Khushal, could this be https://chromium.googlesource.com/chromium/src/+/8360fc950e05eda2a24da3cea3fbc3e199e20fd8 ?

Hmmm, could be. Let me see if I can get a repro on linux.

Looks like this is flaky on linux too: https://test-results.appspot.com/dashboards/flakiness_dashboard.html#testType=cc_unittests&tests=LayerTreeHostScrollTestImplSideInvalidation

This also exonerates your patch because it was flaky before your patch.

The change above made it so we could activate a main frame update before a previous impl-side pending tree is draw to avoid blocking the pipeline. The scroll test was depending on the previous pipelining.

https://chromium-review.googlesource.com/c/587871/ should fix this.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/5e40f323bc67b22cd168b14b019bd5dc1b48c973

commit 5e40f323bc67b22cd168b14b019bd5dc1b48c973
Author: Khushal <khushalsagar@chromium.org>
Date: Sat Jul 29 00:00:03 2017

cc: Unflake LayerTreeHostScrollTestImplSideInvalidation.

Don't use num of draws to end test. The test assumed to see 4 draws
to correspond with 4 pending trees. Since the active tree resulting
from an impl-side pending tree can be merged with a new pending tree,
the number of draws may not always be equal to the number of pending
trees.

Bug:  749218 
Cq-Include-Trybots: master.tryserver.blink:linux_trusty_blink_rel
Change-Id: Ib000838f07b0b00537d037402122539ccb449985
Reviewed-on: https://chromium-review.googlesource.com/587871
Reviewed-by: enne <enne@chromium.org>
Commit-Queue: Khushal <khushalsagar@chromium.org>
Cr-Commit-Position: refs/heads/master@{#490547}
[modify] https://crrev.com/5e40f323bc67b22cd168b14b019bd5dc1b48c973/cc/trees/layer_tree_host_unittest_scroll.cc