[UseZoomForDSF] Guest view support

content/browser/frame_host/render_widget_host_view_child_frame.cc

Index: content/browser/frame_host/render_widget_host_view_child_frame.cc
diff --git a/content/browser/frame_host/render_widget_host_view_child_frame.cc b/content/browser/frame_host/render_widget_host_view_child_frame.cc
index be109468cc4034d53d1e0e6df900c7172f131ec8..ee607581787f53862a26bb37a75765aafb4149c7 100644
--- a/content/browser/frame_host/render_widget_host_view_child_frame.cc
+++ b/content/browser/frame_host/render_widget_host_view_child_frame.cc
@@ -48,7 +48,8 @@ RenderWidgetHostViewChildFrame::RenderWidgetHostViewChildFrame(
           GetSurfaceIdNamespace(), this);
     }
   }
-
+  // Input for child frames are already scaled.
+  host_->set_scale_input_to_viewport(false);

[kenrb, 2016/01/15 20:09:12]

I think that comment might not be generally true. If you are only testing with
BrowserPlugin, then I suspect it is true for mouse events but not touch events
(wjmaclean might be able to speak more definitively on that). For out-of-process
iframes, I don't think any of the input will have been scaled already, because
mouse and mousewheel events do not pass through the top-level RenderWidgetHost
anymore.

[oshima, 2016/01/15 22:24:10]

Thanks for heads up. That's actually good news because it simplifies this logic.
I added TODO for OOPIF support. Touch event seems to be working fine though.

[kenrb, 2016/01/20 21:43:10]

For most of this code I would defer to wjmaclean, but will this break input
events on OOPIFs? I would strongly prefer we not do that.

   host_->SetView(this);
 }