I suspect we want to continue using RenderWidgetHostInputEventRouter for mus, but we shouldn't need the hit testing part, just the bubbling/dispatching. This will largely depend upon where Ken goes with 750755.

Ken, is this work you plan on doing?

I have a WIP CL to route hit test regions to the CompositorFrameSinkSupport through the browser process' content layer, and the tentative plan is to have RenderWidgetHostInputEventRouter own a viz::HitTestQuery which it will use to obtain the FrameSinkId for the target RenderWidgetHostView.

I can own this work.

Thanks Ken!
I suspect you're going to run into a problem mapping FrameSinkIds to windows in --mash (and possibly --mus). I'm happy to help out here once you get far enough along.

This is no longer a blocker for 731255 as we're going with the browser assigning all FrameSinkIds, and we're going with the classic path for registering RenderWidgetHostInputEventRouter, which means it all just work for --mus now. That isn't the case for --mus=viz (or --mash) though.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/f1677be12bc5bd2f1445fb3fd82ce3983b677cf9

commit f1677be12bc5bd2f1445fb3fd82ce3983b677cf9
Author: Ken Buchanan <kenrb@chromium.org>
Date: Tue Dec 19 22:39:26 2017

Allow RenderWidgetHostInputEventRouter to use viz hit testing

Tentative CL for using the alternative browser process hit testing path
when running with --enable-viz.

Bug: 780279
Change-Id: I314808fd8f8264168aff3ee128d7782d37525e9d
Reviewed-on: https://chromium-review.googlesource.com/817621
Commit-Queue: Ken Buchanan <kenrb@chromium.org>
Reviewed-by: kylechar <kylechar@chromium.org>
Reviewed-by: Ria Jiang <riajiang@chromium.org>
Cr-Commit-Position: refs/heads/master@{#525165}
[modify] https://crrev.com/f1677be12bc5bd2f1445fb3fd82ce3983b677cf9/content/browser/renderer_host/browser_compositor_view_mac.h
[modify] https://crrev.com/f1677be12bc5bd2f1445fb3fd82ce3983b677cf9/content/browser/renderer_host/browser_compositor_view_mac.mm
[modify] https://crrev.com/f1677be12bc5bd2f1445fb3fd82ce3983b677cf9/content/browser/renderer_host/render_widget_host_input_event_router.cc
[modify] https://crrev.com/f1677be12bc5bd2f1445fb3fd82ce3983b677cf9/content/browser/renderer_host/render_widget_host_input_event_router.h
[modify] https://crrev.com/f1677be12bc5bd2f1445fb3fd82ce3983b677cf9/content/browser/renderer_host/render_widget_host_view_aura.cc
[modify] https://crrev.com/f1677be12bc5bd2f1445fb3fd82ce3983b677cf9/content/browser/renderer_host/render_widget_host_view_aura.h
[modify] https://crrev.com/f1677be12bc5bd2f1445fb3fd82ce3983b677cf9/content/browser/renderer_host/render_widget_host_view_base.cc
[modify] https://crrev.com/f1677be12bc5bd2f1445fb3fd82ce3983b677cf9/content/browser/renderer_host/render_widget_host_view_base.h
[modify] https://crrev.com/f1677be12bc5bd2f1445fb3fd82ce3983b677cf9/content/browser/renderer_host/render_widget_host_view_mac.h
[modify] https://crrev.com/f1677be12bc5bd2f1445fb3fd82ce3983b677cf9/content/browser/renderer_host/render_widget_host_view_mac.mm

The integration of this and the blocking bugs should be sufficient to use Viz hit testing for M66/M67, in order to get rid of the current performance penalty of 'slow path' hit testing. That would seem to mean this is an appropriate tracking bug for that.

I thought this is specific to mus/mash? We shouldn't need this for M66/67, right? In order to perform better compared to the current scheme, we'll need this 732392 version.

You might be right. I was thinking it was going to be similar work on the content side, but there is more on this that what we need, such as switching to real FrameSinkIds. I'll file a separate bug for that then.