Detected 3 new flakes for test/step "WebMediaPlayerMSTest.CreateHardwareFrames". To see the actual flakes, please visit https://chromium-try-flakes.appspot.com/all_flake_occurrences?key=ahVzfmNocm9taXVtLXRyeS1mbGFrZXNyNAsSBUZsYWtlIilXZWJNZWRpYVBsYXllck1TVGVzdC5DcmVhdGVIYXJkd2FyZUZyYW1lcww. 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).

[ RUN      ] WebMediaPlayerMSTest.CreateHardwareFrames
[233263:1394281871:0312/225157.871864:335304965:ERROR:base_paths_fuchsia.cc(37)] Not implemented reached in bool base::PathProviderFuchsia(int, base::FilePath *)
../../content/renderer/media/stream/webmediaplayer_ms_unittest.cc:1075: Failure
Actual function call count doesn't match EXPECT_CALL(*this, DoStartRendering())...
         Expected: to be called once
           Actual: never called - unsatisfied and active
../../content/renderer/media/stream/webmediaplayer_ms_unittest.cc:1074: Failure
Actual function call count doesn't match EXPECT_CALL(*this, DoSetWebLayer(true))...
         Expected: to be called once
           Actual: never called - unsatisfied and active
../../content/renderer/media/stream/webmediaplayer_ms_unittest.cc:1077: Failure
Actual function call count doesn't match EXPECT_CALL(*this, DoReadyStateChanged( blink::WebMediaPlayer::kReadyStateHaveMetadata))...
         Expected: to be called once
           Actual: never called - unsatisfied and active
../../content/renderer/media/stream/webmediaplayer_ms_unittest.cc:1079: Failure
Actual function call count doesn't match EXPECT_CALL(*this, DoReadyStateChanged( blink::WebMediaPlayer::kReadyStateHaveEnoughData))...
         Expected: to be called once
           Actual: never called - unsatisfied and active
../../content/renderer/media/stream/webmediaplayer_ms_unittest.cc:1081: Failure
Actual function call count doesn't match EXPECT_CALL(*this, CheckSizeChanged(gfx::Size(kStandardWidth, kStandardHeight)))...
         Expected: to be called once
           Actual: never called - unsatisfied and active
../../content/renderer/media/stream/webmediaplayer_ms_unittest.cc:1091: Failure
Actual function call count doesn't match EXPECT_CALL(*this, DoStopRendering())...
         Expected: to be called once
           Actual: never called - unsatisfied and active
[  FAILED  ] WebMediaPlayerMSTest.CreateHardwareFrames (230 ms

disabling on fuscia^H^H^H^H^H^Hpink in https://chromium-review.googlesource.com/#/c/chromium/src/+/958666

(scottmg: I picked you in case this is related to Issue 766356)

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/a3f0e87ea2bf994888abfc988abc7d2587bc60d2

commit a3f0e87ea2bf994888abfc988abc7d2587bc60d2
Author: Trent Apted <tapted@chromium.org>
Date: Tue Mar 13 03:49:47 2018

Disable WebMediaPlayerMSTest.CreateHardwareFrames on Fushsia

Flaky.

TBR=scottmg@chromium.org

Bug:  819924 
Change-Id: Id6a337392ec94bbc1b3ae2f83a914bc4d2565996
Reviewed-on: https://chromium-review.googlesource.com/958666
Reviewed-by: Trent Apted <tapted@chromium.org>
Commit-Queue: Trent Apted <tapted@chromium.org>
Cr-Commit-Position: refs/heads/master@{#542722}
[modify] https://crrev.com/a3f0e87ea2bf994888abfc988abc7d2587bc60d2/testing/buildbot/filters/fuchsia.content_unittests.filter

WebMediaPlayerMSTest.StopsCreatingHardwareFramesWhenHiddenOrClosed  failed in https://build.chromium.org/p/chromium.fyi/builders/Fuchsia/builds/14779 with a crash:

