All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=778610

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=fe340e994ee63ea6c345fe336d65906f7596fa26cf698df632307fcd5aa1c396


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

android-nexus7v2

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8964677035433680768

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

Hi lgarron@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 : Lucas Garron
  Commit : 32d266526d8c245ed486d88ec870bb3964877616
  Date   : Wed Oct 18 17:19:18 2017
  Subject: Preload HSTS for youtube.com, including subdomains.

Bisect Details
  Configuration: android_nexus7_perf_bisect
  Benchmark    : thread_times.key_mobile_sites_smooth
  Metric       : thread_renderer_main_cpu_time_per_frame/http___cuteoverload.com
  Change       : 10.94% | 4.79669024842 -> 5.42297960417

Revision             Result                   N
chromium@509760      4.79669 +- 0.925513      9       good
chromium@509780      4.62562 +- 0.862683      9       good
chromium@509790      4.65737 +- 1.36705       9       good
chromium@509795      4.90668 +- 1.46004       14      good
chromium@509797      4.89409 +- 1.41326       14      good
chromium@509798      4.8701 +- 0.57589        6       good
chromium@509799      5.42298 +- 0.600284      9       bad       <--

To Run This Test
  src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=http...cuteoverload.com thread_times.key_mobile_sites_smooth

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8964677035433680768


For feedback, file a bug with component Speed>Bisection

Ooh, exciting.

Security for youtube.com is important, so this is probably WontFix.

I presume the test loads content from YouTube. Where is the content of the test located?

The content of the test is in cloud storage: you can re-run the test with the "To Run This Test" command above to run it and load the content from web-page-replay. +Ned for more info on getting the test content.

Lucas: can you explain the nature of this regression?

We are now forcing HTTPS for youtube.com
If any test was using HTTP, this will add an extra redirect.

Then I believe that there is bug, because the page being regressed here is cuteoverload (see thread_renderer_main_cpu_time_per_frame/http___cuteoverload.com)

Re: #8: I haven't checked the recorded version, but the live version of cuteoverload.com definitely has embedded youtube videos.

cuteoverload.com appears to be an HTTP site with embedded HTTP YouTube videos.

Ah interesting. The recorded version online contain HTTP, so this should mean that HTTPS connections are failing?

Assigning to elawrence who is the new owner of the HSTS list, but I suspect there's nothing we can do about this regression. That said, I'm surprised the internal redirect introduced by HSTS would cause a 10% regression?

I expect the CPU regression here comes from the use of HTTPS, not from the internal redirection itself. 

YouTube is now enforcing HTTPS, and CPU improvements in HTTPS performance are not within the scope of domain security policy. 

The test should probably be recorded with a new baseline.