content_shell crash fails |
||
Issue descriptionWhat steps will reproduce the problem? (1) "checkout_libaom": True (2) ./out/Default/content_shell --run-layout-test chrome://crash --enable-crash-reporter --crash-dumps-dir=/tmp/crash What happens instead? https://logs.chromium.org/v/?s=chromium%2Fbb%2Ftryserver.chromium.linux%2Flinux_chromium_rel_ng%2F592106%2F%2B%2Frecipes%2Fsteps%2Fcontent_shell_crash_test__with_patch_%2F0%2Fstdout [31170:31170:1121/101302.257775:FATAL:feature_list.cc(259)] Check failed: !g_initialized_from_accessor. Seems like it might be related to the feature check somehow? This might be something I broke setting up the flag, so I'll try to dig into it too. Rest of stack trace: #0 0x00000241888c base::debug::StackTrace::StackTrace() #1 0x00000242e62c logging::LogMessage::~LogMessage() #2 0x00000241bce8 base::FeatureList::InitializeInstance() #3 0x000001e9ce16 content::BrowserMainLoop::PreCreateThreads() #4 0x0000021cf2d7 content::StartupTaskRunner::RunAllTasksNow() #5 0x000001e9d48c content::BrowserMainLoop::CreateStartupTasks() #6 0x000001ea20c8 content::BrowserMainRunnerImpl::Initialize() #7 0x0000023cf1ad LayoutTestBrowserMain() #8 0x0000023ce702 content::ShellMainDelegate::RunProcess() #9 0x000001de0040 content::RunNamedProcessTypeMain() #10 0x000001de0a55 content::ContentMainRunnerImpl::Run() #11 0x000003603bcf service_manager::Main() #12 0x0000015ad861 content::ContentMain() #13 0x000001040152 main #14 0x7f71fc5362b1 __libc_start_main #15 0x00000104002a _start Failed to generate minidump.Received signal 6 #0 0x00000241888c base::debug::StackTrace::StackTrace() #1 0x0000024183f1 base::debug::(anonymous namespace)::StackDumpSignalHandler() #2 0x7f7200f7d0c0 <unknown> #3 0x7f71fc548fcf gsignal #4 0x7f71fc54a3fa abort #5 0x000002417065 base::debug::BreakDebugger() #6 0x00000242ea41 logging::LogMessage::~LogMessage() #7 0x00000241bce8 base::FeatureList::InitializeInstance() #8 0x000001e9ce16 content::BrowserMainLoop::PreCreateThreads() #9 0x0000021cf2d7 content::StartupTaskRunner::RunAllTasksNow() #10 0x000001e9d48c content::BrowserMainLoop::CreateStartupTasks() #11 0x000001ea20c8 content::BrowserMainRunnerImpl::Initialize() #12 0x0000023cf1ad LayoutTestBrowserMain() #13 0x0000023ce702 content::ShellMainDelegate::RunProcess() #14 0x000001de0040 content::RunNamedProcessTypeMain() #15 0x000001de0a55 content::ContentMainRunnerImpl::Run() #16 0x000003603bcf service_manager::Main() #17 0x0000015ad861 content::ContentMain() #18 0x000001040152 main #19 0x7f71fc5362b1 __libc_start_main #20 0x00000104002a _start r8: 0000000000000000 r9: 00007ffc44e6f660 r10: 0000000000000008 r11: 0000000000000246 r12: 00007ffc44e6fd60 r13: 00007ffc44e6fd50 r14: 00007ffc44e6fd58 r15: 00007ffc44e6f910 di: 0000000000000002 si: 00007ffc44e6f660 bp: 00007ffc44e6f8a0 bx: 0000000000000006 dx: 0000000000000000 ax: 0000000000000000 cx: 00007f71fc548fcf sp: 00007ffc44e6f6d8 ip: 00007f71fc548fcf efl: 0000000000000246 cgf: 002b000000000033 erf: 0000000000000000 trp: 0000000000000000 msk: 0000000000000000 cr2: 0000000000000000 [end of stack trace] Calling _exit(1). Core file will not be generated.
,
Nov 21 2017
OIC this is a bot run. Will see what I can find locally.
,
Nov 21 2017
Probably this is due to mime util being initialized earlier on and accessing the feature list before its ready.
,
Nov 21 2017
Not sure if there is an auto-crash test for chrome ... it came up in this change: https://chromium-review.googlesource.com/c/chromium/src/+/780581 on this bot: https://ci.chromium.org/buildbot/tryserver.chromium.linux/linux_chromium_rel_ng/592106 I can repro locally and after disabling av1 (enable_av1_decoder = false) the test passes.
,
Nov 22 2017
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/d69958f1aa564e532d7edc7f57f7ba4e6dd77f43 commit d69958f1aa564e532d7edc7f57f7ba4e6dd77f43 Author: Dale Curtis <dalecurtis@chromium.org> Date: Wed Nov 22 22:12:27 2017 Don't invoke base::FeatureList access before browser startup. base::FeatureList is initialized in the PreCreateThreads() section of browser startup, but when running layout tests we invoke media::RemoveProprietaryMediaTypesAndCodecsForTests() prior to even starting the browser client. Because this method checks a base::Feature it thus implicitly sets up features ahead of the source of truth in PreCreateThread() In the near term we will be deleting the media method, but that probably won't land for a couple weeks, so in the mean time so that we're ready for the next dev release w/ AV1, move the call into LayoutTestContentRendererClient::RenderThreadStarted(); this isn't needed in the browser anyways. BUG= 787522 , 787575 TEST=no content_shell crash w/ av1 enabled. Change-Id: I4d97d181431862d41e8abbc0ea4c4e90758f3ed5 Reviewed-on: https://chromium-review.googlesource.com/783514 Commit-Queue: Dale Curtis <dalecurtis@chromium.org> Reviewed-by: Peter Beverloo <peter@chromium.org> Cr-Commit-Position: refs/heads/master@{#518768} [modify] https://crrev.com/d69958f1aa564e532d7edc7f57f7ba4e6dd77f43/content/shell/app/shell_main_delegate.cc [modify] https://crrev.com/d69958f1aa564e532d7edc7f57f7ba4e6dd77f43/content/shell/renderer/layout_test/layout_test_content_renderer_client.cc
,
Nov 22 2017
|
||
►
Sign in to add a comment |
||
Comment 1 by dalecur...@chromium.org
, Nov 21 2017