Runtime hangs on exit on a spinlock

[mikedn]

From time to time the jit diff tool hangs on exit:

ntdll.dll!NtDelayExecution()
KernelBase.dll!SleepEx()
coreclr.dll!SpinLock::SpinToAcquire() Line 270
coreclr.dll!CallCounter::IsCallCountingEnabled(MethodDesc * pMethodDesc) Line 56
coreclr.dll!TieredCompilationManager::GetInitialOptimizationTier(MethodDesc * pMethodDesc) Line 115
coreclr.dll!PrepareCodeConfig::GetJitOptimizationTier(PrepareCodeConfig * config, MethodDesc * methodDesc) Line 1215
coreclr.dll!ETW::MethodLog::SendMethodEvent(MethodDesc * pMethodDesc, unsigned long dwEventOptions, int bIsJit, SString * namespaceOrClassName, SString * methodName, SString * methodSignature, unsigned __int64 pNativeCodeStartAddress, PrepareCodeConfig * pConfig) Line 6402
coreclr.dll!ETW::MethodLog::SendEventsForJitMethodsHelper(LoaderAllocator * pLoaderAllocatorFilter, unsigned long dwEventOptions, int fLoadOrDCStart, int fUnloadOrDCEnd, int fSendMethodEvent, int fSendILToNativeMapEvent, int fGetCodeIds) Line 6889
coreclr.dll!ETW::MethodLog::SendEventsForJitMethods(BaseDomain * pDomainFilter, LoaderAllocator * pLoaderAllocatorFilter, unsigned long dwEventOptions) Line 7046
coreclr.dll!ETW::EnumerationLog::IterateDomain(BaseDomain * pDomain, unsigned long enumerationOptions) Line 7139
coreclr.dll!ETW::EnumerationLog::IterateAppDomain(AppDomain * pAppDomain, unsigned long enumerationOptions) Line 7096
coreclr.dll!ETW::EnumerationLog::EnumerationHelper(Module * moduleFilter, BaseDomain * enumerationOptions, unsigned long) Line 7412
coreclr.dll!ETW::EnumerationLog::ProcessShutdown() Line 5674
coreclr.dll!EEShutDownHelper(int fIsDllUnloading) Line 1327
coreclr.dll!EEShutDown(int fIsDllUnloading) Line 1803
coreclr.dll!EEDllMain(HINSTANCE__ * hInst, unsigned long dwReason, void * lpReserved) Line 1973
coreclr.dll!DllMain(void * hInstance, unsigned long dwReason, void * lpReserved) Line 156
coreclr.dll!CoreDllMain(void * hInstance, unsigned long dwReason, void * lpReserved) Line 107
ntdll.dll!LdrpCallInitRoutine()

This is the only remaining thread in the process and all it does it to spin there. Presumably the spinlock was abandoned by a terminated thread.

Known issue? Perhaps already fixed? The jit utils are using the released 3.0, not the current coreclr build.

[jkotas]

cc @kouvel

[kouvel]

Fixed by:

• master - Process/send etw rundown events only during graceful shutdown and not during process detach (a form of abrupt shutdown) coreclr#27238
• 3.1 - [3.1] Protect against a rare invalid lock acquision attempt during etw processing during shutdown coreclr#27241

Did not meet the bar for 3.0.