All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=838463

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=52582c47c4d7d48424d2a17cb96b41d9a11d0bd024d759595e63bfaee701468f


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

chromium-rel-win7-dual

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/1484343dc40000

📍 Found significant differences after each of 3 commits.
https://pinpoint-dot-chromeperf.appspot.com/job/1484343dc40000

[signin] Add metrics for refresh token revocation. by droger@chromium.org
https://chromium.googlesource.com/chromium/src/+/c6bf01d352c240cdda3c0e9c7595db214a6eeb74

Update FileManagerBrowserTestBase to avoid unprocessed messages by noel@chromium.org
https://chromium.googlesource.com/chromium/src/+/a59359507e747e61c620610cd4c9fb110ecba5ac

Import wpt@cc364d6f98b111e488aa1d88064af21566ec4184 by blink-w3c-test-autoroller@chromium.org
https://chromium.googlesource.com/chromium/src/+/ec2d4ae90a08738eb616bb2ab1c0fee5a551e0d6

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

Based on Pinpoint data, it looks like the first and second commits caused the regression. Since the first one regressed more, let's start investigating that.

I don't believe my CL could have performance impact: it only adds histograms which are rarely called. These are called on browser signout operations, when a OAuth2 token is revoked. This should not affect the renderer, and is probably not even happening in blink tests at all.

Looks like the graphs recovered.