Update assert to handle object[] element type

[AaronRobinsonMSFT]

Fixes #105186

@jakobbotsch this is very odd. This code path is the same on all platforms, but only seems to be taken with these tests on non-Windows platforms. That is the reason it didn't fail on Windows. Any ideas why exceptions are being thrown on non-Windows, but not on Windows?

[dotnet-policy-service]

Tagging subscribers to this area: @mangod9
See info in area-owners.md if you want to be subscribed.

[jakobbotsch]

Any ideas why exceptions are being thrown on non-Windows, but not on Windows?

The JIT is generating direct call to CORINFO_HELP_OVERFLOW for these tests on both Windows and Linux, so seems like there is something different there. It is bound to JIT_Overflow that does COMPlusThrow(kOverflowException);, but apparently that does not end up in this code on Windows.
Maybe @janvorli has better idea what is different here.

BTW, while stepping through this on Windows I ended up in this weird code:

runtime/src/coreclr/vm/excep.cpp

Lines 7354 to 7357 in 98b165d

	while (count < 20 && sp < stopPoint)
	{
	sp += 1;
	}

[janvorli]

BTW, while stepping through this on Windows I ended up in this weird code:

I've noticed that recently too. This is a remainder of logging of a piece of stack into stresslog, see this in coreclr 3.1:
https://github.com/dotnet/coreclr/blob/v3.1.32/src/vm/excep.cpp#L8139-L8147.
The body of the loop was removed , but blame on github somehow isn't able to show me the change that did that.

[jkotas]

his is a remainder of logging of a piece of stack into stresslog, see this in coreclr 3.1:
https://github.com/dotnet/coreclr/blob/v3.1.32/src/vm/excep.cpp#L8139-L8147.
The body of the loop was removed , but blame on github somehow isn't able to show me the change that did that.

This was deleted in
https://github.com/dotnet/runtime/pull/81927/files#diff-47cbfa47fb88845e5e76fd4df0a84e9da494508ffe80da116511fc2d0becd9baL7614

Deleting the rest #105346

[AaronRobinsonMSFT]

Maybe @janvorli has better idea what is different here.

I tried to debug this and it is mostly related to the SEH simulation on non-Windows. Particularly usage of NativeExceptionHolderBase.