All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=872126

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=d3eedc6a065bf55f16df7ecc19c1d5892ac93c3c634566eaa0a299f72387c5bf


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

mac-10_12_laptop_low_end-perf

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/12ec8ec8640000

📍 Found significant differences after each of 3 commits.
https://pinpoint-dot-chromeperf.appspot.com/job/12ec8ec8640000

Move associated interface helper impls to Blink by rockot@chromium.org
https://chromium.googlesource.com/chromium/src/+/26e4e80de6ec7fa293f67a20759a8606b24919db
1.086 → 1.081 (-0.004776)

Roll src/third_party/skia 0bf5408286ea..74c6ed3d1f17 (1 commits) by skia-chromium-autoroll@skia-buildbots.google.com.iam.gserviceaccount.com
https://chromium.googlesource.com/chromium/src/+/148fcd5f84f04f473fba5ab7ae97ee32c8f52537
1.081 → 1.095 (+0.0137)

[Display Cutout] Pull out WebContentsObserver by beccahughes@chromium.org
https://chromium.googlesource.com/chromium/src/+/d11d650ce098aa271d168b5de822042cb7abea87
1.104 → 1.098 (-0.00615)

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

for the benchmark, lower is better. so, the first and the third significant differences should not matter.

skia roll is the real culprit.

let me run another bisect with a narrow range so that it can dig the roll to find a suspected cl in the submodule.

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/14a9e51c640000

📍 Found a significant difference after 1 commit.
https://pinpoint-dot-chromeperf.appspot.com/job/14a9e51c640000

Roll src/third_party/skia 0bf5408286ea..74c6ed3d1f17 (1 commits) by skia-chromium-autoroll@skia-buildbots.google.com.iam.gserviceaccount.com
https://chromium.googlesource.com/chromium/src/+/148fcd5f84f04f473fba5ab7ae97ee32c8f52537
1.079 → 1.098 (+0.0181)

Assigning to sheriff scroggo@chromium.org because "Roll src/third_party/skia 0bf5408286ea..74c6ed3d1f17 (1 commits)" is a roll.

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

Removing myself from CC as my CL is limited to Android and this regression is on Mac.

+Speed>Bisection
Bisect couldn't dig up the third-party module roll. Is this expected?

The only CL in this roll is herb@'s https://skia-review.googlesource.com/c/skia/+/144302

Oops, Herb is out this week. Jim, can you or Brian take a look?

I can take a look, but I'm leaving for PTO in a few hours myself. My concern is that this will not be easy to revert as other CLs landed after it. I also seem to recall that a small perf regression was expected, which would be addressed in later CLs.

Yeah, there's not much I can do that wouldn't be more risky in other ways. Brian's out as well, so reassigning back to Herb who can look at it Monday.