ANRs after disabling concurrent GC

[feric-isc]

Android framework version

net8.0-android

Affected platform version

8.0.17

Description

Very similar to issue #9365 (comment).

Many months back, we switched from Xamarin.Android to .NET8. At first we were having many NEW crashes in that version.

They were mostly SIGSEGV, SIGABRT and SIGQUIT. Note that even before, in Xamarin, we had these kind of errors, but there were a LOT more with .NET8.

Following the recommendations in an issue we opened dotnet/runtime#105389 (comment), it was recommended we disable the concurrent GC. We did so and this had good results as it brought us back to the SIGSEGV levels that we had prior. It's not good to have these issues and we hope to find a resolution in our another ticket but at least we were back to our original level of errors for SIGSEGV, etc.

Now, in .NET8, we still have the SIGSEGV errors that we had prior in Xamarin but some new ANR errors started popping up.

Here are a few stacktraces found in Crashlytics:

anr_stacktrace.txt
anr_stacktrace2.txt
anr_stacktrace3.txt

The first 2 seemed to occur sometimes in our Xamarin versions but not as much as with .NET8 versions. The third one only seems to occur since .NET8 version (I base this on the data provided from Crashlytics that specifies when an issue was first discovered.

Let me know what I can do to help troubleshoot the issue

Steps to Reproduce

We do not have exact steps to reproduce. This occurs apparently randomly. We run the same automated tests regularly and the error will come back but we cannot find a reproduction pattern.

Did you find any workaround?

No

Relevant log output

[jonathanpeppers]

Can you update to .NET 9? .NET 8 is out of support for mobile:

• https://aka.ms/maui-support-policy

If the issue is still happening in .NET 9, we could possibly address the problem in that version.

[yeahg-dev]

We are also experiencing persistent ANRs, and we've observed a stack trace similar to anr_stacktrace3.txt. We're currently using .NET 9.

anrLog.txt

[feric-isc]

Hello @jonathanpeppers . I will need a little time to get my "proof" on version 9 but I will get started on it. I will try to get back to you ASAP but it will take a week or two. This being said, in the meantime, the log provided by @yeahg-dev indeed looks similar to problem and she has it on .NET9

[jonathanantoine]

Any update :) ?

[feric-isc]

On our side we have made preliminary tests with version 9 but it's not conclusive. Knowing this error seems very random, it's hard to make any conclusions before the fix is in production. We'll only have news early Q1 if it helps or not.