All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=747523

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=c436ee0a727b15996d7e9dcd5941373650268bcd5cc1c1df462c563fe25f4c53


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

chromium-rel-win7-gpu-intel

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8973433076857227984

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

Hi hajimehoshi@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 : Hajime Hoshi
  Commit : 52448951331fb305cd6342b4fc84542dc4d60d62
  Date   : Thu Jul 20 09:30:57 2017
  Subject: Enable new ownership edges for base::SharedMemory

Bisect Details
  Configuration: winx64intel_perf_bisect
  Benchmark    : media.tough_video_cases_tbmv2
  Metric       : memory:chrome:all_processes:reported_by_chrome:sharedbitmap:effective_size_avg/video.html?src_crowd720_vp9.webm
  Change       : 106300.00% | 4352.0 -> 4630528.0

Revision             Result              N
chromium@488141      4352.0 +- 0.0       6      good
chromium@488168      4352.0 +- 0.0       6      good
chromium@488172      4352.0 +- 0.0       6      good
chromium@488173      4352.0 +- 0.0       6      good
chromium@488174      4630528 +- 0.0      6      bad       <--
chromium@488175      4630528 +- 0.0      6      bad
chromium@488181      4630528 +- 0.0      6      bad
chromium@488194      4630528 +- 0.0      6      bad

To Run This Test
  src/tools/perf/run_benchmark -v --browser=release_x64 --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=video.html.src.crowd720.vp9.webm media.tough_video_cases_tbmv2

More information on addressing performance regressions:
  http://g.co/ChromePerformanceRegressions

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8973433076857227984


For feedback, file a bug with component Speed>Bisection

 Issue 747526  has been merged into this issue.

 Issue 747537  has been merged into this issue.

Trace before the problem: https://console.developers.google.com/m/cloudstorage/b/chrome-telemetry-output/o/trace-file-id_2-2017-07-20_02-46-22-99576.html
Trace after the problem: https://console.developers.google.com/m/cloudstorage/b/chrome-telemetry-output/o/trace-file-id_2-2017-07-20_08-21-34-3242.html

* The latter includes GPU process data, but the former doesn't. Why?
* The latter includes shared memory usages, but the former doesn't. Our CL seems to blame, but I am not 100% sure.

Oops, I saw a different trace. The former is 

https://console.developers.google.com/m/cloudstorage/b/chrome-telemetry-output/o/trace-file-id_2-2017-07-20_02-35-12-93480.html

The latter is not changed.

So there is a big bump jump from 0 to 0.5MB in 'sharedbitmap' effective size, but it looks like the same reduction in 'shared_memory' effective size. I think this is because of our new ownership edges, and now we can see the more accurate (effective) memory usage in sharedbitmap. Thus, I think this is an expected behavior.

primiano@, wdyt?

Took a look at some random traces following before & after in https://chromeperf.appspot.com/group_report?bug_id=747523. Looks like memory is moving form shared_memory to the actual owners (sharedbitmap in this case), which is a definitely WAI result. This looks like a happy "finally works as intended" to me :)
ssid@, if you have some mins can you give another pair of eyes to this before we close it?

Issue 747926 has been merged into this issue.

 Issue 747524  has been merged into this issue.

 Issue 747538  has been merged into this issue.

 Issue 747538  has been merged into this issue.

 Issue 747538  has been merged into this issue.

 Issue 747538  has been merged into this issue.

ping ssid@

This is not a real regression, but I think this is still a real issue.
The dump "cc/resource_provider/provider_0/resource_17" in most of the traces does not call pmd->CreateSharedMemoryOwnershipEdge() at all i guess. The dump mentioned is double counted in "sharedbitmap" in browser because the renderer does not have correct importance, because CreateSharedMemoryOwnershipEdge() is not called.

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8964725317606990544

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8964725285734323616

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

Bisect Details
  Configuration: winx64intel_perf_bisect
  Benchmark    : media.desktop
  Metric       : memory:chrome:all_processes:reported_by_chrome:sharedbitmap:effective_size_avg/video.html?src_crowd720_vp9.webm


To Run This Test
  src/tools/perf/run_benchmark -v --browser=release_x64 --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=video.html.src.crowd720.vp9.webm media.desktop

More information on addressing performance regressions:
  http://g.co/ChromePerformanceRegressions

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8964725317606990544


For feedback, file a bug with component Speed>Bisection

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

Bisect Details
  Configuration: winx64intel_perf_bisect
  Benchmark    : media.desktop
  Metric       : memory:chrome:all_processes:reported_by_chrome:sharedbitmap:effective_size_avg/video.html?src_crowd720_vp9.webm


To Run This Test
  src/tools/perf/run_benchmark -v --browser=release_x64 --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=video.html.src.crowd720.vp9.webm media.desktop

More information on addressing performance regressions:
  http://g.co/ChromePerformanceRegressions

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8964725285734323616


For feedback, file a bug with component Speed>Bisection

ssid@, so what we should do is to create dump even when the shared memory is not mapped?

WontFix. Bisect didn't work and bug is stale.

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8962810214353116160

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

Bisect Details
  Configuration: winx64intel_perf_bisect
  Benchmark    : media.desktop
  Metric       : memory:chrome:all_processes:reported_by_chrome:sharedbitmap:effective_size_avg/video.html?src_crowd720_vp9.webm


To Run This Test
  src/tools/perf/run_benchmark -v --browser=release_x64 --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=video.html.src.crowd720.vp9.webm media.desktop

More information on addressing performance regressions:
  http://g.co/ChromePerformanceRegressions

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8962810214353116160


For feedback, file a bug with component Speed>Bisection