All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=809485

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=6bfddda721b4c58e3caf762c560aa4dce643846ddf464ca761edef11f2582820


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

chromium-rel-win7-x64-dual

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/16c50391840000

📍 Found significant differences after each of 4 commits.
https://pinpoint-dot-chromeperf.appspot.com/job/16c50391840000

Update the region lines property to latest spec by k2.nagaraju@samsung.com
chromium @ 2fb26fdd1ccb836c2f02e4dbf213afaed0a00e50

cc: Defer impl-side invalidations for main frame with sync scheduling. by khushalsagar@chromium.org
chromium @ 10d3a66963310f7dc33b4f6478304439784da28e

bindings: Allow `any' dictionary members to be null. by raphael.kubo.da.costa@intel.com
chromium @ 05865333959b92ae34caa4c61bc0e76358c71b8d

Rename editing/inserting/4960120-2.html to insert_line_break_at_end_of_anonymous_block.html by yosin@chromium.org
chromium @ 4c95848c61b19cb858476e610becc6006229d4b0

Understanding performance regressions:
  http://g.co/ChromePerformanceRegressions

Remove myself because my change renames layout test, it can't affect 
browser executable.

Its definitely not 10d3a66963310f7dc33b4f6478304439784da28e, that only affects the code path for compositor scheduling on Android Webview.

yukishiino, looking at the graph I think this is "bindings: Allow `any' dictionary members to be null". by raphael.kubo.da.costa@intel.com, but crbug won't allow me to assign to raphael.kubo.da.costa. Can one of you take a look?

Also adding benchmark owner wangxianzhu in case he has comments on importance; this only regressed on a single bot.

I'm here. I remember looking at the graph at the time and not being convinced it was my CL because https://cs.chromium.org/chromium/src/third_party/WebKit/PerformanceTests/Paint/fixed-and-many-layers-scroll.html?&sq=package:chromium&l=1 doesn't seem to be referencing any IDL operation or attribute with the `any` type.

lfg@, could you triage this issue?

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/16d13c0e440000

This regressions seems to be caused by enabling RLS.

📍 Found significant differences after each of 16 commits.
https://pinpoint-dot-chromeperf.appspot.com/job/16d13c0e440000

Move non client-library code from services/network/public/cpp out of the public directory. by jam@chromium.org
https://chromium.googlesource.com/chromium/src/+/89c02aea795ae75eab45419574d7c5db092a4a17

Add layout tests exercising suspend/resume of src= media elements. by dalecurtis@chromium.org
https://chromium.googlesource.com/chromium/src/+/99a9b48a8b0294acebee03bd38cb96e9569de41a

Duplicate SerializedHandle definition in //ipc/ipc.mojom by evem@chromium.org
https://chromium.googlesource.com/chromium/src/+/383325a1f76a746acd9bd0c1e2a750091cc249ed

Share common code of Document::createElement() and createElementNS(). by tkent@chromium.org
https://chromium.googlesource.com/chromium/src/+/b8a7fc604162683306efeebf4d5b6f989566002e

Roll src/third_party/angle/ 61d5325e6..15443ebad (1 commit) by angle-chromium-autoroll@skia-buildbots.google.com.iam.gserviceaccount.com
https://chromium.googlesource.com/chromium/src/+/6e5da8b0062c892e3f2662a885d7ca124f1621bc

Directional code cleanup in editing module by tanvir.rizvi@samsung.com
https://chromium.googlesource.com/chromium/src/+/7ee0cb4bd0396c0ed26bbf9611cf310fcc096bb3

Define legacy getLocalStreams() on top of track-based API. by hbos@chromium.org
https://chromium.googlesource.com/chromium/src/+/55a6f822232bb39ce9296fe9de7baae0df12b244

Change some SSL logging to use DLOG by jonasolsson@webrtc.org
https://webrtc.googlesource.com/src/+/addc3801682034da98055a09fd1566b65f41ff06

Roll src/third_party/webrtc/ c38d32068..06c944f03 (4 commits) by webrtc-chromium-autoroll@skia-buildbots.google.com.iam.gserviceaccount.com
https://chromium.googlesource.com/chromium/src/+/5ad5a3d2ef938e9a98e93ff8c498b7662ba6de3b

Add traffic annotation to SocketBIOAdapter. by rhalavati@chromium.org
https://chromium.googlesource.com/chromium/src/+/35565a5d3d623f5420d066a4bf9814e6c5075998

custom-elements: createElement() should not throw for unknown "is" value. by tkent@chromium.org
https://chromium.googlesource.com/chromium/src/+/ae25beae260de5f2560003a65c46710eb296570f

Capture all DOM nodes reachable from V8 in heap snapshot. by ulan@chromium.org
https://chromium.googlesource.com/chromium/src/+/75edf20e769f39825d6e17fab52e16589350e7f2

Use base::ResetAndReturn() in //net. by bnc@chromium.org
https://chromium.googlesource.com/chromium/src/+/bc1de2973401b6b13b3ab1b2b2a12981f0daa42d

[RootLayerScrolls] Flip RuntimeEnabledFeatures flag to experimental by szager@chromium.org
https://chromium.googlesource.com/chromium/src/+/fe8c4db4003022f5cb12506a15cc78ea6633bbdf

[PE] Correct anchor position in printed subframe by wangxianzhu@chromium.org
https://chromium.googlesource.com/chromium/src/+/11b92d75fea81c74ce5c9d3e71a5088bd1a4ff20

Revert "Capture all DOM nodes reachable from V8 in heap snapshot." by steimel@chromium.org
https://chromium.googlesource.com/chromium/src/+/0c2648420c80bff20154a6abd2d267a9a53cc6ad

Understanding performance regressions:
  http://g.co/ChromePerformanceRegressions

Still most likely RLS.

Is this regression recovered now?

This has recovered. The progression range is:
https://chromium.googlesource.com/chromium/src/+log/ed5ff53b4265585e42e6cdfc0681f8c221375fa3%5E..ab61a76223b2b14b4e3b348dae07e727be3cc2c7?pretty=fuller&n=1000

I think this was fixed by:
[Blink] Avoid O(n) lookup during clip parent computation
http://crrev.com/542605
trchen@chromium.org