All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=823942

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=c3f50c24b4d9007bc0aa1d6f65887fc41292ed8ce257a81fc2a29ab9f1810467


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

android-webview-nexus6

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/13178059440000

😿 Pinpoint job stopped with an error.
https://pinpoint-dot-chromeperf.appspot.com/job/13178059440000

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/1708b079440000

Cannot triage further without a working bisect.

😿 Pinpoint job stopped with an error.
https://pinpoint-dot-chromeperf.appspot.com/job/1708b079440000

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/154a93a7440000

📍 Found significant differences after each of 2 commits.
https://pinpoint-dot-chromeperf.appspot.com/job/154a93a7440000

WebUI cr-toast: clear timeout before showing by aee@chromium.org
https://chromium.googlesource.com/chromium/src/+/0c1d42b7de12ced90144d045742412b4017ea890

Disable app shell installer when use_sysroot=false by thomasanderson@chromium.org
https://chromium.googlesource.com/chromium/src/+/01ab4cc587e13c2f2cf944e100ca101191b8c1ae

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

"Disable app shell installer when use_sysroot=false" should be unrelated to this regression.

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/15cf404bc40000

😿 Pinpoint job stopped with an error.
https://pinpoint-dot-chromeperf.appspot.com/job/15cf404bc40000

'ReadHistogramsJsonValue' object has no attribute '_results_filename'

aee: PTAL. Looks like this should have been assigned to you (back in April!), but it fell off of everyone's radar. Looking at the graph, it seems the perf regression caused by the following CL still exists:

WebUI cr-toast: clear timeout before showing by aee@chromium.org
https://chromium.googlesource.com/chromium/src/+/0c1d42b7de12ced90144d045742412b4017ea890

@miu: The CL updates a UI component using by chrome:// pages. How can this be regressing blink performance? Are the blink_perf.bindings tests operate on chrome:// pages? If not, I think the bisection is not correct.

Also note that this particular code does not run on Android at all, which is referenced by comment #1.

I'm removing myself as owner. I removed myself from cc back in May since this seemed unrelated to the cr-toast CL.

Long-term, seems things are noisy and mostly recovered.