All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=895391

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=5c115400757c159117f022f3a641121fd68bbaaf9aa67242a364cec8ed89a348


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

Android Nexus5X WebView Perf

blink_perf.paint - Benchmark documentation link:
  https://bit.ly/blink-perf-benchmarks

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/15206b2ee40000

https://chromium.googlesource.com/chromium/src/+/a2bc49cf7603a49aa935fc20b1f212ad9487269c is in the regression range, and it's plausible that "Recalc style from first to last sibling." could impact perf.

Nexus 5 is harsh. It catches all the regressions for some reason.

📍 Found significant differences after each of 3 commits.
https://pinpoint-dot-chromeperf.appspot.com/job/15206b2ee40000

Recalc style from first to last sibling. by futhark@chromium.org
https://chromium.googlesource.com/chromium/src/+/a2bc49cf7603a49aa935fc20b1f212ad9487269c
large-table-background-change-with-invisible-collapsed-borders: 3594 → 4450 (+856.1)

Automated Commit: LKGM 11143.0.0 for chromeos. by 3su6n15k.default@developer.gserviceaccount.com
https://chromium.googlesource.com/chromium/src/+/f3ad10fd9a92ae781897d9a60038a045b8be17eb
large-table-background-change-with-invisible-collapsed-borders: 4450 → 4407 (-43.05)

Add instrumentation to debug a crash in ThrottlingUrlLoader. by falken@chromium.org
https://chromium.googlesource.com/chromium/src/+/709b126d1e0ac09ff884ec5fea23bd25ccf2ce1b
large-table-background-change-with-invisible-collapsed-borders: 4429 → 4517 (+88.43)

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

Benchmark documentation link:
  https://bit.ly/blink-perf-benchmarks

The test generates a lot of ComputedStyle objects upfront, but only changes one per measured test afterwards. The allocation order of the ComputedStyles changed with the style recalc order change, so the memory access pattern on the ComputedStyles during the repaint could be the reason for the regression.

I don't think it's worth looking into.