Revert 207278 "Make sure that the UI window created by base::Mes..."

Revert 207278 "Make sure that the UI window created by base::Mes..."

Speculative revert: Suspected to break NavigationControllerTest.PurgeScreenshot.

> Make sure that the UI window created by base::MessagePumpForUI is destoyed on the same thread (Windows).
> 
> Currently the window created base::MessagePumpForUI can be destroyed on a wrong thread. base::MessagePumpForUI is a ref-counted class so it can (and does) outlive the owning base::MessageLoop. As the result DestroyWindow() can be called on a wrong thread. This makes TSAN unhappy and it reports races deep unside user32.dll.
> 
> Changes in this CL:
> - The message pump is now notified when the owning message loop is being destroyed. The notification is used to free all resources that hve to be released on the base::MessageLoop's thread.
> - MessagePumpForUI::ScheduleWork() synchronizes access to the message-only window handle to avoid posting messages to the window during or after its destruction.
> 
> BUG=241939
> 
> Review URL: https://chromiumcodereview.appspot.com/15709015

TBR=alexeypa@chromium.org

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=207327

[kinuko, 2013-06-20 06:35:29]

Committed patchset #1 manually as r207327.

[kinuko, 2013-06-20 06:37:13]

On 2013/06/20 06:35:29, kinuko wrote:
> Committed patchset #1 manually as r207327.

PurgeScreenshot has started failing after around this change. It's a speculative
revert, I'll revert this revert if it doesn't fix.

http://test-results.appspot.com/dashboards/flakiness_dashboard.html#group=%40...
http://build.chromium.org/p/chromium.win/builders/Win7%20Tests%20(3)/builds/2...

[alexeypa (please no reviews), 2013-06-20 06:37:41]

On 2013/06/20 06:35:29, kinuko wrote:
> Committed patchset #1 manually as r207327.

Could you please post a link to the log?

[kinuko, 2013-06-20 06:54:25]

On 2013/06/20 06:37:41, alexeypa wrote:
> On 2013/06/20 06:35:29, kinuko wrote:
> > Committed patchset #1 manually as r207327.
> 
> Could you please post a link to the log?

Please see links in my message #3.

(I'll also update this issue again after a few cycles)

[kinuko, 2013-06-20 09:37:56]

On 2013/06/20 06:54:25, kinuko wrote:
> On 2013/06/20 06:37:41, alexeypa wrote:
> > On 2013/06/20 06:35:29, kinuko wrote:
> > > Committed patchset #1 manually as r207327.
> > 
> > Could you please post a link to the log?
> 
> Please see links in my message #3.
> 
> (I'll also update this issue again after a few cycles)

It looks the revert has fixed the test on both Win7 bots.
It might be an error in the test (because that's the only test that was
failing), but it's great if you could take a look.

http://test-results.appspot.com/dashboards/flakiness_dashboard.html#group=%40...

Some more sample logs:
Before revert:
http://build.chromium.org/p/chromium.win/builders/Win%207%20Tests%20x64%20(3)...
After revert:
http://build.chromium.org/p/chromium.win/builders/Win%207%20Tests%20x64%20(3)...

[alexeypa (please no reviews), 2013-06-20 16:34:15]

On 2013/06/20 09:37:56, kinuko wrote:
> It looks the revert has fixed the test on both Win7 bots.

Thanks for looking into this! I'll investigate.