The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/e4003e2ea8587d47f18c7807885cd364c32246c3

commit e4003e2ea8587d47f18c7807885cd364c32246c3
Author: tansell <tansell@chromium.org>
Date: Tue Jun 06 01:40:10 2017

LayoutTest: fast/workers/worker-document-leak.html is flaky on Windows.

BUG= 729836 
TBR=kbr@google.com,sigbjornf@opera.com

Review-Url: https://codereview.chromium.org/2923093002
Cr-Commit-Position: refs/heads/master@{#477148}

[modify] https://crrev.com/e4003e2ea8587d47f18c7807885cd364c32246c3/third_party/WebKit/LayoutTests/TestExpectations

This failure is potentially problematic. While I'm quite sure the old bug here has not reoccurred (that creating a Worker led to retaining the page's Document instance forever), I wonder whether the Blink scheduler may have caused the asyncGC function used by this layout test to no longer work reliably.

https://cs.chromium.org/chromium/src/third_party/WebKit/LayoutTests/resources/js-test.js?q=asyncGC+package:%5Echromium$&l=750

Here are the two recent failures:

https://storage.googleapis.com/chromium-layout-test-archives/WebKit_Win7/52981/layout-test-results/results.html

https://storage.googleapis.com/chromium-layout-test-archives/WebKit_Win10/22685/layout-test-results/results.html

In both of these cases it looks like all of the Document instances associated with the iframes were leaked. This could indicate that a severe (though intermittent) memory leak was recently introduced.

---------

--- E:\b\rr\tmpeazvbv\w\layout-test-results\fast/workers/worker-document-leak-expected.txt
+++ E:\b\rr\tmpeazvbv\w\layout-test-results\fast/workers/worker-document-leak-actual.txt
@@ -3,7 +3,7 @@
 On success, you will see a series of "PASS" messages, followed by "TEST COMPLETE".
 
 
-PASS did not leak documents during test run
+FAIL leaked documents during test run (started with 1, ended with 7)
 PASS successfullyParsed is true
 
 TEST COMPLETE

Failures apparently started on Jun 2:
https://test-results.appspot.com/dashboards/flakiness_dashboard.html#testType=webkit_tests&tests=fast%2Fworkers%2Fworker-document-leak

Anyone with a Windows box want to bisect? Is the TE team equipped to do bisects for layout test failures?

I don't think we've changed anything related to timer scheduling around June 2nd, but yes, a bisect would be helpful.

I believe this has been low level flaky for a while, I wouldn't use the date I logged this bug as a restriction on the date range for this problem.

The date comes from the flakiness dashboard:
https://test-results.appspot.com/dashboards/flakiness_dashboard.html#testType=webkit_tests&tests=fast%2Fworkers%2Fworker-document-leak

The red and grey boxes for Windows started on June 2nd.

Is there other data source that indicates it's been flaky for longer than that?

Looking at that data, it seems to have always been flaky on Win as far back as there is data?

This is a confusing limitation of the flakiness dashboard. It does not collect data until the test failed once. So the empty boxes actually can mean the test ran and passed. It could also mean the test did not exist or was not run. But according to the commit history this test existed since 2013 and I don't see a Skip expectation in TestExpectations any time recently.

Just to reply to C#3, TE team dont have set up to bisect such layout test failures.

nhiroki@
Do you think this is related to your worker shutdown work?

Can we have the latest update on this issue?

Updating labels and removing from bisect bucket since TE cannot repro.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/e286250b9e5c11b0a9e14d6b243dc206999c4a50

commit e286250b9e5c11b0a9e14d6b243dc206999c4a50
Author: Quinten Yearsley <qyearsley@chromium.org>
Date: Fri Jan 05 00:42:57 2018

Update TestExpectations with update-flaky-expectations

This CL was initially based on changes made by the third_party/WebKit/Tools/Scripts/update-flaky-expectations, but includes manual changes as well, including formatting changes and making expectations more specific based on visually checking the flakiness dashboard results.

Flakiness dashboard link:
https://test-results.appspot.com/dashboards/flakiness_dashboard.html#testType=webkit_tests&tests=http/tests/devtools/coverage/coverage-view.js,inspector-protocol/layout-fonts/cjk-ideograph-fallback-by-lang.js,http/tests/websocket/close-code-and-reason.html,external/wpt/css/css-writing-modes/writing-mode-vertical-rl-001.xht,external/wpt/css/css-writing-modes/writing-mode-vertical-rl-002.xht,external/wpt/css/css-fonts/font-display/font-display.html,fast/workers/chromium/worker-document-leak.html,fast/dom/Window/window-resize-contents.html,fast/text/ellipsis-in-relative-inline.html,fast/spatial-navigation/snav-1st-stop.html,fast/spatial-navigation/snav-display-contents-crash.html,fast/spatial-navigation/snav-div-in-anchor-and-img-crash.html,fast/spatial-navigation/snav-div-in-anchor.html,fast/spatial-navigation/snav-hidden-focusable-element.html,fast/spatial-navigation/snav-iframe-nested.html,fast/spatial-navigation/snav-iframe-no-focusable-content.html,fast/spatial-navigation/snav-iframe-no-scrollable-content.html,fast/spatial-navigation/snav-iframe-with-offscreen-focusable-element.html,fast/spatial-navigation/snav-media-elements.html,fast/spatial-navigation/snav-multiple-select-focusring.html,fast/spatial-navigation/snav-offscreen-content.html,fast/spatial-navigation/snav-only-clipped-overflow-content.html,fast/spatial-navigation/snav-simple-content-overflow.html,fast/spatial-navigation/snav-single-select-list.html,fast/spatial-navigation/snav-zero-margin-content.html,external/wpt/clear-site-data/storage.https.html,http/tests/devtools/network/network-xhr-same-url-as-main-resource.js,virtual/mouseevent_fractional/fast/events/touch/gesture/gesture-tap-hover-state-iframe.html

Bug:  518989 ,  520737 , 520739,  564109 ,  729836 ,  731535 ,  746904 , 757165, 767040,  767269 ,  773353 ,  785980 ,  795250 
Change-Id: I03ac7e658ee42ccf3798e7f3994d30fc74880565
Reviewed-on: https://chromium-review.googlesource.com/847835
Commit-Queue: Quinten Yearsley <qyearsley@chromium.org>
Reviewed-by: Philip Jägenstedt <foolip@chromium.org>
Cr-Commit-Position: refs/heads/master@{#527171}
[modify] https://crrev.com/e286250b9e5c11b0a9e14d6b243dc206999c4a50/third_party/WebKit/LayoutTests/TestExpectations

Apparently this is no longer flaky according to c#12's update-flaky-expectations.