All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=847379

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=e0571453c4395e608c107cc163412f767977b1b03e144bdf40e92687cde4a269


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

android-nexus5X

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/16e7bf26240000

📍 Found significant differences after each of 2 commits.
https://pinpoint-dot-chromeperf.appspot.com/job/16e7bf26240000

Fix regression in thread_times.key_mobile_sites_smooth by yiyix@chromium.org
https://chromium.googlesource.com/chromium/src/+/be2578711ee47f8b8a5ba8b15108dedb50a27367

Headless: allow creating HeadlessDevToolsClient for given external connection. by pfeldman@chromium.org
https://chromium.googlesource.com/chromium/src/+/f169143da1c6baea333f0bc6119bc031368f51a6

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

I don't think headless could affect chrome performance, especially on Android.

https://chromium-review.googlesource.com/1067147 might be relevant

the improvement was due to this cl, https://chromium-review.googlesource.com/c/chromium/src/+/1029733 and it has cause more 20 regression in other metrics. the cl mentioned here, has fixed those regressions and  smoothness.top_25_smooth timing is back to the point before the fake improvement. No bug.