[00332.786] 01197.01240> <== general fault, PC at 0x552e4c0e4907
...
#01: content::MockMediaStreamVideoRenderer::InjectFrame() at ??:?
#02: base::debug::TaskAnnotator::RunTask(char const*, base::PendingTask*) at ??:?
#03: base::MessageLoop::RunTask(base::PendingTask*) at ??:?
#04: base::MessageLoop::DoDelayedWork(base::TimeTicks*) at ??:?
#05: base::MessagePumpDefault::Run(base::MessagePump::Delegate*) at ??:?
#06: base::RunLoop::RunUntilIdle() at ??:?
#07: content::WebMediaPlayerMSTest::~WebMediaPlayerMSTest() at ??:?
#08: content::WebMediaPlayerMSTest_OpacityChange_Test::~WebMediaPlayerMSTest_OpacityChange_Test() at ??:?
#09: testing::TestInfo::Run() at ??:?
#10: testing::TestCase::Run() at ??:?
#11: testing::internal::UnitTestImpl::RunAllTests() at ??:?
#12: testing::UnitTest::Run() at ??:?
#13: base::TestSuite::Run() at ??:?
#14: base::(anonymous namespace)::LaunchUnitTestsInternal(base::OnceCallback<int ()>, unsigned long, int, bool, base::OnceCallback<void ()>) at ??:?
#15: base::LaunchUnitTests(int, char**, base::OnceCallback<int ()>) at ??:?
#16: main at ??:?
#17: pc 0x59e1984f661e sp 0x147ec5a55fe0 (libc.so,0x1361e)
#18: pc 0 sp 0x147ec5a56000

Will broaden the suppression for this test suite.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/00eae0d98fb6a1f83aeaa901b080f7543d43318d

commit 00eae0d98fb6a1f83aeaa901b080f7543d43318d
Author: Wez <wez@chromium.org>
Date: Wed Mar 14 01:43:55 2018

Filter a new WebMediaPlayerMSTest test under Fuchsia.

The new StopsCreatingHardwareFramesWhenHiddenOrClosed crashes under
Fuchsia, so filter it for now.

TBR: sergeyu
Bug:  819924 
Change-Id: Iecec7d2dae681af105e2c9adde84e120343317bd
Reviewed-on: https://chromium-review.googlesource.com/961735
Reviewed-by: Wez <wez@chromium.org>
Commit-Queue: Wez <wez@chromium.org>
Cr-Commit-Position: refs/heads/master@{#542981}
[modify] https://crrev.com/00eae0d98fb6a1f83aeaa901b080f7543d43318d/testing/buildbot/filters/fuchsia.content_unittests.filter

WebMediaPlayerMSTest.RotationChange failed in https://ci.chromium.org/buildbot/chromium.fyi/Fuchsia/15151 with the following:

[00328.633] 01201.01235> <== fatal exception: process content_unittests__exec[224715] thread initial-thread[224778]
...
#01: content::MockMediaStreamVideoRenderer::InjectFrame() at ??:?
#02: base::debug::TaskAnnotator::RunTask(char const*, base::PendingTask*) at ??:?
#03: base::MessageLoop::RunTask(base::PendingTask*) at ??:?
#04: base::MessageLoop::DoDelayedWork(base::TimeTicks*) at ??:?
#05: base::MessagePumpDefault::Run(base::MessagePump::Delegate*) at ??:?
#06: base::RunLoop::RunUntilIdle() at ??:?
#07: content::WebMediaPlayerMSTest::~WebMediaPlayerMSTest() at ??:?
#08: content::WebMediaPlayerMSTest_OpacityChange_Test::~WebMediaPlayerMSTest_OpacityChange_Test() at ??:?
#09: testing::TestInfo::Run() at ??:?
#10: testing::TestCase::Run() at ??:?
#11: testing::internal::UnitTestImpl::RunAllTests() at ??:?
#12: testing::UnitTest::Run() at ??:?
#13: base::TestSuite::Run() at ??:?
#14: base::(anonymous namespace)::LaunchUnitTestsInternal(base::OnceCallback<int ()>, unsigned long, int, bool, base::OnceCallback<void ()>) at ??:?
#15: base::LaunchUnitTests(int, char**, base::OnceCallback<int ()>) at ??:?
#16: main at ??:?
#17: pc 0x40cf83c8d66e sp 0x61666bc2cfe0 (libc.so,0x1366e)
#18: pc 0 sp 0x61666bc2d000

Notice that both stacks take place during teardown of the OpacityChange test; the crash may be getting mis-attributed by the TestLauncher e.g. due to truncated output from the crashing batch process.

The InjectFrame() function injects a frame and then posts a delayed task to inject the next frame (see https://cs.chromium.org/chromium/src/content/renderer/media/stream/webmediaplayer_ms_unittest.cc?l=349). The posted task uses base::Unretained(). It's not clear whether that is intentional; the class itself is ref-counted, but Unretained() avoids the test itself affecting the ref-counting of implementation code.  Some of the tests have a pending delayed task when they exit (i.e. they have scheduled more frames to be injected than the test actually waits for) - the tests only pass (most of the time) because the delayed task is silently dropped when the test instance' message queue is torn down.

