Mac: move content::RenderWidgetResizeHelper to ui::WindowResizeHelperMac (acelerated_widget_mac)

Mac: move content::RenderWidgetResizeHelper to ui::WindowResizeHelperMac (acelerated_widget_mac)

We need composited UI on Mac to cooperate with the GPU process in a
similar way to how a Mac RenderWidget does this for composited content.

By separating out the content dependencies from
RenderWidgetResizeHelper, it then becomes a reasonably straightforward
move to src/ui/accelerated_widget_mac.

This allows resize operations performed on views::BridgedNativeWidget to
hook into the same resize task coordination paths as the
RenderWidgetHostView.

One tricky bit: a call to BrowserThread::GetMessageLoopProxyForThread(
BrowserThread::UI)). ui/accelerated_widget_mac can't depend on content,
so it needs to be informed about the UI thread's task runner before any
renderers are created. Doing the initialization in
BrowserMainLoop::PreCreateThreads() ensures the initialization happens
before the IO thread is created.

BUG=531831
Committed: https://crrev.com/ef4c91b0545cd2190ef0650d03662e823d1cf6b0
Cr-Commit-Position: refs/heads/master@{#350449}

[tapted, 2015-09-17 08:48:25]

tapted@chromium.org changed reviewers:
+ ccameron@chromium.org

[tapted, 2015-09-17 08:53:19]

Hi Chris please take a look.

Follow-up to this (for MacViews) is http://crrev.com/1352743002

Once this one looks good, I'll loop in some owners for browser_main_loop and
thread_restrictions

Thanks!

[ccameron, 2015-09-17 18:58:48]

This lgtm

[tapted, 2015-09-17 22:56:32]

tapted@chromium.org changed reviewers:
+ jam@chromium.org

[tapted, 2015-09-17 22:59:30]

+jam please take a look for OWNERS in thread_restrictions (just a rename) and
content/browser/browser_main_loop.cc (I *think* this is the right place for it -
see CL description).

Thanks!

[jam, 2015-09-17 23:25:16]

https://codereview.chromium.org/1354533002/diff/1/content/browser/browser_mai...
File content/browser/browser_main_loop.cc (right):

https://codereview.chromium.org/1354533002/diff/1/content/browser/browser_mai...
content/browser/browser_main_loop.cc:706: // The WindowResizeHelper synchronizes
IO thread tasks with the UI thread.
can you describe this some more? are we making the UI thread wait on the IO, or
vice versa?

[tapted, 2015-09-18 00:36:55]

https://codereview.chromium.org/1354533002/diff/1/content/browser/browser_mai...
File content/browser/browser_main_loop.cc (right):

https://codereview.chromium.org/1354533002/diff/1/content/browser/browser_mai...
content/browser/browser_main_loop.cc:706: // The WindowResizeHelper synchronizes
IO thread tasks with the UI thread.
On 2015/09/17 23:25:16, jam wrote:
> can you describe this some more? are we making the UI thread wait on the IO,
or
> vice versa?

Updated the comment, now says

  // The WindowResizeHelper allows the UI thread to wait on specific renderer
  // and GPU messages from the IO thread. Initializing it before the IO thread
  // starts ensures the affected IO thread messages always have somewhere to go.

There's also a lot of specific details in the WindowResizeHelper class comment
in ui/accelerated_widget_mac/window_resize_helper_mac.h

[jam, 2015-09-21 16:49:15]

On 2015/09/18 00:36:55, tapted wrote:
>
https://codereview.chromium.org/1354533002/diff/1/content/browser/browser_mai...
> File content/browser/browser_main_loop.cc (right):
> 
>
https://codereview.chromium.org/1354533002/diff/1/content/browser/browser_mai...
> content/browser/browser_main_loop.cc:706: // The WindowResizeHelper
synchronizes
> IO thread tasks with the UI thread.
> On 2015/09/17 23:25:16, jam wrote:
> > can you describe this some more? are we making the UI thread wait on the IO,
> or
> > vice versa?
> 
> Updated the comment, now says
> 
>   // The WindowResizeHelper allows the UI thread to wait on specific renderer
>   // and GPU messages from the IO thread. Initializing it before the IO thread
>   // starts ensures the affected IO thread messages always have somewhere to
go.
> 
> There's also a lot of specific details in the WindowResizeHelper class comment
> in ui/accelerated_widget_mac/window_resize_helper_mac.h

I see. This is new to me, but since it was existing behavior I won't press
further.

However: the files in ui/ should remove all mentions of content & its classes.

[tapted, 2015-09-22 05:05:22]

Patchset #3 (id:40001) has been deleted

[tapted, 2015-09-22 05:06:55]

Patchset #3 (id:60001) has been deleted

[tapted, 2015-09-22 06:24:18]

On 2015/09/21 16:49:15, jam wrote:
> On 2015/09/18 00:36:55, tapted wrote:
> >
>
https://codereview.chromium.org/1354533002/diff/1/content/browser/browser_mai...
> > File content/browser/browser_main_loop.cc (right):
> > 
> >
>
https://codereview.chromium.org/1354533002/diff/1/content/browser/browser_mai...
> > content/browser/browser_main_loop.cc:706: // The WindowResizeHelper
> synchronizes
> > IO thread tasks with the UI thread.
> > On 2015/09/17 23:25:16, jam wrote:
> > > can you describe this some more? are we making the UI thread wait on the
IO,
> > or
> > > vice versa?
> > 
> > Updated the comment, now says
> > 
> >   // The WindowResizeHelper allows the UI thread to wait on specific
renderer
> >   // and GPU messages from the IO thread. Initializing it before the IO
thread
> >   // starts ensures the affected IO thread messages always have somewhere to
> go.
> > 
> > There's also a lot of specific details in the WindowResizeHelper class
comment
> > in ui/accelerated_widget_mac/window_resize_helper_mac.h
> 
> I see. This is new to me, but since it was existing behavior I won't press
> further.
> 
> However: the files in ui/ should remove all mentions of content & its classes.

Done.

[jam, 2015-09-23 18:14:39]

lgtm

[tapted, 2015-09-24 00:34:50]

The CQ bit was checked by tapted@chromium.org

[tapted, 2015-09-24 00:34:51]

The patchset sent to the CQ was uploaded after l-g-t-m from
ccameron@chromium.org
Link to the patchset: https://codereview.chromium.org/1354533002/#ps80001
(title: "rejig class comments")

[commit-bot: I haz the power, 2015-09-24 00:38:34]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1354533002/80001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1354533002/80001

[commit-bot: I haz the power, 2015-09-24 02:51:46]

Committed patchset #3 (id:80001)

[commit-bot: I haz the power, 2015-09-24 02:52:31]

Patchset 3 (id:??) landed as
https://crrev.com/ef4c91b0545cd2190ef0650d03662e823d1cf6b0
Cr-Commit-Position: refs/heads/master@{#350449}