WebRtcTwoDeviceDepthCaptureBrowserTest's flaky in asan-clang-phone |
|||||||||
Issue descriptionThe following tests have recently started showing flaky crashes on asan-clang-phone: test build status WebRtcTwoDeviceDepthCaptureBrowserTest.GetBothStreamsAndCheckForFeaturesPresence (CRASHED) 2850 --F-~F~-~-------~~--~---~ WebRtcTwoDeviceDepthCaptureBrowserTest.GetStreamsByVideoKind (CRASHED) 2850 -F-F~-~-~-------~~--~---~ Build: https://uberchromegw.corp.google.com/i/internal.client.clank/builders/asan-clang-phone/builds/2849 Log: https://luci-logdog.appspot.com/v/?s=chrome%2Fbb%2Finternal.client.clank%2Fasan-clang-phone%2F2849%2F%2B%2Frecipes%2Fsteps%2Fcontent_browsertests%2F0%2Fstdout Build: https://uberchromegw.corp.google.com/i/internal.client.clank/builders/asan-clang-phone/builds/2848 Log: https://luci-logdog.appspot.com/v/?s=chrome%2Fbb%2Finternal.client.clank%2Fasan-clang-phone%2F2848%2F%2B%2Frecipes%2Fsteps%2Fcontent_browsertests%2F0%2Fstdout From one of the logs: C 6820.021s Main Detailed Logs C 6820.021s Main ******************************************************************************** C 6820.025s Main [CRASH] WebRtcTwoDeviceDepthCaptureBrowserTest.GetStreamsByVideoKind: C 6820.025s Main [ RUN ] WebRtcTwoDeviceDepthCaptureBrowserTest.GetStreamsByVideoKind C 6820.025s Main [WARNING:dns_config_service_posix.cc(326)] Failed to read DnsConfig. C 6820.025s Main [ERROR:devtools_http_handler.cc(226)] Cannot start http server for devtools. Stop devtools. C 6820.025s Main [ERROR:child_process_launcher_helper_posix.cc(118)] Ignoring invalid file assets/snapshot_blob_64.bin C 6820.025s Main [WARNING:simple_synchronous_entry.cc(1102)] Could not open platform files for entry. C 6820.025s Main [INFO:CONSOLE(140)] "Calling getStreamsByVideoKind", source: http://127.0.0.1:43998/media/getusermedia-depth-capture.html (140) C 6820.025s Main [INFO:SkFontMgr_android_parser.cpp(609)] [SkFontMgr Android Parser] '/system/etc/fonts.xml' could not be opened C 6820.025s Main C 6820.025s Main [INFO:CONSOLE(29)] "Test Success", source: http://127.0.0.1:43998/media/webrtc_test_utilities.js (29) C 6820.025s Main [INFO:SkFontMgr_android_parser.cpp(609)] [SkFontMgr Android Parser] '/vendor/etc/fallback_fonts.xml' could not be opened C 6820.025s Main C 6820.025s Main [FATAL:interface_endpoint_client.cc(33)] Check failed: !is_valid. The callback passed to ImageCapture::GetCapabilities() was never run. C 6820.025s Main #00 0x7a777501 /data/app-lib/org.chromium.content_browsertests_apk-1/libbase.cr.so+0x00173501 C 6820.025s Main #01 0x7bbaf515 /data/app-lib/org.chromium.content_browsertests_apk-1/libbindings.cr.so+0x00040515 C 6820.025s Main #02 0x7bbaebfd /data/app-lib/org.chromium.content_browsertests_apk-1/libbindings.cr.so+0x0003fbfd C 6820.025s Main #03 0x909158eb /data/app-lib/org.chromium.content_browsertests_apk-1/libcontent.cr.so+0x0230b8eb [...]
,
Jun 6 2017
..sorry - intended to link to the build it failed at https://uberchromegw.corp.google.com/i/internal.client.clank/builders/asan-clang-phone/builds/2875
,
Jun 6 2017
+ mcasas@ It seems this happens during shutdown. I wonder if this could be related to the Browser main loop shutdown sequence not being aware of and not waiting for video capture internal threads still doing work, see issue 725271 .
,
Jun 19 2017
Issue 734558 has been merged into this issue.
,
Jun 19 2017
Test is there for since 02.2017. I understand that it got flaky recently. johnme@ submitted a skip-workaround https://crrev.com/2944003002 for Android. I'm not able to access the https://uberchromegw.corp.google.com/* links, and don't see any recent change in master that could trigger this. Will try to reproduce it locally.
