All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=764344

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=cc0b8d0b4dddeee01e4d0f0f89e072380f87a2ea5c4c56f313c1486951ca2cc5


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

chromium-rel-win7-gpu-ati

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8968648614395850992

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

Hi elawrence@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 : Eric Lawrence
  Commit : 7dbeb2d882cbc41dc9c66021c4ac335f033e4322
  Date   : Wed Sep 06 20:58:40 2017
  Subject: Add field trial config for HTTPBad Phase 2

Bisect Details
  Configuration: winx64ati_perf_bisect
  Benchmark    : loading.desktop
  Metric       : timeToFirstContentfulPaint_avg/pcv1-cold/IndiaTimes
  Change       : 9.29% | 1230.46561905 -> 1344.71728572

Revision             Result                  N
chromium@500026      1230.47 +- 782.768      21      good
chromium@500073      1197.89 +- 64.9652      9       good
chromium@500079      1199.43 +- 201.545      9       good
chromium@500082      1207.24 +- 521.156      14      good
chromium@500083      1143.88 +- 54.5512      6       good
chromium@500084      1332.42 +- 207.481      9       bad       <--
chromium@500085      1370.15 +- 716.64       14      bad
chromium@500096      1339.31 +- 365.266      9       bad
chromium@500119      1344.72 +- 752.401      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=IndiaTimes loading.desktop

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8968648614395850992


For feedback, file a bug with component Speed>Bisection

Interesting. http://www.indiatimes.com triggers the HTTPBad Phase 2 editing notification on page load.

It uses Modernizr on page load to find out what features the client supports; one of those things involves editing an input control in such a way that it fires the DidChangeValueInTextField event.

http://webdbg.com/test/forms/modernizr.html is the minimal repro, showing "Not secure" even without the user interacting with the page in any way.

I'm kind of confused how drawing the security chip affects time to first contentful paint. Does animating it in block the first contentful paint...? If so, that seems bad, so should we not animate the chip (just appear it) if it appears before contentful paint?

cc-ing the test-owners for comments.

The functional bug that led to the performance regression is now fixed in 63.0.3217.0, but we still don't have an explanation for why showing the security chip would impact time to first paint. There are other scenarios in which this chip will show (e.g. a HTTP page with a password field) and these scenarios will become more common as we start showing "Not Secure" on all HTTP sites.

kouhei, ksakamoto: ping on helping security chip folks understand its impact on first paint?

Assigning to a test owner for follow up on the question in #6. 

While this specific regression no longer exists, we do not expect the warning UI previously firing here to impact page load time in any way, and we do plan to start showing this warning UI more often in the future, so understanding why the test was impacted would be helpful.

Also adding Ned re #6 and #10 since this likely impacts more benchmarks too.

It sounds like the regression is fixed.
If there's still a question about how the loading metrics work, tdresser's team might be able to help over email.

RE #12: Thanks for the pointer.

This immediate regression was fixed by not entering this state on that site, but the regression is likely to come right back when https://chromium-review.googlesource.com/c/chromium/src/+/843456 gets enabled and shows the UI for all HTTP sites.