All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=758579

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=78829483f7a8acf12411579737f3ac8925847c977dd93bd7f20cb7f87279b3fb


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

android-webview-nexus5X

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8970369966572392432

=== Auto-CCing suspected CL author ducbui@google.com ===

Hi ducbui@google.com, 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 : Duc Bui
  Commit : 7be7aceb006f81a3638bba4249c4367ddf659351
  Date   : Thu Aug 17 03:03:05 2017
  Subject: Reland "Reland "[GRC] Track Expected Task Queueing Duration of tabs in GRC.""

Bisect Details
  Configuration: android_webview_arm64_aosp_perf_bisect
  Benchmark    : thread_times.key_idle_power_cases
  Metric       : tasks_per_second_total_all/blank.html
  Change       : 5.79% | 2.98838568073 -> 3.16137869935

Revision             Result                    N
chromium@494943      2.98839 +- 0.112539       6      good
chromium@495020      2.97101 +- 0.0599398      6      good
chromium@495058      3.02941 +- 0.122469       9      good
chromium@495061      3.00378 +- 0.148298       6      good
chromium@495062      3.2787 +- 0.114239        6      bad       <--
chromium@495063      3.16799 +- 0.202332       9      bad
chromium@495068      3.20391 +- 0.0856272      6      bad
chromium@495077      3.21251 +- 0.113767       6      bad
chromium@495096      3.18601 +- 0.130532       6      bad
chromium@495248      3.16138 +- 0.0912825      6      bad

To Run This Test
  src/tools/perf/run_benchmark -v --browser=android-webview --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=blank.html thread_times.key_idle_power_cases

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8970369966572392432


For feedback, file a bug with component Speed>Bisection

+lpy@ and fmeawad@

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8970346879851204016

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

Suspected Commit
  Author : Duc Bui
  Commit : 7be7aceb006f81a3638bba4249c4367ddf659351
  Date   : Thu Aug 17 03:03:05 2017
  Subject: Reland "Reland "[GRC] Track Expected Task Queueing Duration of tabs in GRC.""

Bisect Details
  Configuration: android_webview_arm64_aosp_perf_bisect
  Benchmark    : thread_times.key_idle_power_cases
  Metric       : tasks_per_second_total_all/blank.html
  Change       : 7.26% | 2.97149700931 -> 3.18725419033

Revision             Result                    N
chromium@494943      2.9715 +- 0.0945934       6      good
chromium@495020      2.95424 +- 0.163431       6      good
chromium@495058      2.99549 +- 0.223182       6      good
chromium@495061      2.97967 +- 0.0926899      6      good
chromium@495062      3.19583 +- 0.213791       6      bad       <--
chromium@495063      3.21197 +- 0.207337       6      bad
chromium@495068      3.2457 +- 0.157876        6      bad
chromium@495077      3.20359 +- 0.109432       6      bad
chromium@495096      3.1792 +- 0.0909962       6      bad
chromium@495248      3.18725 +- 0.192781       6      bad

To Run This Test
  src/tools/perf/run_benchmark -v --browser=android-webview --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=blank.html thread_times.key_idle_power_cases

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8970346879851204016


For feedback, file a bug with component Speed>Bisection

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8970335323715755600

Transfer owner to my host fmeawad@

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

Suspected Commit
  Author : Duc Bui
  Commit : 7be7aceb006f81a3638bba4249c4367ddf659351
  Date   : Thu Aug 17 03:03:05 2017
  Subject: Reland "Reland "[GRC] Track Expected Task Queueing Duration of tabs in GRC.""

Bisect Details
  Configuration: android_webview_arm64_aosp_perf_bisect
  Benchmark    : thread_times.key_idle_power_cases
  Metric       : tasks_per_second_total_all/blank.html
  Change       : 6.66% | 2.99647757228 -> 3.19598392707

Revision             Result                    N
chromium@494943      2.99648 +- 0.0991047      6      good
chromium@495020      2.9882 +- 0.110984        6      good
chromium@495058      2.98008 +- 0.116639       6      good
chromium@495061      3.00426 +- 0.129278       6      good
chromium@495062      3.18732 +- 0.238803       6      bad       <--
chromium@495063      3.2129 +- 0.1661          6      bad
chromium@495068      3.19617 +- 0.0994556      6      bad
chromium@495077      3.22925 +- 0.116538       6      bad
chromium@495096      3.21239 +- 0.134423       6      bad
chromium@495248      3.19598 +- 0.0999289      6      bad

To Run This Test
  src/tools/perf/run_benchmark -v --browser=android-webview --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=blank.html thread_times.key_idle_power_cases

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8970335323715755600


For feedback, file a bug with component Speed>Bisection

I tracked this to 6 (over the existing 60) extra tasks on the IO thread that are caused by sending the EQT to the browser process using a mojo message when the IO thread is already doing work. This change did not wake up the IO thread after an idle period or added noticeable CPU work. Inclined to WontFix.

lpy?

When tasks are being throttled or suspended, we don't send EQT, right? If we don't wake up the IO thread or the process itself, I am fine closing it as WONTFIX.

yes.

 Issue 758577  has been merged into this issue.