The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/0440dc171f05c5354c261c00aa4f6ed866c44214

commit 0440dc171f05c5354c261c00aa4f6ed866c44214
Author: Scott Graham <scottmg@chromium.org>
Date: Fri Oct 20 02:21:16 2017

fuchsia: Disable RenderWidgetHostViewAuraOverscrollTest.* in content_unittests

NOTRY=true

TBR: dtapuska@chromium.org
Bug: 776424
Change-Id: I26211dfcaf3fc20457d9047a8b53624cd05d6abf
Reviewed-on: https://chromium-review.googlesource.com/728321
Commit-Queue: Scott Graham <scottmg@chromium.org>
Reviewed-by: Scott Graham <scottmg@chromium.org>
Cr-Commit-Position: refs/heads/master@{#510299}
[modify] https://crrev.com/0440dc171f05c5354c261c00aa4f6ed866c44214/testing/buildbot/filters/fuchsia.content_unittests.filter

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/f732be588c806b44eab63e1370a7914d61cc7705

commit f732be588c806b44eab63e1370a7914d61cc7705
Author: Scott Graham <scottmg@chromium.org>
Date: Wed Oct 25 05:24:05 2017

fuchsia: Disable some more tests to try to get content_unittests on to waterfall

Bug: 776424, 761214, 777916, 777917
Change-Id: I9906ff9ec53c3afd045ece861f0f9cdfd1a24727
Reviewed-on: https://chromium-review.googlesource.com/733955
Reviewed-by: Wez <wez@chromium.org>
Commit-Queue: Scott Graham <scottmg@chromium.org>
Cr-Commit-Position: refs/heads/master@{#511380}
[modify] https://crrev.com/f732be588c806b44eab63e1370a7914d61cc7705/testing/buildbot/filters/fuchsia.content_unittests.filter

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/b2f8374043f52c891a6927b28b00a1b4dcc02249

commit b2f8374043f52c891a6927b28b00a1b4dcc02249
Author: Scott Graham <scottmg@chromium.org>
Date: Sat Oct 28 00:12:11 2017

fuchsia: Disable yet another flaky RWHVA test

TBR: wez@chromium.org
Bug: 776424
Change-Id: I3d67e9aab9cbeab7296d834001dd456f8890de16
Reviewed-on: https://chromium-review.googlesource.com/742651
Reviewed-by: Scott Graham <scottmg@chromium.org>
Cr-Commit-Position: refs/heads/master@{#512363}
[modify] https://crrev.com/b2f8374043f52c891a6927b28b00a1b4dcc02249/testing/buildbot/filters/fuchsia.content_unittests.filter

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/b8b08a20d346be226261ed758738f4aab5caf5ee

commit b8b08a20d346be226261ed758738f4aab5caf5ee
Author: Wez <wez@chromium.org>
Date: Wed Jan 10 20:02:00 2018

Update Fuchsia filters for content_unittests for new test fixtures.

Several timing-dependent tests were already filtered-out under Fuchsia,
but https://chromium-review.googlesource.com/c/chromium/src/+/851054
added some new variants with names not matching the suppression.

This CL updates the Fuchsia filters and removes the preprocessor
conditional disabling, to make it easier for us to diagnose and fix them
under Fuchsia.

TBR: jochen
Bug:  800771 ,  800415 ,  800841 , 776424
Change-Id: I524b77b65e9550e303c774dd7c1eb8751bbe8039
Reviewed-on: https://chromium-review.googlesource.com/860730
Reviewed-by: Wez <wez@chromium.org>
Reviewed-by: Scott Graham <scottmg@chromium.org>
Commit-Queue: Wez <wez@chromium.org>
Cr-Commit-Position: refs/heads/master@{#528399}
[modify] https://crrev.com/b8b08a20d346be226261ed758738f4aab5caf5ee/content/browser/renderer_host/render_widget_host_view_aura_unittest.cc
[modify] https://crrev.com/b8b08a20d346be226261ed758738f4aab5caf5ee/testing/buildbot/filters/fuchsia.content_unittests.filter

 Issue 800841  has been merged into this issue.

 Issue 800415  has been merged into this issue.

