All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=604266

Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICghOuBvwoM


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

android-nexus7v2

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

Hi dmurph@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 : [BlobAsync] Faster shortcuttin, make renderer controller leaky & alive.
Author  : dmurph
Commit description:
  
This should fix strange memory access crashes observed in the
following bugs by doing the following things:
1. We make the controller a leaky lazy instance so it isn't destructed.
2. We make one less message loop hop before we call AddProcessRef, so
   hopefully our process won't be shutting down in the middle of a
   transfer.
3. We send the 'descriptions' as early as possible if we're below the
   IPC threshold, which should help some of the speed regressions.
4. We prevent sudden shutdowns (like tab bar closing) by using
   blink::Platform::current()->suddenTerminationChanged(false);

BUG=600462,600435, 599490 , 599416 , 375297 

Review URL: https://codereview.chromium.org/1853333003

Cr-Commit-Position: refs/heads/master@{#385472}
Commit  : 90774e39bb24b1c531e9b3aba16c77ab96675fb3
Date    : Wed Apr 06 16:21:48 2016


===== TESTED REVISIONS =====
Revision                Mean Value  Std. Dev.   Num Values  Good?
chromium@385464         6.457       0.04995     5           good
chromium@385469         6.489       0.044215    5           good
chromium@385471         6.476       0.032863    5           good
chromium@385472         6.807       0.091692    5           bad         <-
chromium@385473         6.773       0.048296    5           bad
chromium@385481         6.817       0.040249    5           bad
chromium@385497         6.835       0.06364     5           bad

Bisect job ran on: android_nexus7_perf_bisect
Bug ID: 604266

Test Command: src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --also-run-disabled-tests service_worker.service_worker_micro_benchmark
Test Metric: concurrent_1_response_50_percentile/concurrent_1_response_50_percentile
Relative Change: 5.85%
Score: 99.9

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/android_nexus7_perf_bisect/builds/2915
Job details: https://chromeperf.appspot.com/buildbucket_job_status/9015087286045254272


Not what you expected? We'll investigate and get back to you!
  https://chromeperf.appspot.com/bad_bisect?try_job_id=604266

| 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!

Remove performance-sheriff, as the blink-worker team takes care of SW microbenchmark regressions, not perf sheriffs.

I'm inclined to accept the regression as it looks like a constructive change (fix strange memory access crashes). dmurph@ do you expect this to slow down blob transferring somehow?

The switch to async-transfering of blobs can slow down the transfer in some cases. In large amounts of data, it'll be a little slower, as it's not a request-response IPC structure instead of a synchronous ipc system.

This can be played with in the future by allocating more shared memory buffers, but this is is necessary for allowing us to remove the 500MB limit for blobs. This small regression was necessary to make the small blob path faster

Marking as wontfix.