All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=780190

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=63f4909d534b1e5ee20fa63be5f2dc83bf11a8d6b4d73e3698c42809689fe4e6


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

chromium-rel-win8-dual

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8964186423015490416

=== Auto-CCing suspected CL author yutak@chromium.org ===

Hi yutak@chromium.org, the bisect results pointed to your CL, please take a look at the
results.


=== BISECT JOB RESULTS ===
Perf regression found with culprit

Suspected Commit
  Author : Yuta Kitamura
  Commit : f8350aeae266267a394fd462fe83131ec2818b7b
  Date   : Tue Oct 24 05:22:13 2017
  Subject: Revert "Ensure scrollbars are laid out before hit testing"

Bisect Details
  Configuration: win_8_perf_bisect
  Benchmark    : rasterize_and_record_micro.top_25
  Metric       : record_time/file___static_top_25_google.html
  Change       : 3.72% | 0.0985 -> 0.0948333333333

Revision             Result                       N
chromium@511042      0.0985 +- 0.00273861         6      good
chromium@511043      0.0956667 +- 0.0023094       6      bad       <--
chromium@511044      0.0958333 +- 0.00168325      6      bad
chromium@511045      0.0958889 +- 0.00329983      9      bad
chromium@511047      0.0953333 +- 0.0011547       6      bad
chromium@511051      0.0963333 +- 0.002           9      bad
chromium@511060      0.0956667 +- 0.00182574      6      bad
chromium@511078      0.0948333 +- 0.00168325      6      bad

To Run This Test
  src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=file...static.top.25.google.html rasterize_and_record_micro.top_25

More information on addressing performance regressions:
  http://g.co/ChromePerformanceRegressions

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8964186423015490416


For feedback, file a bug with component Speed>Bisection

This is a revert of pdr's change.

This is interesting.

My patch likely did affect performance but was actually rolled back in quickly (only change was an android test suppression) and the graph did not drop back down. Here's the timeline:
first land - r510961 - https://crrev.com/9960cbaff4e136069ce2c3026d747b7679c8ef29 - Mon Oct 23 23:45:42 2017
revert - r511043 - https://crrev.com/f8350aeae266267a394fd462fe83131ec2818b7b - Tue Oct 24 05:22:13 2017
reland - r511279 - https://crrev.com/4fc60db274fcd51069d5830b112c3bc76e4234e4 - Tue Oct 24 22:17:26 2017

Here are some more graphs of the same regression:
https://chromeperf.appspot.com/report?sid=8d6616e5e5ec879ccbb8be010d2a17b4b41b1a537bc1579491b62ce7c14342d2

I'm going to try a perf bisect on chromium-rel-win7-x64-dual to see if we can get the underlying regression.

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8964110399131108016

=== BISECT JOB RESULTS ===
NO Perf regression found

Bisect Details
  Configuration: win_x64_perf_bisect
  Benchmark    : rasterize_and_record_micro.top_25
  Metric       : record_time/file___static_top_25_google.html

Revision             Result                       N
chromium@511175      0.086619 +- 0.00607885       21      good
chromium@511272      0.0875238 +- 0.00460848      21      bad

To Run This Test
  src/tools/perf/run_benchmark -v --browser=release_x64 --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=file...static.top.25.google.html rasterize_and_record_micro.top_25

More information on addressing performance regressions:
  http://g.co/ChromePerformanceRegressions

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8964110399131108016


For feedback, file a bug with component Speed>Bisection

Bisect didn't turn up anything :/

I'm not sure what to do because there's a real regression but no obvious regressing patch. Brian, do you have any suggestions? If not, we can probably close as WONTFIX because the regression is fairly small overall.

WontFix-ing per #8