Applying a simple Sleep() during the test teardown can provoke this crash under other platforms, e.g. Linux:

--- a/content/renderer/media/stream/webmediaplayer_ms_unittest.cc
+++ b/content/renderer/media/stream/webmediaplayer_ms_unittest.cc
@@ -469,6 +469,7 @@ class WebMediaPlayerMSTest
         rendering_(false),
         background_rendering_(false) {}
   ~WebMediaPlayerMSTest() override {
+    base::PlatformThread::Sleep(base::TimeDelta::FromMilliseconds(500));
     player_.reset();
     base::RunLoop().RunUntilIdle();
   }

This appears to affect the RotationChange, OpacityChange and StopsCreatingHardwareFramesWhenHiddenOrClosed tests.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/33d4d98b9d6ff6b6be10fb414bda6e3b20d118d3

commit 33d4d98b9d6ff6b6be10fb414bda6e3b20d118d3
Author: Wez <wez@chromium.org>
Date: Fri Mar 23 19:45:16 2018

Fix WebMediaPlayerMSTest.* not to use-after-free on slow systems.

These tests were posting delayed tasks to inject video frames, with raw
pointers to the target object. The tests would usually pass because the
delayed task would be silently dropped during MessageLoop teardown, but
on slow systems (e.g. under QEMU) these delayed tasks could end up
being executed during the teardown RunUntilIdle(), after the target
had been torn down.

The target is a ref-counted object, so post delayed tasks to it without
using Unretained(), to ensure it remains valid.

Bug:  819924 
Change-Id: I5a781cbaab1519b5707967753decf2672ba826d5
Reviewed-on: https://chromium-review.googlesource.com/978637
Reviewed-by: Emircan Uysaler <emircan@chromium.org>
Commit-Queue: Wez <wez@chromium.org>
Cr-Commit-Position: refs/heads/master@{#545559}
[modify] https://crrev.com/33d4d98b9d6ff6b6be10fb414bda6e3b20d118d3/content/renderer/media/stream/webmediaplayer_ms_unittest.cc
[modify] https://crrev.com/33d4d98b9d6ff6b6be10fb414bda6e3b20d118d3/testing/buildbot/filters/fuchsia.content_unittests.filter

Flaked again in https://ci.chromium.org/p/chromium/builders/luci.chromium.ci/fuchsia-fyi-arm64-rel/496 with:

[ RUN      ] WebMediaPlayerMSTest.CreateHardwareFrames
../../content/renderer/media/stream/webmediaplayer_ms_unittest.cc:1137: Failure
Value of: frame != nullptr
  Actual: false
Expected: true
Stack trace:
#00: testing::internal::AssertHelper::operator=(testing::Message const&) const at gtest.cc:?
#01: content::WebMediaPlayerMSTest_CreateHardwareFrames_Test::TestBody() at webmediaplayer_ms_unittest.cc:?

The test runs in 175ms, so this doesn't look like a timeout issue.

emircan: Under what circumstances would we get the assertion in #15 failing?

Flaked again, in https://ci.chromium.org/p/chromium/builders/luci.chromium.ci/fuchsia-fyi-arm64-rel/578 with:

[ RUN      ] WebMediaPlayerMSTest.CreateHardwareFrames
../../content/renderer/media/stream/webmediaplayer_ms_unittest.cc:1131: Failure
Actual function call count doesn't match EXPECT_CALL(*this, CheckSizeChanged(provider->get_standard_size()))...
         Expected: to be called once
           Actual: never called - unsatisfied and active
Stack trace:
#00: testing::internal::AssertHelper::operator=(testing::Message const&) const at gtest.cc:?
#01: testing::internal::GoogleTestFailureReporter::ReportFailure(testing::internal::FailureReporterInterface::FailureType, char const*, int, std::__2::basic_string<char, std::__2::char_traits<char>, std::__2::allocator<char> > const&) at gmock-internal-utils.cc:?
#02: testing::internal::UntypedFunctionMockerBase::VerifyAndClearExpectationsLocked() at gmock-spec-builders.cc:?
#03: testing::Mock::VerifyAndClearExpectationsLocked(void*) at gmock-spec-builders.cc:?
#04: testing::Mock::VerifyAndClearExpectations(void*) at gmock-spec-builders.cc:?
#05: content::WebMediaPlayerMSTest_CreateHardwareFrames_Test::TestBody() at webmediaplayer_ms_unittest.cc:?

[  FAILED  ] WebMediaPlayerMSTest.CreateHardwareFrames (279 ms)