All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=783647

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=8228dd5015326850725204f72bef503616dd43ad433f7a85adb5685ae8d3c244


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

android-nexus7v2

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8963323243922912768

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

Hi sullivan@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 : Annie Sullivan
  Commit : 0ea04e064baa8615e5ef9702050ce6a605435c57
  Date   : Fri Nov 03 18:38:38 2017
  Subject: Revert "Add CrashKey to LazyLineBreakIterator on Android"

Bisect Details
  Configuration: android_nexus7_perf_bisect
  Benchmark    : thread_times.key_silk_cases
  Metric       : thread_raster_cpu_time_per_frame/infinite_scrolling.html
  Change       : 6.91% | 1.02974581782 -> 1.1009099652

Revision             Result                    N
chromium@513769      1.02975 +- 0.0271627      6      good
chromium@513834      1.02892 +- 0.0313632      6      good
chromium@513851      1.02548 +- 0.0305348      6      good
chromium@513852      1.06477 +- 0.0410162      9      bad       <--
chromium@513853      1.06637 +- 0.0313211      6      bad
chromium@513855      1.07269 +- 0.0399098      6      bad
chromium@513859      1.07436 +- 0.020937       6      bad
chromium@513867      1.11062 +- 0.0437609      6      bad
chromium@513899      1.10091 +- 0.0257574      6      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=infinite.scrolling.html thread_times.key_silk_cases

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8963323243922912768


For feedback, file a bug with component Speed>Bisection

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8963318339066922720

Removing a crashkey doesn't sound like it should regress thread times.  Lets run that bisect again.

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

Suspected Commit
  Author : Annie Sullivan
  Commit : 0ea04e064baa8615e5ef9702050ce6a605435c57
  Date   : Fri Nov 03 18:38:38 2017
  Subject: Revert "Add CrashKey to LazyLineBreakIterator on Android"

Bisect Details
  Configuration: android_nexus7_perf_bisect
  Benchmark    : thread_times.key_silk_cases
  Metric       : thread_raster_cpu_time_per_frame/infinite_scrolling.html
  Change       : 7.94% | 1.02414127779 -> 1.10542204541

Revision             Result                    N
chromium@513700      1.02414 +- 0.0314858      6      good
chromium@513810      1.02941 +- 0.0274118      6      good
chromium@513838      1.00988 +- 0.0164761      6      good
chromium@513845      1.03073 +- 0.0259515      6      good
chromium@513849      1.02969 +- 0.0302798      6      good
chromium@513851      1.01657 +- 0.0238507      6      good
chromium@513852      1.09715 +- 0.0201925      6      bad       <--
chromium@513865      1.10259 +- 0.0343485      6      bad
chromium@513920      1.10542 +- 0.0209536      6      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=infinite.scrolling.html thread_times.key_silk_cases

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8963318339066922720


For feedback, file a bug with component Speed>Bisection

That's weird.

 Issue 783724  has been merged into this issue.

Super weird, but the original CL caused widespread performance regressions and crashes, so I think a small regression on revert is okay.

Adding original CL author kojii and thread_times.key_silk_cases owner vmiura to see how much they want to investigate.

Unassigning myself as I'm actually going OOO through Monday.

Sorry, no idea why revert made the perf worse.