All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=695522

Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDghK6vvgkM


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

android-nexus7v2

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8986846520998169520

=== BISECT JOB RESULTS ===
NO Perf regression found, tests failed to produce values

Bisect Details
  Configuration: android_nexus7_perf_bisect
  Benchmark    : tracing_perftests
  Metric       : TraceEventPerfTest.Submit_10000_TRACE_EVENT_with_TracedValue/events_with_value


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=events.with.value tracing_perftests

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8986846520998169520

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=6431681591902208


| 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!

My change enables some extra code in TaskStopwatch, even when the profiler is off.
There's an extra bool that's set at construction <https://cs.chromium.org/chromium/src/base/tracked_objects.cc?l=935>, which is then tested twice (in Start/Stop). There's also an additional ThreadHeapUsageTracker constructed, but that's just a nullptr assignment. If this test is effectively testing creation/destruction of TaskStopwatch, it's possible that this adds 7% to that.

Thanks siggi@
Will take a look

Hmm afaict this is just doing lot of TRACE_EVENT_INSTANT1 in a tight loop, nothing to do with ThreadUsageTracker, or just threads at all.
Let's see what kraynov finds out.

tl;dr Looks like we have regression in creation of TracedValue

More details here:
https://codereview.chromium.org/2711313005

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8986229476054864528

There is no evidence of regression.
1. This looks quite device-specific https://chromeperf.appspot.com/report?sid=49caea6d5d399e27b61b940c93ca17fca689b129e23b3ffbf3643b04f2a235d7

2. Benchmark https://codereview.chromium.org/2711313005 seems to be within noise level and plots (see file attached) shows non-deterministinc result of comparison with and witout enable_task_profiling.

3. No code of task profiler ever called in tracing_perftests.

4. No extra allocator dispatches were registered.

So I believe that test has been affected by external factor and is not related to task profiler.

Deleted: Screenshot from 2017-03-02 15:55:57.png 41.9 KB

	Screenshot from 2017-03-02 15:55:57.png 41.9 KB View Download

I think we need to add ref builds testing on that benchmark. I will file a separate bug.

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8982624366916778560

=== BISECT JOB RESULTS ===
Bisect failed for unknown reasons

Please contact the team (see below) and report the error.


Bisect Details
  Configuration: android_nexus7_perf_bisect
  Benchmark    : tracing_perftests
  Metric       : TraceEventPerfTest.Submit_10000_TRACE_EVENT_with_TracedValue/events_with_value


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=events.with.value tracing_perftests

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8982624366916778560

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5287561829810176


| 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!