ForceMaximizeOnFirstRunTest.PRE_TwoRuns is flaky in single_process_mash_browser_tests |
|||||
Issue descriptionHit a CHECK in WebContentsObserverSanityChecker::DidStopLoading. Example build: https://ci.chromium.org/p/chromium/builders/luci.chromium.ci/linux-chromeos-rel/17290 Failure log: ==== [22364:22364:1214/124553.601380:FATAL:web_contents_observer_sanity_checker.cc(319)] Check failed: is_loading_. #0 0x55bb714bd65f base::debug::StackTrace::StackTrace() #1 0x55bb71440043 logging::LogMessage::~LogMessage() #2 0x55bb719d1284 content::WebContentsObserverSanityChecker::DidStopLoading() #3 0x55bb701cfc4d content::WebContentsImpl::LoadingStateChanged() #4 0x55bb701dac01 content::WebContentsImpl::DidStopLoading() #5 0x55bb6ff2d430 content::FrameTreeNode::DidStopLoading() #6 0x55bb6ff2d29f content::FrameTreeNode::~FrameTreeNode() #7 0x55bb6ff5777d content::RenderFrameHostImpl::OnRenderProcessGone() #8 0x55bb6ff57625 _ZN3IPC8MessageTI35FrameHostMsg_RenderProcessGone_MetaNSt3__15tupleIJiiEEEvE8DispatchIN7content19RenderFrameHostImplES8_vMS8_FviiEEEbPKNS_7MessageEPT_PT0_PT1_T2_ #9 0x55bb6ff563f6 content::RenderFrameHostImpl::OnMessageReceived() #10 0x55bb700cad3a content::RenderProcessHostImpl::ProcessDied() #11 0x55bb700ca735 content::RenderProcessHostImpl::FastShutdownIfPossible() #12 0x55bb715b90fc browser_shutdown::OnShutdownStarting() #13 0x55bb715b8b0f chrome::ExitCleanly() #14 0x55bb714d2cb5 base::debug::TaskAnnotator::RunTask() #15 0x55bb7144614f base::MessageLoopImpl::RunTask() #16 0x55bb71446633 base::MessageLoopImpl::DoWork() #17 0x55bb714cf1a9 base::MessagePumpLibevent::Run() #18 0x55bb714643f5 base::RunLoop::Run() #19 0x55bb715206c7 InProcessBrowserTest::QuitBrowsers() #20 0x55bb715205ac InProcessBrowserTest::PostRunTestOnMainThread() #21 0x55bb719a3f65 content::BrowserTestBase::ProxyRunTestOnMainThreadLoop() #22 0x55bb7156d5c0 ChromeBrowserMainParts::PreMainMessageLoopRunImpl() #23 0x55bb7156c5a2 ChromeBrowserMainParts::PreMainMessageLoopRun() #24 0x55bb6edb0e02 chromeos::ChromeBrowserMainPartsChromeos::PreMainMessageLoopRun() #25 0x55bb6fe17f82 content::BrowserMainLoop::PreMainMessageLoopRun() #26 0x55bb7019f697 content::StartupTaskRunner::RunAllTasksNow() #27 0x55bb6fe16d7d content::BrowserMainLoop::CreateStartupTasks() #28 0x55bb6fe19db5 content::BrowserMainRunnerImpl::Initialize() #29 0x55bb6fe15282 content::BrowserMain() #30 0x55bb710b26df content::ContentMainRunnerImpl::RunServiceManager() #31 0x55bb710b2369 content::ContentMainRunnerImpl::Run() #32 0x55bb731a6173 service_manager::Main() #33 0x55bb710b0844 content::ContentMain() #34 0x55bb719a39e1 content::BrowserTestBase::SetUp() #35 0x55bb7151ef94 InProcessBrowserTest::SetUp() #36 0x55bb6e8ab27c chromeos::OobeBaseTest::SetUp() #37 0x55bb6e94dae2 policy::LoginPolicyTestBase::SetUp() #38 0x55bb6f255b70 testing::Test::Run() #39 0x55bb6f25690f testing::TestInfo::Run() #40 0x55bb6f256e37 testing::TestCase::Run() #41 0x55bb6f263297 testing::internal::UnitTestImpl::RunAllTests()
,
Dec 17
This has failed twice in ForceMaximizeOnFirstRunTest.PRE_TwoRuns, the run linked and one on 9/28 according to findit: https://findit-for-me.appspot.com/waterfall/list-flakes?test_name=ForceMaximizeOnFirstRunTest.PRE_TwoRuns
,
Dec 17
I found a couple of other examples of failures in tests based on policy::LoginPolicyTestBase, so we might be skipping some shutdown code there: ScreenTimeControllerTest.LockOverride: https://findit-for-me.appspot.com/waterfall/list-flakes?test_name=ScreenTimeControllerTest.LockOverride UserCloudPolicyManagerChildTest.PolicyForChildUser: https://findit-for-me.appspot.com/waterfall/list-flakes?test_name=UserCloudPolicyManagerChildTest.PolicyForChildUser
,
Dec 17
,
Dec 17
Also found a failure in a test based on OobeBaseTest (which LoginPolicyTestBase is based on): https://findit-for-me.appspot.com/waterfall/list-flakes?test_name=EulaTest.LoadOffline
,
Dec 17
,
Dec 17
I am reasonably convinced that these failures are not specific to Single Process Mash, we just happened to see two of them in this test on single_process_mash_browser_tests, probably due to a subtle timing issue. Removing the label and marking Available.
,
Dec 17
,
Dec 17
May also be related to issue 795479
,
Dec 18
Happening again here: https://ci.chromium.org/p/chromium/builders/luci.chromium.ci/linux-chromeos-rel/17459
,
Jan 8
Seems similar to issue 919347 , viz_browser_tests "ShelfAppBrowserTestNoDefaultBrowser.AltNumberBrowserTabbing is flaky" Find-It blamed that bug on *disabling* SingleProcessMash. I suspect this is related to the set of posted tasks in queue during shutdown, so features like SPM and Viz change things.
,
Jan 11
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/51bdda57a8d838112c315e6b95e45b979f41fd6e commit 51bdda57a8d838112c315e6b95e45b979f41fd6e Author: James Cook <jamescook@chromium.org> Date: Fri Jan 11 18:33:58 2019 Remove CHECK in WebContentsObserverSanityChecker::DidStopLoading The check causes low-level browser_test flake across a wide variety of test suites, usually due to check failures during test suite teardown. This is test-only code, so relax the check until the root cause can be investigated further. Bug: 466089, 871010, 913006, 915360, 919347 Test: browser_tests Change-Id: Ife2ab6a6d260471f79873b6f0fd2914a4c7bb71d Reviewed-on: https://chromium-review.googlesource.com/c/1403353 Reviewed-by: Camille Lamy <clamy@chromium.org> Commit-Queue: James Cook <jamescook@chromium.org> Cr-Commit-Position: refs/heads/master@{#622067} [modify] https://crrev.com/51bdda57a8d838112c315e6b95e45b979f41fd6e/content/test/web_contents_observer_sanity_checker.cc
,
Jan 14
|
|||||
►
Sign in to add a comment |
|||||
Comment 1 by steve...@chromium.org
, Dec 17Owner: steve...@chromium.org
Status: Started (was: Untriaged)