 Issue 800771  has been merged into this issue.

sahel: This is the tracking bug for the RenderWidgetHostViewAura*Tests which we found were flaky under Fuchsia. I think the recent flakes are just due to test names changing, so that our existing filters didn't disable them (see below).

We run Chromium tests on Fuchsia under QEMU, which consumes some CPU cycles for the VM emulation (e.g. emulating hardware devices), potentially leading to a CPU appearing to stall for a little while. Fuchsia itself then has different thread-scheduling properties to typical desktop operating systems, at present. Then recent changes to move more things to TaskScheduler have introduced scope for some work occurring on background-priority worker threads, meaning it may be heavily delayed if the system is under load.

The tests look to use the real TimeTicks clock, which means that any assumptions about e.g. certain tasks occurring within small intervals (e.g. I see time deltas of 17ms in various places in the tests) are especially likely to be invalid on the Fuchsia bots.

Instructions on building & running tests for Fuchsia are here:
https://cs.chromium.org/chromium/src/docs/fuchsia_build_instructions.md

and you can find the current set of content_unittests filters for Fuchsia here:
https://cs.chromium.org/chromium/src/testing/buildbot/filters/fuchsia.content_unittests.filter?l=42

For most of the test flakes we have investigated under Fuchsia, we have been able to reproduce equivalent flakes under Linux/Mac/Win by introducing sleep()s at specific points, FWIW.

 Issue 800877  has been merged into this issue.

 Issue 800923  has been merged into this issue.

 Issue 800946  has been merged into this issue.

sahel: I think we should zero the bug # in the try-flakes tool, so it doesn't re-use the Fuchsia-specific reports for new flakes, once we have the filters landed (which they should be by now - see https://chromium-review.googlesource.com/c/chromium/src/+/860730) - WDYT?

Detected 13 new flakes for test/step "RenderWidgetHostViewAuraTest.SetCanScrollForWebMouseWheelEvent". To see the actual flakes, please visit https://chromium-try-flakes.appspot.com/all_flake_occurrences?key=ahVzfmNocm9taXVtLXRyeS1mbGFrZXNySQsSBUZsYWtlIj5SZW5kZXJXaWRnZXRIb3N0Vmlld0F1cmFUZXN0LlNldENhblNjcm9sbEZvcldlYk1vdXNlV2hlZWxFdmVudAw. This message was posted automatically by the chromium-try-flakes app. Since flakiness is ongoing, the issue was moved back into Sheriff Bug Queue (unless already there).

re #12: So far all the flakes that I've seen for RenderWidgetHostViewAura* content_unittests are on Fuchsia, do you have any counter examples on mind? if not, I am not sure how we can benefit from setting the bug number to zero.

Re #14: Now that these tests are filtered-out, they should no longer be running on Fuchsia, so any new flakes are presumably some other issue, is my rationale. :)

https://chromium-review.googlesource.com/c/chromium/src/+/861864 clears the filters for the following tests and fixes the flaky tests:
RenderWidgetHostViewAuraOverscrollTest.*
RenderWidgetHostViewAuraOverscrollWithoutWheelScrollLatchingTest.*
RenderWidgetHostViewAuraOverScrollAsyncWheelEventsEnabledTest.*


I'd rather to wait for one day and see if the flakes happen again or not, and if not then reset the bug number, WDYT?

That sounds great to me - thanks for the quick fix.

Removing the label since an owner looks assigned.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/f3a04acf71e4d53036200acce97157861b9f107c

commit f3a04acf71e4d53036200acce97157861b9f107c
Author: Sahel Sharify <sahel@chromium.org>
Date: Mon Jan 15 17:57:03 2018

Fix flaky content_unittests on Fuchsia

This cl fixes the following flaky tests on Fuchsia:
RenderWidgetHostViewAuraOverscrollTest.*
RenderWidgetHostViewAuraOverscrollWithoutWheelScrollLatchingTest.*
RenderWidgetHostViewAuraOverScrollAsyncWheelEventsEnabledTest.*

