All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=814251

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=48d57a0ec1b0f89365b51208b909b79d5e75d99c49431ed952e537dba611460e


Bot(s) for this bug's original alert(s):

android-nexus7v2

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/118cea8f840000

📍 Couldn't reproduce a difference.
https://pinpoint-dot-chromeperf.appspot.com/job/118cea8f840000

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/15f24a8f840000

📍 Found significant differences after each of 4 commits.
https://pinpoint-dot-chromeperf.appspot.com/job/15f24a8f840000

Revert "Enable root layer scrolling." by skobes@chromium.org
https://chromium.googlesource.com/chromium/src/+/066296a5708f57ba89b68605ae8e65292eb77d4d

[test_env.py] Warm up vpython virtualenv cache on swarming task shards. by iannucci@chromium.org
https://chromium.googlesource.com/chromium/src/+/8348d507e280f4bd02a7737bf5577f838d64437e

Revert "[test_env.py] Warm up vpython virtualenv cache on swarming task shards." by kbr@chromium.org
https://chromium.googlesource.com/chromium/src/+/7dde857919af2f59fcab10264d696301a15c64da

[RLS] Skip compositing input step if a scroll can be applied directly. by chrishtr@chromium.org
https://chromium.googlesource.com/chromium/src/+/32419f4422bf130d9c52331a0c7dafba9337cc99

Understanding performance regressions:
  http://g.co/ChromePerformanceRegressions

This is an RLS regression, not due to my patch.

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/12807470440000

Graphs seem to be recovering, but noisy.

I think this regression has to do with InvalidatePaintUsingContainer somehow
doing more work in the presence of RLS. Pre-RLS, SetBackingNeedsPaintInvalidationInRect
would always follow the conditional at the bottom. Now, it's one of three conditionals
in the middle, depending.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/1768c76d47b9d2058f18dcf37e9a4097be43417c

commit 1768c76d47b9d2058f18dcf37e9a4097be43417c
Author: Chris Harrelson <chrishtr@chromium.org>
Date: Sat Mar 10 03:00:55 2018

[PE] Optimize InvalidatePaintUsingContainer.

If a paint invalidation container returns true from IsPaintInvalidationContainer(),
that means it is composited, which in turn must mean the containing LayoutView is composited.

This allows us to simplify the code and avoid a call to View(), which may be expensive
because it has to first dereference the document.

Bug:  814251 
Cq-Include-Trybots: master.tryserver.blink:linux_trusty_blink_rel;master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2
Change-Id: I79cc37ee79271c2030a9f5c832ba5640ca8445c9
Reviewed-on: https://chromium-review.googlesource.com/957962
Reviewed-by: Philip Rogers <pdr@chromium.org>
Commit-Queue: Chris Harrelson <chrishtr@chromium.org>
Cr-Commit-Position: refs/heads/master@{#542333}
[modify] https://crrev.com/1768c76d47b9d2058f18dcf37e9a4097be43417c/third_party/WebKit/Source/core/paint/ObjectPaintInvalidator.cpp

This was a progression in the first place, I had read the graphs backwards.