All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=701810

Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDgtNun7QgM


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

android-nexus5

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8985041131577718592

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

Hi chaopeng@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 : chaopeng
  Commit : b2b25bd2d961354db5081b6090d3782c10513fab
  Date   : Mon Mar 13 20:52:51 2017
  Subject: Show Overlay Scrollbar when GestureScrollUpdate

Bisect Details
  Configuration: android_nexus5_perf_bisect
  Benchmark    : system_health.memory_mobile
  Metric       : memory:chrome:all_processes:reported_by_os:gpu_memory:proportional_resident_size_avg/browse_news/browse_news_nytimes
  Change       : 21.74% | 73259690.6667 -> 89184938.6667

Revision             Result                   N
chromium@456471      73259691 +- 1779821      6      good
chromium@456473      72609792 +- 2387880      6      good
chromium@456474      89184939 +- 1779821      6      bad       <--
chromium@456477      89509888 +- 0.0          6      bad
chromium@456483      89184939 +- 1779821      6      bad
chromium@456494      89184939 +- 1779821      6      bad
chromium@456516      89509888 +- 0.0          6      bad
chromium@456561      89184939 +- 1779821      6      bad

Please refer to the following doc on diagnosing memory regressions:
  https://chromium.googlesource.com/chromium/src/+/master/docs/memory-infra/memory_benchmarks.md

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=browse.news.nytimes system_health.memory_mobile

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8985041131577718592

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=6381107781566464


| 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 Speed>Bisection.  Thank you!

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8985033056416994432

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

Suspected Commit
  Author : chaopeng
  Commit : b2b25bd2d961354db5081b6090d3782c10513fab
  Date   : Mon Mar 13 20:52:51 2017
  Subject: Show Overlay Scrollbar when GestureScrollUpdate

Bisect Details
  Configuration: android_nexus5_perf_bisect
  Benchmark    : system_health.memory_mobile
  Metric       : memory:chrome:all_processes:reported_by_os:gpu_memory:proportional_resident_size_avg/browse_news/browse_news_nytimes
  Change       : 21.29% | 73259690.6667 -> 88859989.3333

Revision             Result                   N
chromium@456471      73259691 +- 1779821      6      good
chromium@456473      73259691 +- 1779821      6      good
chromium@456474      89509888 +- 0.0          6      bad       <--
chromium@456477      89184939 +- 1779821      6      bad
chromium@456483      89184939 +- 1779821      6      bad
chromium@456494      89184939 +- 1779821      6      bad
chromium@456516      89509888 +- 0.0          6      bad
chromium@456561      88859989 +- 2251315      6      bad

Please refer to the following doc on diagnosing memory regressions:
  https://chromium.googlesource.com/chromium/src/+/master/docs/memory-infra/memory_benchmarks.md

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=browse.news.nytimes system_health.memory_mobile

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8985033056416994432

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=6089819475673088


| 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 Speed>Bisection.  Thank you!

Sigh...ok, it looks pretty confident but I don't understand it...I'd just do a revert to confirm it's your patch. If you confirm it - try to repro locally.

I can not reproduce it locally. I got same result with/without my patch.

```
"browse:news:nytimes": {
  "std": 0.0,
  "name": "memory:chrome:all_processes:reported_by_os:gpu_memory:proportional_resident_size_avg",
  "page_id": 54,
  "grouping_keys": {
    "case": "browse",
    "group": "news"
  },
  "improvement_direction": "down",
  "important": false,
  "values": [
    97210368
  ],
  "units": "sizeInBytes",
  "tir_label": "browse_news",
  "type": "list_of_scalar_values",
  "description": "total proportional resident size (PSS) of GPU memory (Android memtrack) used by all processes in Chrome"
}
```

I can reproduce locally on Nexus5

1. ScrollbarAnimationController::DidResize did not consider currently_scrolling_, but this is not the reason of memory issue.

2. 

```
void ScrollbarAnimationController::DidResize() {
  StopAnimation();
  Show();

  // As an optimization, we avoid spamming fade delay tasks during active fast
  // scrolls.
  if (!currently_scrolling_) {
    PostDelayedFadeOut(true); <-- false can fix the memory
  } else {
    show_in_fast_scroll_ = true;
  }
}
```

We use 300ms for normal fadeout(false) and 2000ms for resize fadeout(true). https://cs.chromium.org/chromium/src/content/renderer/gpu/render_widget_compositor.cc?l=395

Patch relanded.