Bug: 776424
Change-Id: I6e83584cd3d86c649960ec9b1dad79ac20934f87
Reviewed-on: https://chromium-review.googlesource.com/861864
Reviewed-by: Wez <wez@chromium.org>
Reviewed-by: Timothy Dresser <tdresser@chromium.org>
Commit-Queue: Sahel Sharifymoghaddam <sahel@chromium.org>
Cr-Commit-Position: refs/heads/master@{#529309}
[modify] https://crrev.com/f3a04acf71e4d53036200acce97157861b9f107c/content/browser/renderer_host/render_widget_host_view_aura_unittest.cc
[modify] https://crrev.com/f3a04acf71e4d53036200acce97157861b9f107c/testing/buildbot/filters/fuchsia.content_unittests.filter

We appear to still be getting flakes on Fuchsia/x64 FYI:

from https://ci.chromium.org/buildbot/chromium.fyi/Fuchsia/13042:
[ RUN      ] RenderWidgetHostViewAuraOverscrollTest.GestureScrollDebounceTimerOverscroll
../../content/browser/renderer_host/render_widget_host_view_aura_unittest.cc:4957: Failure
Expected equality of these values:
  0U
    Which is: 0
  GetAndResetDispatchedMessages().size()
    Which is: 1
../../content/browser/renderer_host/render_widget_host_view_aura_unittest.cc:4961: Failure
Expected equality of these values:
  OVERSCROLL_EAST
    Which is: 4
  overscroll_mode()
    Which is: 0
../../content/browser/renderer_host/render_widget_host_view_aura_unittest.cc:4962: Failure
Expected equality of these values:
  OverscrollSource::TOUCHSCREEN
    Which is: 4-byte object <02-00 00-00>
  overscroll_source()
    Which is: 4-byte object <00-00 00-00>
../../content/browser/renderer_host/render_widget_host_view_aura_unittest.cc:4963: Failure
Expected equality of these values:
  OVERSCROLL_EAST
    Which is: 4
  overscroll_delegate()->current_mode()
    Which is: 0
../../content/browser/renderer_host/render_widget_host_view_aura_unittest.cc:4964: Failure
Expected equality of these values:
  55.f
    Which is: 55
  overscroll_delta_x()
    Which is: 0
../../content/browser/renderer_host/render_widget_host_view_aura_unittest.cc:4965: Failure
Expected equality of these values:
  5.f
    Which is: 5
  overscroll_delegate()->delta_x()
    Which is: 0
../../content/browser/renderer_host/render_widget_host_view_aura_unittest.cc:4980: Failure
Expected equality of these values:
  "GestureScrollEnd"
  GetMessageNames(events)
    Which is: ""
[  FAILED  ] RenderWidgetHostViewAuraOverscrollTest.GestureScrollDebounceTimerOverscroll (239 ms)

and from RenderWidgetHostViewAuraOverscrollTest.TouchGestureEndDispatchedAfterOverscrollComplete:

[ RUN      ] RenderWidgetHostViewAuraOverscrollTest.TouchGestureEndDispatchedAfterOverscrollComplete
../../content/browser/renderer_host/render_widget_host_view_aura_unittest.cc:5187: Failure
Expected equality of these values:
  0U
    Which is: 0
  events.size()
    Which is: 1
../../content/browser/renderer_host/render_widget_host_view_aura_unittest.cc:5195: Failure
Expected equality of these values:
  "GestureScrollEnd"
  GetMessageNames(events)
    Which is: ""
[  FAILED  ] RenderWidgetHostViewAuraOverscrollTest.TouchGestureEndDispatchedAfterOverscrollComplete (344 ms)

I saw the build history after 15th of jan, looks like RenderWidgetHostViewAuraOverscrollTest.TouchGestureEndDispatchedAfterOverscrollComplete and RenderWidgetHostViewAuraOverscrollTest.GestureScrollDebounceTimerOverscroll are still flaky.. , I will take a look.

 Issue 802375  has been merged into this issue.

 Issue 802389  has been merged into this issue.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/21ee38c2e8ebd4f0646552e3face31472810d60d

commit 21ee38c2e8ebd4f0646552e3face31472810d60d
Author: Wez <wez@chromium.org>
Date: Tue Jan 16 22:36:12 2018

Filter out RenderWidgetHostViewAuraOverscrollTest.* under Fuchsia.

