System.ComponentModel.EventBasedAsync.Tests.BackgroundWorkerTests.TestFinalization failed on mono interpreter mode

[safern]

System.ComponentModel.EventBasedAsync.Tests.BackgroundWorkerTests.TestFinalization

Console Log Summary

    System.ComponentModel.EventBasedAsync.Tests.BackgroundWorkerTests.TestFinalization [FAIL]
      Assert.True() Failure
      Expected: True
      Actual:   False
      Stack Trace:
        /_/src/libraries/System.ComponentModel.EventBasedAsync/tests/BackgroundWorkerTests.cs(343,0): at System.ComponentModel.EventBasedAsync.Tests.BackgroundWorkerTests.TestFinalization()
        /_/src/mono/netcore/System.Private.CoreLib/src/System/Reflection/RuntimeMethodInfo.cs(359,0): at System.Reflection.RuntimeMethodInfo.Invoke(Object obj, BindingFlags invokeAttr, Binder binder, Object[] parameters, CultureInfo culture)

Builds

Build	Pull Request	Test Failure Count
#662934	Rolling	2

Configurations

• net5.0-Linux-Release-x64-mono_interpreter_release-Debian.9.Amd64.Open

Helix Logs

Build	Pull Request	Console	Core	Test Results	Run Client
#662934	Rolling	console.log		testResults.xml	run_client.py
#662934	Rolling	console.log		testResults.xml	run_client.py

cc: @steveisok

[safern]

I'll put up a PR to disable the test on interpreter mode.

Tagging subscribers to this area: @safern
Notify danmosemsft if you want to be subscribed.

Tagging subscribers to this area: @BrzVlad, @lewurm
Notify danmosemsft if you want to be subscribed.

[BrzVlad]

This is not really interpreter related. This is due to conservative pinning and stack pinning just happens in different patterns when using the interpreter, compared to jit code. We should disable this test based on [ConditionalFact(typeof(PlatformDetection), nameof(PlatformDetection.IsPreciseGcSupported))]. As other tests that call GC.WaitForPendingFinalizers ()

[safern]

Thanks for looking into this @BrzVlad -- makes sense.