Fix race in BrowserAccessibilityManagerWin creation.

Fix race in BrowserAccessibilityManagerWin creation.

Previously, BrowserAccessibilityManagerWin needed its parent
HWND and parent IAccessible upon construction, and if RWHVA
didn't have those, it didn't create the BAMW. This resulted
in at least two races:

* If an accessibility tree update came from the renderer
but the BAMW wasn't created because the window wasn't available,
the tree update would get discarded - possibly leading to an
assertion failure / renderer crash later on the next update
from being out of sync.

* If we tried to create the BAMW before the LegacyRWHH existed,
we would fall through and create a BAM instead of a BAMW,
making that whole tab inaccessible.

To fix both issues, we always create a BAMW when needed, and
we update the BAMW with the LegacyRWHH as soon as it's
available.

BUG=372478, 379019
NOTRY=true

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=276950

[jam, 2014-06-11 16:02:35]

lgtm for files outside of accessibility directories

[ananta, 2014-06-12 18:30:17]

https://codereview.chromium.org/329053004/diff/1/content/browser/renderer_hos...
File content/browser/renderer_host/render_widget_host_view_aura.cc (right):

https://codereview.chromium.org/329053004/diff/1/content/browser/renderer_hos...
content/browser/renderer_host/render_widget_host_view_aura.cc:1024:
legacy_render_widget_host_HWND_->set_browser_accessibility_manager(
This will crash in Windows 8 ASH. You may want to move this code below under the
if (legacy_render_widget_host_HWND_) block.

[dmazzoni, 2014-06-12 22:01:00]

Please take a look again. I want to merge to M36 to fix a top crasher, so I
rewrote it as minimally as possible. I've tested it and confirmed it also solves
the problem.

I still may want to do the refactoring from the first patch set in the future,
but this is a better patch to merge.

[ananta, 2014-06-12 22:17:45]

https://codereview.chromium.org/329053004/diff/20001/content/browser/renderer...
File content/browser/renderer_host/render_widget_host_view_aura.cc (right):

https://codereview.chromium.org/329053004/diff/20001/content/browser/renderer...
content/browser/renderer_host/render_widget_host_view_aura.cc:1263:
LegacyRenderWidgetHostHWND* parent_hwnd =
We should be allocating the BrowserAccessibilityManagerWin instance only if the 
LegacyRenderWidgetHostHWND instance is valid?

[jam, 2014-06-12 23:07:00]

lgtm

[dmazzoni, 2014-06-12 23:23:28]

https://codereview.chromium.org/329053004/diff/20001/content/browser/renderer...
File content/browser/renderer_host/render_widget_host_view_aura.cc (right):

https://codereview.chromium.org/329053004/diff/20001/content/browser/renderer...
content/browser/renderer_host/render_widget_host_view_aura.cc:1263:
LegacyRenderWidgetHostHWND* parent_hwnd =
On 2014/06/12 22:17:45, ananta wrote:
> We should be allocating the BrowserAccessibilityManagerWin instance only if
the 
> LegacyRenderWidgetHostHWND instance is valid?

No, this is intentional. We want to create the BAMW every time. The rest of this
patch makes it so that BAMW will handle receiving a NULL LRWHH and so that the
LRWHH will be passed to the BAMW if it's created second.

[ananta, 2014-06-12 23:38:22]

On 2014/06/12 23:23:28, dmazzoni wrote:
>
https://codereview.chromium.org/329053004/diff/20001/content/browser/renderer...
> File content/browser/renderer_host/render_widget_host_view_aura.cc (right):
> 
>
https://codereview.chromium.org/329053004/diff/20001/content/browser/renderer...
> content/browser/renderer_host/render_widget_host_view_aura.cc:1263:
> LegacyRenderWidgetHostHWND* parent_hwnd =
> On 2014/06/12 22:17:45, ananta wrote:
> > We should be allocating the BrowserAccessibilityManagerWin instance only if
> the 
> > LegacyRenderWidgetHostHWND instance is valid?
> 
> No, this is intentional. We want to create the BAMW every time. The rest of
this
> patch makes it so that BAMW will handle receiving a NULL LRWHH and so that the
> LRWHH will be passed to the BAMW if it's created second.

lgtm

[dmazzoni, 2014-06-12 23:44:37]

The CQ bit was checked by dmazzoni@chromium.org

[commit-bot: I haz the power, 2014-06-12 23:48:41]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/dmazzoni@chromium.org/329053004/40001

[dmazzoni, 2014-06-13 05:24:08]

The CQ bit was unchecked by dmazzoni@chromium.org

[dmazzoni, 2014-06-13 05:24:23]

The CQ bit was checked by dmazzoni@chromium.org

[commit-bot: I haz the power, 2014-06-13 05:26:19]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/dmazzoni@chromium.org/329053004/40001

[commit-bot: I haz the power, 2014-06-13 06:57:59]

Change committed as 276950