Maintain key state in the plugin delegate if UIPI is enabled.

Maintain key state in the plugin delegate if UIPI is enabled.

UIPI blocks keystate across process boundaries in certain situations. To be safe we maintain our own key state and intercept calls to GetKeyState. 

BUG=86895
TEST=Adobe has test coverage.
Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=96069

[jschuh, 2011-08-09 03:43:46]

@cpu for code and @jam for owners (plus relevant plugin knowledge).

[jam, 2011-08-09 16:26:18]

http://codereview.chromium.org/7529041/diff/3005/webkit/plugins/npapi/webplug...
File webkit/plugins/npapi/webplugin_delegate_impl_win.cc (right):

http://codereview.chromium.org/7529041/diff/3005/webkit/plugins/npapi/webplug...
webkit/plugins/npapi/webplugin_delegate_impl_win.cc:95: size_t
g_saved_key_state[256 / sizeof(size_t)];
I'm not sure I follow why size_t is used here?

http://codereview.chromium.org/7529041/diff/3005/webkit/plugins/npapi/webplug...
webkit/plugins/npapi/webplugin_delegate_impl_win.cc:479: base::win::GetVersion()
>= base::win::VERSION_VISTA &&
nit: probably you want to do this OS check when you enable the quirk

[jschuh, 2011-08-09 16:48:04]

http://codereview.chromium.org/7529041/diff/3005/webkit/plugins/npapi/webplug...
File webkit/plugins/npapi/webplugin_delegate_impl_win.cc (right):

http://codereview.chromium.org/7529041/diff/3005/webkit/plugins/npapi/webplug...
webkit/plugins/npapi/webplugin_delegate_impl_win.cc:95: size_t
g_saved_key_state[256 / sizeof(size_t)];
On 2011/08/09 16:26:18, John Abd-El-Malek wrote:
> I'm not sure I follow why size_t is used here?

No major reason. I figured it made sense to use a machine-word type width. Is
there a better type to use?

http://codereview.chromium.org/7529041/diff/3005/webkit/plugins/npapi/webplug...
webkit/plugins/npapi/webplugin_delegate_impl_win.cc:479: base::win::GetVersion()
>= base::win::VERSION_VISTA &&
On 2011/08/09 16:26:18, John Abd-El-Malek wrote:
> nit: probably you want to do this OS check when you enable the quirk

Done.

[jschuh, 2011-08-09 18:06:45]

http://codereview.chromium.org/7529041/diff/3005/webkit/plugins/npapi/webplug...
File webkit/plugins/npapi/webplugin_delegate_impl_win.cc (right):

http://codereview.chromium.org/7529041/diff/3005/webkit/plugins/npapi/webplug...
webkit/plugins/npapi/webplugin_delegate_impl_win.cc:95: size_t
g_saved_key_state[256 / sizeof(size_t)];
On 2011/08/09 16:48:04, Justin Schuh wrote:
> On 2011/08/09 16:26:18, John Abd-El-Malek wrote:
> > I'm not sure I follow why size_t is used here?
> 
> No major reason. I figured it made sense to use a machine-word type width. Is
> there a better type to use?

Just talked to you but I figured I'd record for reference. I'd intended to write
this as a static bit-array, but I see that I forgot one important detail. I'll
have new patch up shortly.

[jschuh, 2011-08-09 19:24:24]

So, my bit array was using far more space than it should. This cleans it up and
adds some comments to make it clearer.

[jam, 2011-08-09 19:32:03]

lgtm

[cpu_(ooo_6.6-7.5), 2011-08-09 20:07:31]

lgtm

http://codereview.chromium.org/7529041/diff/8001/webkit/plugins/npapi/webplug...
File webkit/plugins/npapi/webplugin_delegate_impl_win.cc (right):

http://codereview.chromium.org/7529041/diff/8001/webkit/plugins/npapi/webplug...
webkit/plugins/npapi/webplugin_delegate_impl_win.cc:110: }
zeromemory is just a macro to memset

http://codereview.chromium.org/7529041/diff/8001/webkit/plugins/npapi/webplug...
webkit/plugins/npapi/webplugin_delegate_impl_win.cc:995:
SetSavedKeyState(wparam);
wparam % 256

Here or inside, your choice.

[jschuh, 2011-08-09 20:57:27]

Okay, changes made.

http://codereview.chromium.org/7529041/diff/8001/webkit/plugins/npapi/webplug...
File webkit/plugins/npapi/webplugin_delegate_impl_win.cc (right):

http://codereview.chromium.org/7529041/diff/8001/webkit/plugins/npapi/webplug...
webkit/plugins/npapi/webplugin_delegate_impl_win.cc:110: }
On 2011/08/09 20:07:31, cpu wrote:
> zeromemory is just a macro to memset

I know. It's just a habit I have, but I'm happy to change it.

http://codereview.chromium.org/7529041/diff/8001/webkit/plugins/npapi/webplug...
webkit/plugins/npapi/webplugin_delegate_impl_win.cc:995:
SetSavedKeyState(wparam);
On 2011/08/09 20:07:31, cpu wrote:
> wparam % 256
> 
> Here or inside, your choice.

I assumed we shouldn't ever see a value larger than 254. So, adding a CHECK
instead.