All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=792449

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=9a2f2d04e8ca602db146c655ac93dec9100b0654acf5011c03f1d4323a337090


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

android-one

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8960949479456872512

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

Hi xlai@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 : xlai
  Commit : 94b2f4f94c6d0e92fd078f7e776f7bff7fcc8571
  Date   : Mon Dec 04 23:56:03 2017
  Subject: Remove Canvas Display List 2D mode

Bisect Details
  Configuration: android_one_perf_bisect
  Benchmark    : smoothness.tough_canvas_cases
  Metric       : frame_times/tough_canvas_cases_rendering_throughput_put_get_image_data.html
  Change       : 38.60% | 269.929162961 -> 374.298616372

Revision             Result                  N
chromium@521388      269.929 +- 17.5567      6      good
chromium@521466      277.071 +- 46.8445      6      good
chromium@521505      270.012 +- 35.0588      5      good
chromium@521525      261.854 +- 2.45862      5      good
chromium@521530      269.792 +- 34.1629      5      good
chromium@521532      264.575 +- 17.7049      5      good
chromium@521533      375.096 +- 9.15153      6      bad       <--
chromium@521535      372.161 +- 9.49246      6      bad
chromium@521544      374.299 +- 15.4065      7      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=tough.canvas.cases.rendering.throughput.put.get.image.data.html smoothness.tough_canvas_cases

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8960949479456872512


For feedback, file a bug with component Speed>Bisection

junov@: I think it's because the DisableAccleration() in html canvas was wrongly using the RecordingImageBufferSurface previously whilst display list 2d mode is disabled and this CL unmask it, as this regression happens in Android One only. But I'm wondering why Canvas2DLayerBridge(disabled mode) cannot do what's been done previously in RecordingImageBufferSurface.

Yeah, we cancelled an unintended performance boost that we were getting from a deprecated (not to mention buggy) code path. Let's not fix this.  The way forward is to forge ahead towards implementing canvas OOP raster, that should improve performance in a similar way, without the bugs.