Force MimeHandlerView to always use BrowserPlugin

content/browser/frame_host/render_widget_host_view_guest.cc

Index: content/browser/frame_host/render_widget_host_view_guest.cc
diff --git a/content/browser/frame_host/render_widget_host_view_guest.cc b/content/browser/frame_host/render_widget_host_view_guest.cc
index d433e64c164122f7deae254bb6a39da4f45014db..634fa9af97fce484a86a6ae081d189667cd7347b 100644
--- a/content/browser/frame_host/render_widget_host_view_guest.cc
+++ b/content/browser/frame_host/render_widget_host_view_guest.cc
@@ -353,9 +353,13 @@ gfx::NativeView RenderWidgetHostViewGuest::GetNativeView() const {
   if (!guest_)
     return gfx::NativeView();
 
-  RenderWidgetHostView* rwhv = guest_->GetOwnerRenderWidgetHostView();
+  if (guest_->IsMimeHandlerViewGuest())
+    return guest_->GetTopLevelRenderWidgetHostView()->GetNativeView();

[lfg, 2016/07/20 19:49:57]

What's the problem when returning the OwnerRenderWidgetHostView() ?

[EhsanK, 2016/07/20 20:49:19]

When we show a PDF inside an OOPIF-<webview>, the owner's RWHV is a
RWHVChildFrame. But RWHVCF::GetNativeView() is NOTIMPLEMENTED().

Now my question is, what stops us from using top-level in all scenarios?  i.e.,
just return GetTopLevelRenderWidgetHost()->GetNativeView() regardless of whether
this is a mime handler or not.

[lfg, 2016/07/20 21:05:34]

In that case, shouldn't IsMimeHandlerViewGuest() return false?
Right, the other question I have is why are we calling GetNativeView() on the
guest? I'd rather try to clean this up before adding
GetTopLevelRenderWidgetHostView() to RenderWidgetHostViewGuest.

[EhsanK, 2016/07/21 19:53:06]

This is called from tests as well as used in sending screen rects. That being
said, whether or not we keep this in, wjmclean@ is about to add something like
GetTopLevelRWHV() to RWHVCF. I can then rebase after his CL lands.

+
+  RenderWidgetHostView* rwhv = GetOwnerRenderWidgetHostView();
   if (!rwhv)
     return gfx::NativeView();
+
   return rwhv->GetNativeView();
 }