Nice Catch, thanks.

The DCHECK failure is expected since when latching is enabled bubbling happens for the GSB event rather than GSU events. This means the SendGestureScrollEnd will get called with the GSB event.

I will update the DCHECK to check event types properly when latching is enabled.

I notice that the DCHECKs in SendGestureScrollBegin/End allow for GesturePinchBegin/End events but go on to access the scroll_update union member. We should be checking the event type to make sure we're using the right union member.

Regarding the issue of the scroll stopping, it looks like we're hitting https://cs.chromium.org/chromium/src/content/browser/renderer_host/render_widget_host_input_event_router.cc?rcl=a337dcd89c49d249cb19ec1e39036b68e1522a01&l=658 right away, and then all of the subsequent events get dropped here https://cs.chromium.org/chromium/src/content/browser/renderer_host/render_widget_host_input_event_router.cc?rcl=a337dcd89c49d249cb19ec1e39036b68e1522a01&l=649

It looks like the router is seeing the GSE created when it called SendGestureScrollEnd and cancels the scroll bubbling.

It looks like we just need to check |target_view == first_bubbling_scroll_target_.target| when we process the non-GSB events.

In particular, when |target_view != first_bubbling_scroll_target_.target| then the GSEs are from SendGestureScrollEnd. The non-latching code drops those here: https://cs.chromium.org/chromium/src/content/browser/renderer_host/render_widget_host_input_event_router.cc?rcl=a337dcd89c49d249cb19ec1e39036b68e1522a01&l=679

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/40324fd4616a8ee0d3d6c291c90a89a1b0621bd7

commit 40324fd4616a8ee0d3d6c291c90a89a1b0621bd7
Author: Sahel Sharify <sahel@chromium.org>
Date: Tue Sep 19 06:57:08 2017

SendGestureScrollEnd uses GSB data when wheel scroll latching is enabled.

The added test will fail with wheel scroll latching enabled and without
applying this cl.

Bug:  763422 
Test: SitePerProcessBrowserTest.ScrollBubblingFromOOPIFTest
Change-Id: I67ad586211f9d740b4652ed40ec0d260bd2c4abe
Reviewed-on: https://chromium-review.googlesource.com/658217
Reviewed-by: Avi Drissman <avi@chromium.org>
Reviewed-by: Timothy Dresser <tdresser@chromium.org>
Reviewed-by: Kevin McNee <mcnee@chromium.org>
Commit-Queue: Sahel Sharifymoghaddam <sahel@chromium.org>
Cr-Commit-Position: refs/heads/master@{#502793}
[modify] https://crrev.com/40324fd4616a8ee0d3d6c291c90a89a1b0621bd7/content/browser/renderer_host/render_widget_host_input_event_router.cc
[modify] https://crrev.com/40324fd4616a8ee0d3d6c291c90a89a1b0621bd7/content/browser/site_per_process_browsertest.cc

That CL fixes the DCHECK, but the RWHIER is still interrupting itself with its own GSE as described in #3 and #4.

I am sorry, I missed your previous comments, this cl should fix the issue:
https://chromium-review.googlesource.com/c/chromium/src/+/678536

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/c7a86bf619e8c1db94fbf722657a121a0d53a7b6

commit c7a86bf619e8c1db94fbf722657a121a0d53a7b6
Author: Sahel Sharify <sahel@chromium.org>
Date: Wed Sep 27 19:24:15 2017

Generated GSE events don't reset the bubbling target.

In case of nested OOPIFs the GSB event is bubbled one target at a time;
when GSB is bubbling up from an intermediate target, a synthetic GSE is
sent to the target to finish the scrolling sequence on it.
Once GSB has found its final target, the rest of the scroll events are
directly bubbled to that target rather than getting bubbled up one
target at a time.

https://chromium-review.googlesource.com/658217 cl has fixed the DCHECK
hit while sending a synthetic GSE to an intermediate target.
However, the generated GSE still stops scroll bubbling by resetting the
bubbling target to null. This cl drops generated GSEs that are sent to
intermediate bubbling targets and prevents them from stopping the
scroll bubbling.

Bug:  763422 
Change-Id: I6a49d431b6f838b3f56e477ef6b0af7b8e6bf0c2
Reviewed-on: https://chromium-review.googlesource.com/678536
Reviewed-by: Kevin McNee <mcnee@chromium.org>
Reviewed-by: Ken Buchanan <kenrb@chromium.org>
Reviewed-by: Avi Drissman <avi@chromium.org>
Commit-Queue: Sahel Sharifymoghaddam <sahel@chromium.org>
Cr-Commit-Position: refs/heads/master@{#504749}
[modify] https://crrev.com/c7a86bf619e8c1db94fbf722657a121a0d53a7b6/content/browser/renderer_host/render_widget_host_input_event_router.cc
[modify] https://crrev.com/c7a86bf619e8c1db94fbf722657a121a0d53a7b6/content/browser/site_per_process_browsertest.cc