These tests are currently flaky under Fuchsia bots.

Bug:  802375 , 776424
Change-Id: I7d7e6d2a5d2392eddf84cb35ef698022f673622b
Reviewed-on: https://chromium-review.googlesource.com/868935
Commit-Queue: Wez <wez@chromium.org>
Reviewed-by: Scott Graham <scottmg@chromium.org>
Cr-Commit-Position: refs/heads/master@{#529511}
[modify] https://crrev.com/21ee38c2e8ebd4f0646552e3face31472810d60d/testing/buildbot/filters/fuchsia.content_unittests.filter

Detected 4 new flakes for test/step "RenderWidgetHostViewAuraAsyncWheelEventsEnabledTest.TimerBasedWheelEventPhaseInfo". To see the actual flakes, please visit https://chromium-try-flakes.appspot.com/all_flake_occurrences?key=ahVzfmNocm9taXVtLXRyeS1mbGFrZXNyXAsSBUZsYWtlIlFSZW5kZXJXaWRnZXRIb3N0Vmlld0F1cmFBc3luY1doZWVsRXZlbnRzRW5hYmxlZFRlc3QuVGltZXJCYXNlZFdoZWVsRXZlbnRQaGFzZUluZm8M. This message was posted automatically by the chromium-try-flakes app. Since flakiness is ongoing, the issue was moved back into Sheriff Bug Queue (unless already there).

Failures in the flake report from #25 are:

[ RUN      ] RenderWidgetHostViewAuraAsyncWheelEventsEnabledTest.TimerBasedWheelEventPhaseInfo
../../content/browser/renderer_host/render_widget_host_view_aura_unittest.cc:1794: Failure
Expected equality of these values:
  WebMouseWheelEvent::kPhaseChanged
    Which is: 4
  wheel_event->phase
    Which is: 1
which then crashes with:
#01: operator-> at ../../../c/b/linux/src/buildtools/third_party/libc++/trunk/include/memory:2515
      (inlined by) Allocate at ../../../c/b/linux/src/cc/base/list_container_helper.cc:458
#02: AllocateAndConstruct<viz::SharedQuadState> at ../../../c/b/linux/src/cc/base/list_container.h:104
      (inlined by) CreateAndAppendSharedQuadState at ../../../c/b/linux/src/components/viz/common/quads/render_pass.cc:260
#03: AppendQuads at ../../../c/b/linux/src/cc/layers/picture_layer_impl.cc:254
#04: TimerBasedWheelEventPhaseInfo at /b/c/b/linux/src/out/Release/../../buildtools/third_party/libc++/trunk/include/memory:?
#05: os_stack_trace_getter at ../../../c/b/linux/src/third_party/googletest/src/googletest/src/gtest.cc:4935
      (inlined by) Run at ../../../c/b/linux/src/third_party/googletest/src/googletest/src/gtest.cc:2480
#06: os_stack_trace_getter at ../../../c/b/linux/src/third_party/googletest/src/googletest/src/gtest.cc:4935
      (inlined by) Run at ../../../c/b/linux/src/third_party/googletest/src/googletest/src/gtest.cc:2655
#07: Run at ../../../c/b/linux/src/third_party/googletest/src/googletest/src/gtest.cc:2768
#08: RunAllTests at ../../../c/b/linux/src/third_party/googletest/src/googletest/src/gtest.cc:4664
#09: Run at ../../../c/b/linux/src/third_party/googletest/src/googletest/src/gtest.cc:4277
#10: RUN_ALL_TESTS at ../../../c/b/linux/src/third_party/googletest/src/googletest/include/gtest/gtest.h:2233
      (inlined by) Run at ../../../c/b/linux/src/base/test/test_suite.cc:272
#11: Run at ../../../c/b/linux/src/base/callback.h:94
      (inlined by) LaunchUnitTestsInternal at ../../../c/b/linux/src/base/test/launcher/unit_test_launcher.cc:220
#12: LaunchUnitTests at ../../../c/b/linux/src/base/test/launcher/unit_test_launcher.cc:558
#13: main at ../../../c/b/linux/src/content/test/run_all_unittests.cc:20
#14: pc 0x6cd17a52814e sp 0x5e1f81e62fe0 (libc.so,0x1b14e)
#15: pc 0 sp 0x5e1f81e63000

