All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=833754

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=0578f23471fcd58d374db0eb467942342fce84a4f3d2886ee8cc79c16a3c35b9


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

android-webview-nexus5X

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/14f8d94ec40000

📍 Found significant differences after each of 4 commits.
https://pinpoint-dot-chromeperf.appspot.com/job/14f8d94ec40000

Roll AFDO from 67.0.3395.0_rc-r1 to 67.0.3396.0_rc-r1 by afdo-chromium-autoroll@skia-buildbots.google.com.iam.gserviceaccount.com
https://chromium.googlesource.com/chromium/src/+/443f4880b248dd6ade32463d8e383e3c01bba289

Re-evaluate media queries for initial style changes. by futhark@chromium.org
https://chromium.googlesource.com/chromium/src/+/7d899e4f29364d1337e51aa9bb133dd25a2fc389

[signin] Revert Dice default testing config to Milestone 1 by droger@chromium.org
https://chromium.googlesource.com/chromium/src/+/8cb55171dddde570257127a055132b9ad8fed284

Roll clang 328716:329921. by hans@chromium.org
https://chromium.googlesource.com/chromium/src/+/1f109d9c516640dda836493b24fd033da64411ff

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

fyi, 'Roll clang 328716:329921' was filed separately as 833273

tentatively assigned tofuthark for 'Re-evaluate media queries for initial style changes'

It's most likely https://crrev.com/4ef836ad2e67ce25e263514bcda94f808bfb02d6

That is the range of a drop on the linux perf bot as well:

https://chromeperf.appspot.com/report?sid=ab5de1d8ea2e0fe720ef83ebdb29732e430f5d86f16a7ed821b2ca4fa9179bcf&start_rev=546728&end_rev=558256

No. Ran the test through callgrind on master and > 90% of the whole run is spent in wtf string handling. Close to no time is used in textarea style handling.

dtapuska@ I don't know if https://chromium.googlesource.com/chromium/src/+/c59618d0f92b57e4dcfb903f3c99bb0574eac340 could be causing it, but it's the only wtf change in the regression range for the linux bot.

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/12daa991640000

I don't think it is my change since it is an elimination of a compile time macro for one that is exactly the same. Since nobody has run a bisect on the linux bot. I've started that.

😿 Pinpoint job stopped with an error.
https://pinpoint-dot-chromeperf.appspot.com/job/12daa991640000

Buildbucket says the build completed successfully, but Pinpoint can't find the isolate hash.

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/12e1d6ce640000

😿 Pinpoint job stopped with an error.
https://pinpoint-dot-chromeperf.appspot.com/job/12e1d6ce640000

Buildbucket says the build completed successfully, but Pinpoint can't find the isolate hash.

dtu@ any idea why my pinpoint jobs fail?

Re: c12: a lot of changed around late April with the perf builders and test harness, so Pinpoint jobs may not work on revision ranges from back then (without a bunch of custom config).

Looking at the graph from the original job, there's a clear large jump at the Clang roll, so I'm duping onto  issue 833273 .