Making sure that base::MessagePumpForUI from different modules are isolated from each other.

Making sure that base::MessagePumpForUI from different modules are isolated from each other.

This is a minimal version of the original fix that was reverted by https://src.chromium.org/viewvc/chrome?view=rev&revision=137542. This version does not contain any protection against shatter attacks.

BUG=124091
Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=137765

[alexeypa (please no reviews), 2012-05-16 23:21:09]

Please take a look.

This is a minimal version of http://codereview.chromium.org/10134001/ that can
be merged along with r137547 to 1132 if we got OK from TPM. 

I didn't add CHECKs to all currently unchecked API calls to avoid making it
large. I do, however, CHECK all calls touched by this CL.

[jar (doing other things), 2012-05-17 00:57:24]

Some simplifying nits below.... 

then LGTM.

https://chromiumcodereview.appspot.com/10407011/diff/1/base/message_pump_win.cc
File base/message_pump_win.cc (right):

https://chromiumcodereview.appspot.com/10407011/diff/1/base/message_pump_win....
base/message_pump_win.cc:95: if (message_hwnd_ != NULL)
nit: you probably don't need this test.  The constructor does a CHECK in
InitMessageWnd()

https://chromiumcodereview.appspot.com/10407011/diff/1/base/message_pump_win....
base/message_pump_win.cc:98: if (atom_ != 0)
nit: The test is not needed since the constructors ensures a non-zero value. You
could DCHECK if you're paranoid, or even CHECK.

[jar (doing other things), 2012-05-17 02:48:09]

Additional major simplifications courtesy of suggestions by Ricardo.

https://chromiumcodereview.appspot.com/10407011/diff/1/base/message_pump_win.cc
File base/message_pump_win.cc (right):

https://chromiumcodereview.appspot.com/10407011/diff/1/base/message_pump_win....
base/message_pump_win.cc:247: instance_ =
base::GetModuleFromAddress(static_cast<void*>(wc.lpfnWndProc));
Talking to rvargas, he notes that a "minimal" change would be to only fix this
one line, and all would be well.

You don't need the printf, etc.  That would make this notably smaller.... you
don't need atom... since the class name is constant, etc.

[alexeypa (please no reviews), 2012-05-17 18:30:19]

PTAL.

https://chromiumcodereview.appspot.com/10407011/diff/1/base/message_pump_win.cc
File base/message_pump_win.cc (right):

https://chromiumcodereview.appspot.com/10407011/diff/1/base/message_pump_win....
base/message_pump_win.cc:247: instance_ =
base::GetModuleFromAddress(static_cast<void*>(wc.lpfnWndProc));
On 2012/05/17 02:48:09, jar wrote:
> Talking to rvargas, he notes that a "minimal" change would be to only fix this
> one line, and all would be well.
> 
> You don't need the printf, etc.  That would make this notably smaller.... you
> don't need atom... since the class name is constant, etc.

I guess you are right. Sure such a fix will not give us any additional checks
but we don't really have time, so that should be bake in trunk.

[jar (doing other things), 2012-05-17 21:36:24]

lgtm

[commit-bot: I haz the power, 2012-05-17 21:38:23]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/alexeypa@chromium.org/10407011/8001

[commit-bot: I haz the power, 2012-05-17 23:09:25]

Change committed as 137765