Revert 185474 -- speculative. browser_tests stopped starting up on all

Revert 185474 -- speculative. browser_tests stopped starting up on all
xp bots. I'm guessing the GetProcAddressed() function might misbehave on XP.

> Fix touch scaling for high dpi windows.
> 
> Touch events in windows are always expressed in screen coordinates, even if
> the target app is not dpi-aware. This patch determines the dpi scale used
> by the OS, and re-scales touch events accordingly. 
> 
> In some cases the touch events are sent to the wrong window. This is detected
> and corrected.
> 
> BUG=175542
> 
> 
> Review URL: https://chromiumcodereview.appspot.com/12317157

TBR=girard@chromium.org
Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=185502

[Nico, 2013-03-01 09:30:00]

Committed patchset #1 manually as r185502.

[Nico, 2013-03-01 10:00:40]

The revert helped.

Before:
http://build.chromium.org/p/chromium.win/buildstatus?builder=XP%20Tests%20%28...
After:
http://build.chromium.org/p/chromium.win/buildstatus?builder=XP%20Tests%20%28...

(also on all the other xp dbg bots running browser_tests and
content_browsertests)

[girard, 2013-03-01 15:07:27]

On 2013/03/01 10:00:40, Nico wrote:
> The revert helped.
> 
> Before:
>
http://build.chromium.org/p/chromium.win/buildstatus?builder=XP%2520Tests%252...
> After:
>
http://build.chromium.org/p/chromium.win/buildstatus?builder=XP%2520Tests%252...
> 
> (also on all the other xp dbg bots running browser_tests and
> content_browsertests)

Thanks for applying the revert. 

I can't imagine how this code could affect browser_tests, since all of the
modified code paths are behind WM_TOUCH events, and we aren't directly linking
to any unavailable entry points. Inconceivable! (In other words, I'll test on XP
to find out what I've screwed up.)

[girard, 2013-03-04 18:51:57]

On 2013/03/01 15:07:27, girard wrote:
> On 2013/03/01 10:00:40, Nico wrote:
> > The revert helped.
> > 
> > Before:
> >
>
http://build.chromium.org/p/chromium.win/buildstatus?builder=XP%252520Tests%2...
> > After:
> >
>
http://build.chromium.org/p/chromium.win/buildstatus?builder=XP%252520Tests%2...
> > 
> > (also on all the other xp dbg bots running browser_tests and
> > content_browsertests)
> 
> Thanks for applying the revert. 
> 
> I can't imagine how this code could affect browser_tests, since all of the
> modified code paths are behind WM_TOUCH events, and we aren't directly linking
> to any unavailable entry points. Inconceivable! (In other words, I'll test on
XP
> to find out what I've screwed up.)

I've tested under XP, and the unit tests run correctly/consistently. I'm going
to assume the tests were flaky and try a re-submit.

[Nico, 2013-03-04 18:53:25]

On Mon, Mar 4, 2013 at 7:51 PM,  <girard@chromium.org> wrote:
> On 2013/03/01 15:07:27, girard wrote:
>>
>> On 2013/03/01 10:00:40, Nico wrote:
>> > The revert helped.
>> >
>> > Before:
>> >
>
>
>
http://build.chromium.org/p/chromium.win/buildstatus?builder=XP%252520Tests%2...
>>
>> > After:
>> >
>
>
>
http://build.chromium.org/p/chromium.win/buildstatus?builder=XP%252520Tests%2...
>
>> >
>> > (also on all the other xp dbg bots running browser_tests and
>> > content_browsertests)
>
>
>> Thanks for applying the revert.
>
>
>> I can't imagine how this code could affect browser_tests, since all of the
>> modified code paths are behind WM_TOUCH events, and we aren't directly
>> linking
>> to any unavailable entry points. Inconceivable! (In other words, I'll test
>> on
>
> XP
>>
>> to find out what I've screwed up.)
>
>
> I've tested under XP, and the unit tests run correctly/consistently. I'm
> going
> to assume the tests were flaky and try a re-submit.
>
> https://codereview.chromium.org/12379040/

They consistently failed on all 6 xp dbg bots on the main waterfall
the second this landed. I'd be surprised if that didn't happen again.
Feel free to reland, but please do watch the bots closely.