and

[ RUN      ] RenderWidgetHostViewAuraWheelScrollLatchingEnabledTest.TimerBasedWheelEventPhaseInfo
../../content/browser/renderer_host/render_widget_host_view_aura_unittest.cc:1794: Failure
Expected equality of these values:
  WebMouseWheelEvent::kPhaseChanged
    Which is: 4
  wheel_event->phase
    Which is: 8
which crashes with:
#01: operator-> at ../../../c/b/linux/src/buildtools/third_party/libc++/trunk/include/memory:2515
      (inlined by) TimerBasedWheelEventPhaseInfo at ../../../c/b/linux/src/content/browser/renderer_host/render_widget_host_view_aura_unittest.cc:1799
#02: os_stack_trace_getter at ../../../c/b/linux/src/third_party/googletest/src/googletest/src/gtest.cc:4935
      (inlined by) Run at ../../../c/b/linux/src/third_party/googletest/src/googletest/src/gtest.cc:2480
#03: os_stack_trace_getter at ../../../c/b/linux/src/third_party/googletest/src/googletest/src/gtest.cc:4935
      (inlined by) Run at ../../../c/b/linux/src/third_party/googletest/src/googletest/src/gtest.cc:2655
#04: Run at ../../../c/b/linux/src/third_party/googletest/src/googletest/src/gtest.cc:2768
#05: RunAllTests at ../../../c/b/linux/src/third_party/googletest/src/googletest/src/gtest.cc:4664
#06: Run at ../../../c/b/linux/src/third_party/googletest/src/googletest/src/gtest.cc:4277
#07: RUN_ALL_TESTS at ../../../c/b/linux/src/third_party/googletest/src/googletest/include/gtest/gtest.h:2233
      (inlined by) Run at ../../../c/b/linux/src/base/test/test_suite.cc:272
#08: Run at ../../../c/b/linux/src/base/callback.h:94
      (inlined by) LaunchUnitTestsInternal at ../../../c/b/linux/src/base/test/launcher/unit_test_launcher.cc:220
#09: LaunchUnitTests at ../../../c/b/linux/src/base/test/launcher/unit_test_launcher.cc:558
#10: main at ../../../c/b/linux/src/content/test/run_all_unittests.cc:20
#11: pc 0x64a45242b14e sp 0x431c7789efe0 (libc.so,0x1b14e)
#12: pc 0 sp 0x431c7789f000

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/525fac0c8be86c97eec45b7b58a0b602f523306b

commit 525fac0c8be86c97eec45b7b58a0b602f523306b
Author: Wez <wez@chromium.org>
Date: Fri Jan 19 22:57:57 2018

Filter some more flaky RenderWidgetHostViewAura*Test tests.

These tests flake under our Fuchsia/x64 try-bots.

Bug: 776424
Change-Id: I706ed77b4a2e882c9f1f3d042fc11677ceee440e
Reviewed-on: https://chromium-review.googlesource.com/876126
Commit-Queue: Wez <wez@chromium.org>
Commit-Queue: Scott Graham <scottmg@chromium.org>
Reviewed-by: Scott Graham <scottmg@chromium.org>
Cr-Commit-Position: refs/heads/master@{#530645}
[modify] https://crrev.com/525fac0c8be86c97eec45b7b58a0b602f523306b/testing/buildbot/filters/fuchsia.content_unittests.filter

 Issue 818299  has been merged into this issue.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/622bc0ed5af4f6097eafd63059347e6f3fb531ae

commit 622bc0ed5af4f6097eafd63059347e6f3fb531ae
Author: Wez <wez@chromium.org>
Date: Fri Apr 06 23:54:59 2018

Filter GestureScrollDebounceOverscrolls test under Fuchsia.

Many of the tests in this suite have fragile timing dependencies, so
they flake, especially when run under emulation.

