All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=693455

Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg-MCWvgoM


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

android-webview-nexus5X

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8987421914416232336

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

Hi paulmiller@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 : paulmiller
  Commit : 635a45202f8cdaef30cfa4c5d4b602660e81395e
  Date   : Wed Feb 15 19:50:15 2017
  Subject: WebView: prepare to depend on individual GMS services

Bisect Details
  Configuration: android_webview_arm64_aosp_perf_bisect
  Benchmark    : system_health.memory_mobile
  Metric       : memory:webview:all_processes:reported_by_os:system_memory:java_heap:proportional_resident_size_avg/load_games/load_games_spychase
  Change       : 8.35% | 9049088.0 -> 8293376.0

Revision             Result                  N
chromium@450759      9049088 +- 224702       6      good
chromium@450773      9573376 +- 174813       6      good
chromium@450777      9799680 +- 253075       6      good
chromium@450779      9970347 +- 357570       6      good
chromium@450780      9179136 +- 241768       6      bad       <--
chromium@450787      8834731 +- 172657       6      bad
chromium@450815      8293376 +- 1994201      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-webview --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=load.games.spychase system_health.memory_mobile

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8987421914416232336

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=6467441061789696


| 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!

Sorry, I don't know how to interpret this. Do those numbers mean "bytes used"? The numbers went down, so where's the regression? The referenced change removed some libraries from the SystemWebView.apk, which I'd expect would improve memory usage, if that's the target being tested.

Note-I'm going to be merging that and some related changes to m57 to ameliorate a previous memory regression: https://crbug.com/671024.

Re #4: Sorry, the bisect isn't very smart about understanding improvements vs regressions. It looks for a significant change between the start and end revisions, and pinpoints that one. You're right that it's measuring bytes and it found a decrease at your CL.

+perezju, hjd: Looks like this metric could be bimodal? It went up in the graphs, but down in the bisect. But if you look at the graphs it seems to fluctuate between two values.

Yes, looking at the larger picture, I don't think there is a regression here but some noise issues:
https://chromeperf.appspot.com/report?sid=70258930278acde35c90971136a000d0a4b2182b781177433c129d880233780f&start_rev=434783&end_rev=451597

WontFix-ing per #6