,
Jun 19 2017
> I'm not able to access the https://uberchromegw.corp.google.com/* links Yeah, unfortunately this is a downstream bot, so the usual trick of replacing https://uberchromegw.corp.google.com/i/ with https://build.chromium.org/p/ doesn't work here :-| The log output and stacktrace got posted above though, so you're not missing much. There's another flaky crash in https://luci-logdog.appspot.com/v/?s=chromium%2Fbb%2Fchromium.fyi%2FChromium_Mac_10.11%2F9232%2F%2B%2Frecipes%2Fsteps%2Fcontent_browsertests%2F0%2Fstdout which is publicly accessible, though not sure if that one's related to this issue: [ RUN ] WebRtcTwoDeviceDepthCaptureBrowserTest.GetBothStreamsAndCheckForFeaturesPresence [18242:775:0616/134722.632880:12410115371076:INFO:CONSOLE(57)] "Calling getBothStreamsAndCheckForFeaturesPresence", source: http://127.0.0.1:63060/media/getusermedia-depth-capture.html (57) [18242:775:0616/134722.652747:12410135218941:INFO:CONSOLE(29)] "Test Success", source: http://127.0.0.1:63060/media/webrtc_test_utilities.js (29) [18242:775:0616/134722.659268:12410141738350:WARNING:ipc_message_attachment_set.cc(49)] MessageAttachmentSet destroyed with unconsumed attachments: 0/1 [18242:775:0616/134722.678236:12410160708465:ERROR:display_link_mac.cc(184)] CVDisplayLinkStop failed: -6672 BrowserTestBase received signal: Segmentation fault: 11. Backtrace: 0 content_browsertests 0x00000001066e210c base::debug::StackTrace::StackTrace(unsigned long) + 28 1 content_browsertests 0x00000001063936c8 content::(anonymous namespace)::DumpStackTraceSignalHandler(int) + 200 2 libsystem_platform.dylib 0x00007fffad208bba _sigtramp + 26 3 ??? 0x0000000000000016 0x0 + 22 4 content_browsertests 0x000000010731e02e ui::DisplayLinkMac::~DisplayLinkMac() + 14 5 content_browsertests 0x000000010581cbff content::RenderWidgetHostViewMac::~RenderWidgetHostViewMac() + 351 6 content_browsertests 0x000000010581cdfe content::RenderWidgetHostViewMac::~RenderWidgetHostViewMac() + 14 7 libobjc.A.dylib 0x00007fffac713686 object_cxxDestructFromClass(objc_object*, objc_class*) + 127 8 libobjc.A.dylib 0x00007fffac70c0c6 objc_destructInstance + 92 9 libobjc.A.dylib 0x00007fffac70c059 object_dispose + 22 10 AppKit 0x00007fff955c86e9 -[NSResponder dealloc] + 139 11 AppKit 0x00007fff955c7102 -[NSView dealloc] + 381 12 content_browsertests 0x0000000107013158 -[BaseView dealloc] + 56 13 content_browsertests 0x0000000105820f48 -[RenderWidgetHostViewCocoa dealloc] + 168 14 libobjc.A.dylib 0x00007fffac711db7 objc_object::sidetable_release(bool) + 285 15 libobjc.A.dylib 0x00007fffac70ce8d (anonymous namespace)::AutoreleasePoolPage::pop(void*) + 781 16 CoreFoundation 0x00007fff97a61856 _CFAutoreleasePoolPop + 22 17 Foundation 0x00007fff994abf13 -[NSAutoreleasePool drain] + 146 18 content_browsertests 0x00000001077c8aa6 service_manager::Main(service_manager::MainParams const&) + 3142 19 content_browsertests 0x0000000105467a94 content::ContentMain(content::ContentMainParams const&) + 68 20 content_browsertests 0x0000000106393108 content::BrowserTestBase::SetUp() + 1768 21 content_browsertests 0x0000000106383876 content::ContentBrowserTest::SetUp() + 486 22 content_browsertests 0x00000001062f7e11 testing::Test::Run() + 97 23 content_browsertests 0x00000001062f8880 testing::TestInfo::Run() + 288 24 content_browsertests 0x00000001062f8de7 testing::TestCase::Run() + 263 25 content_browsertests 0x00000001062feee7 testing::internal::UnitTestImpl::RunAllTests() + 871 26 content_browsertests 0x00000001062feb53 testing::UnitTest::Run() + 163 27 content_browsertests 0x00000001063bcc43 base::TestSuite::Run() + 163 28 content_browsertests 0x000000010638711a content::ContentTestLauncherDelegate::RunTestSuite(int, char**) + 42 29 content_browsertests 0x00000001063aaa19 content::LaunchTests(content::TestLauncherDelegate*, int, int, char**) + 313 30 content_browsertests 0x00000001063870ba main + 74 31 libdyld.dylib 0x00007fffacffb255 start + 1
,
Jun 21 2017
,
Jun 22 2017
I'm not able to reproduce this locally. Don't see it as a "Priority 1" since there is no depth capture enabled on Android yet. Reducing to Pri 2 as the test is disabled. Continuing with attempt to make it reproducible. While working on Tango support ( Issue 674440 ) I have noticed the crash in compositor, that is not related to this, but plan to add more tests covering it and the case here.
,
Jun 22 2017
> Labels: -Pri-1 Pri-2 Please, feel free to raise it back to Pri-1 if I misevaluated in #8.
,
Jun 23 2017
Thanks Aleksandar, sounds good. [I file P1 bugs by default when disabling tests as sheriff to make sure the bugs get noticed, since it's hard to know how critical the test I'm disabling is, but once feature owners like yourself have looked at the bug I defer to their judgement on prioritisation :-)]
,
Jan 24 2018
,
Jan 24 2018
|
|||||||||
►
Sign in to add a comment |
|||||||||
Comment 1 by awdf@chromium.org
, Jun 6 2017