All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=633202

Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICg2vu0ngsM


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

android-galaxy-s5

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/9005528286070095072

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/9005521224155355392

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/9005430077922837312

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

Hi bashi@chromium.org, the bisect results pointed to your CL below as possibly
causing a regression. Please have a look at this info and see whether
your CL be related.


===== BISECT JOB RESULTS =====
Status: completed


===== SUSPECTED CL(s) =====
Subject : Add IDLDictionaryBase
Author  : bashi
Commit description:
  
Before this CL, given an IDL dictionary (e.g. FooDictionary)
the code generator generated ToV8() directly.
This is problematic when Blink wants to pass a
sub-dictionary of FooDictionary because generated
ToV8() doesn't convert members defined in the
sub-dictionary.

To solve this problem, add a base class which
provides toV8Impl() virtual function. The code
generator overrides toV8Impl() and ToV8() uses
them. ToV8() automatically calls appropriate
toV8Impl().

BUG= 630210 

Review-Url: https://codereview.chromium.org/2183623004
Cr-Commit-Position: refs/heads/master@{#408093}
Commit  : 190cffeadfe46c6419789d586551c877b24f082f
Date    : Wed Jul 27 11:06:30 2016


===== TESTED REVISIONS =====
Revision         Mean     Std Dev  N  Good?
chromium@408081  1089536  0.0      5  good
chromium@408091  1089536  0.0      5  good
chromium@408092  1089536  0.0      5  good
chromium@408093  1094451  1831.79  5  bad    <--
chromium@408094  1093632  0.0      5  bad
chromium@408096  1124762  65060.7  5  bad
chromium@408100  1095270  2243.47  5  bad

Bisect job ran on: android_s5_perf_bisect
Bug ID: 633202

Test Command: src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --also-run-disabled-tests system_health.memory_mobile
Test Metric: load_news-memory:chrome:all_processes:reported_by_os:system_memory:ashmem:private_dirty_size_avg/load_news_wikipedia
Relative Change: 0.53%
Score: 99.0

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/android_s5_perf_bisect/builds/866
Job details: https://chromeperf.appspot.com/buildbucket_job_status/9005521224155355392


Not what you expected? We'll investigate and get back to you!
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5814013825384448

| 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 Tests>AutoBisect.  Thank you!

One alert, looks like noise.

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/9003880399650103152

Given the low standard deviations, this looks like a small, but very clear +4 KiB regression.

bashi: Can you please investigate?

===== BISECT JOB RESULTS =====
Status: completed


===== SUSPECTED CL(s) =====
Subject : Add IDLDictionaryBase
Author  : bashi
Commit description:
  
Before this CL, given an IDL dictionary (e.g. FooDictionary)
the code generator generated ToV8() directly.
This is problematic when Blink wants to pass a
sub-dictionary of FooDictionary because generated
ToV8() doesn't convert members defined in the
sub-dictionary.

To solve this problem, add a base class which
provides toV8Impl() virtual function. The code
generator overrides toV8Impl() and ToV8() uses
them. ToV8() automatically calls appropriate
toV8Impl().

BUG= 630210 

Review-Url: https://codereview.chromium.org/2183623004
Cr-Commit-Position: refs/heads/master@{#408093}
Commit  : 190cffeadfe46c6419789d586551c877b24f082f
Date    : Wed Jul 27 11:06:30 2016


===== TESTED REVISIONS =====
Revision         Mean     Std Dev  N  Good?
chromium@408092  1089536  0.0      5  good
chromium@408093  1094451  1831.79  5  bad    <--
chromium@408094  1094451  1831.79  5  bad

Bisect job ran on: android_s5_perf_bisect
Bug ID: 633202

Test Command: src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --also-run-disabled-tests system_health.memory_mobile
Test Metric: load_news-memory:chrome:all_processes:reported_by_os:system_memory:ashmem:private_dirty_size_avg/load_news_wikipedia
Relative Change: 0.45%
Score: 99.0

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/android_s5_perf_bisect/builds/926
Job details: https://chromeperf.appspot.com/buildbucket_job_status/9003880399650103152


Not what you expected? We'll investigate and get back to you!
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5903168639074304

| 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 Tests>AutoBisect.  Thank you!

I added a base class which has vtable and it will describe the regression but it's very difficult to avoid such cost to fix the bug. I would like to mark this as WontFix. What do you think?

I agree. Thanks for looking into this :-)