All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=850970

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=25b179d702e621daa5408adf9f05a93511d0c1e6126203cdd4e8233fddb122f6


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

android-nexus5
android-nexus5X
android-webview-nexus5X
android-webview-nexus6

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/11956509240000

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/1498a651240000

📍 Found a significant difference after 1 commit.
https://pinpoint-dot-chromeperf.appspot.com/job/11956509240000

Hide camera icon if missing camera permission in file picker. by tedchoc@google.com
https://chromium.googlesource.com/chromium/src/+/ff8ead91ee0187bbf6a79e01b0450279e6a21ef9
1.59 → 3.358 (+1.768)

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

📍 Found a significant difference after 1 commit.
https://pinpoint-dot-chromeperf.appspot.com/job/1498a651240000

Hide camera icon if missing camera permission in file picker. by tedchoc@google.com
https://chromium.googlesource.com/chromium/src/+/ff8ead91ee0187bbf6a79e01b0450279e6a21ef9
1.596 → 3.326 (+1.73)

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

Very doubtful that CL should be the cause.  That only affects the file picker, so unless that test is thrashing file selection, that won't affect anything in the normal rendering flow.

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/12be4521240000

I kicked off another bisect job.  Hopefully it will return the real problem.

cc-ing the benchmark owner in case we need some insight.

📍 Found a significant difference after 1 commit.
https://pinpoint-dot-chromeperf.appspot.com/job/12be4521240000

Revert "GPU: Remove transfer buffer autoflushing." by jdarpinian@chromium.org
https://chromium.googlesource.com/chromium/src/+/b63225bbef75dc91b78029ad8f8b1e516934b268
2.356 → 3.875 (+1.519)

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

Sorry, yes, the results in comments 4 and 5 were not correct, my fault.

Comment 10 should be right.

It's a revert, and naturally it regressed previous improvements in 9 of the 10 metrics.  The remaining one has seen a much bigger regression later on, so closing this bug.