu_util when
setting
> up the trial -- maybe there should be a DCHECK here to double-check that
> invariant?
> 
> (Same applies to the change in render_widget_host_impl.cc)

Done.

https://chromiumcodereview.appspot.com/10811002/diff/1/content/renderer/rende...
content/renderer/render_thread_impl.cc:515:
WebKit::WebCompositor::initialize(NULL);
On 2012/07/20 05:36:05, Ilya Sherman wrote:
> nit: As long as you're editing nearby code, please add curly braces to the
else
> stmt, since the if-stmt needs them.

Done.

[commit-bot: I haz the power, 2012-07-23 17:48:52]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/jbates@chromium.org/10811002/7001

[commit-bot: I haz the power, 2012-07-23 19:57:11]

Change committed as 147907