All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=672577

Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICgl9Cd6AsM


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

chromium-rel-mac10

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8993814856766935632

===== BISECT JOB RESULTS =====
Status: completed


===== SUSPECTED CL(s) =====
Subject : Don't remove CSSPreloaderResourceClient for unused speculative markup preloads
Author  : csharrison
Commit description:
  
The CSS @import scanner attaches a passive resource client to a css
preload request. This passive client should not affect the policy
decisions of the preload and should just observe notifications passively.

This patch fixes a bug where removing a passive client from an otherwise
unused preload ends up cancelling it, which removes the preload from
memory cache. This is very wrong behavior, and causes the optimization
to be less effective, and report bad metrics.

Simply not removing the client will not cause the resource to live
longer than necessary, because the client holds only weak references
to the resource.

BUG= 670295 , 662999 

Review-Url: https://codereview.chromium.org/2542183002
Cr-Commit-Position: refs/heads/master@{#436312}
Commit  : dff0384e2ab5ade040a08a38e72eece575ea0aee
Date    : Mon Dec 05 16:28:44 2016


===== TESTED REVISIONS =====
Revision         Mean     Std Dev  N   Good?
chromium@436272  276.989  200.746  27  good
chromium@436295  269.833  158.83   17  good
chromium@436306  273.342  121.498  17  good
chromium@436307  286.592  193.358  26  unknown
chromium@436310  282.237  192.613  27  unknown
chromium@436311  276.336  181.745  27  good
chromium@436312  307.007  93.5973  27  bad      <--
chromium@436313  307.036  107.713  18  bad
chromium@436318  309.263  81.4776  18  bad

Bisect job ran on: mac_10_10_perf_bisect
Bug ID: 672577

Test Command: src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=http...www.ign.com. page_cycler_v2.typical_25
Test Metric: timeToFirstMeaningfulPaint_avg/pcv1-warm/http___www.ign.com_
Relative Change: 11.65%

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/mac_10_10_perf_bisect/builds/2507
Job details: https://chromeperf.appspot.com/buildbucket_job_status/8993814856766935632


Not what you expected? We'll investigate and get back to you!
  https://chromeperf.appspot.com/bad_bisect?try_job_id=6270988415664128

| O O | Visit http://www.chromium.org/developers/speed-infra/perf-bug-faq
|  X  | for more information addressing perf regression bugs. For feedback,
| / \ | file a bug with component Tests>AutoBisect.  Thank you!