Bug:  818299 , 776424
Change-Id: I2fe239e2638a5a7944a38b65b27534d917d5636e
Reviewed-on: https://chromium-review.googlesource.com/1000716
Commit-Queue: Wez <wez@chromium.org>
Reviewed-by: Scott Graham <scottmg@chromium.org>
Cr-Commit-Position: refs/heads/master@{#548977}
[modify] https://crrev.com/622bc0ed5af4f6097eafd63059347e6f3fb531ae/testing/buildbot/filters/fuchsia.content_unittests.filter

Observed RenderWidgetHostViewAuraTest.SyntheticFlingCancelAtTouchpadScrollBegin
 flake on Fuchsia/ARM64/Rel in https://ci.chromium.org/p/chromium/builders/luci.chromium.ci/fuchsia-fyi-arm64-rel/477

Note that that bot is running content_unittests under Fuchsia under QEMU w/ software ARM64 emulation, so certain things are disproportionately slow. This test completes in ~1.5s when it fails, and ~1.1s when it passes.

Test in #35 was added in https://chromium-review.googlesource.com/1138637 by mcnee@ on July 19th, and has actually been flaking on Fuchsia/x64, not Fuchsia/ARM64.

So it looks like a lot of the flakiness here involves timing based wheel event phase. I would suspect this is the cause of flakiness in SyntheticFlingCancelAtTouchpadScrollBegin as well.

I see that the first failure is
../../content/browser/renderer_host/render_widget_host_view_aura_unittest.cc:2485: Failure
Expected equality of these values:
  0U
    Which is: 0
  dispatched_events.size()
    Which is: 1

...which indicates we have an unexpected event in the middle of the scroll sequence. Perhaps this is a wheel event with phase end?

To address the flakiness due to wheel timer-based phase, the timeout of timer is set to a large value: e.g. https://cs.chromium.org/chromium/src/content/browser/renderer_host/render_widget_host_view_aura_unittest.cc?q=RenderWidgetHostViewAuraOverscrollTest&g=0&l=1943
The same logic can be used in the new test to avoid flakiness.

The remaining flaky tests are due to debounce queue timer rather than wheel phase handler timer.

Re #38: Should the longer timeout be added to individual tests, or to the test base-class?  

The default timeout is 500ms, in the test set up I set this to 100ms (https://cs.chromium.org/chromium/src/content/browser/renderer_host/render_widget_host_view_aura_unittest.cc?type=cs&sq=package:chromium&g=0&l=566). 
I remember I did this since some of the tests needed to wait for the timer to expire and I did not want to slow them down. Then in tests that an active timer was required, I set them to the max timeout. However I briefly checked the tests now and I could not find a test that is waiting for timer expiration.

I will double check the tests, and submit a patch to set the timer to max in set up if it doesn't break anything.

Whenever we are using timers we risk having flaky tests. I generally recommend for unit tests that we mock out any timers/time sources. Especially when we want to guarantee that a timer hasn't expired while we've done something else.

It sounds like the tests might want to mock out MouseWheelPhaseHandler::mouse_wheel_end_dispatch_timer_ with one where the test case controls when it actually fires.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/be677c94d91e9b092e97365c14344b2a2ed9d6d9

commit be677c94d91e9b092e97365c14344b2a2ed9d6d9
Author: Wez <wez@chromium.org>
Date: Fri Jul 27 21:29:25 2018

Filter some delicious buttery, crumbly, flaky tests.

Add the following tests to the Fuchsia filter for content_unittests:

- BluetoothBlocklistTest.InvalidUUID
- RenderWidgetHostViewAuraTest.SyntheticFlingCancelAtTouchpadScrollBegin
- WebBluetoothDeviceIdTest.DefaultConstructor

Bug: 776424, 867502
Change-Id: I7e588b3946e640cb77b46cace1ef8ab79385f5b8
Reviewed-on: https://chromium-review.googlesource.com/1150803
Commit-Queue: Sergey Ulanov <sergeyu@chromium.org>
Reviewed-by: Sergey Ulanov <sergeyu@chromium.org>
Cr-Commit-Position: refs/heads/master@{#578804}
[modify] https://crrev.com/be677c94d91e9b092e97365c14344b2a2ed9d6d9/testing/buildbot/filters/fuchsia.content_unittests.filter