[mono] Use correct cast_class for IntPtr[] #103841
Conversation
This change fixes the behavior to be like modern .NET instead of .NET Framework. Fixes dotnet#97145
ghost
added
the
|
I wonder if there's any tests that capture the modern behavior that we previously disabled in Mono. |
|
cc @BrzVlad is this going to cause any problems for interp? |
|
This seems to have crashed on ios/tvos device runs. I ran @lambdageek's sample locally at least works for iossimulator. I'll try on a local device and see if I can get any info out of it. |
| break; | ||
| case MONO_TYPE_U4: | ||
| #if TARGET_SIZEOF_VOID_P == 4 | ||
| case MONO_TYPE_I: |
There was a problem hiding this comment.
MONO_TYPE_BOOLEAN -> byte_class normalization above looks suspect as well. I do not think regular CoreCLR does normalization like that anywhere.
There was a problem hiding this comment.
@jkotas I think the bool -> int8 fixup comes from when I misread I.8.7.2:
A location type T is compatible-with a location type U if and only if one of the following holds.
- T and U are not managed pointer types and T is compatible-with U according to the definition in §I.8.7.1.
- T and U are both managed pointer types and T is pointer-element-compatible-with U.
A managed pointer type T is pointer-element-compatible-with a managed pointer type U if and
only if T has verification type V and U has verification type W and V is identical to W.
The verification type of a bool is defined in I.8.7 as
- If the reduced type of T is:
a. int8 or bool, then its verification type is int8.
The problem is I think I applied the verification type rules to unmanaged pointer types (ie bool*) rather than to managed pointer types (ie bool&)
|
Weird, local device runs are fine. |
|
Appears there's some infra issues with the tvos and ios device legs. I validated system.runtime tests on a local device. |
5 participants
This change fixes the behavior to be like modern .NET instead of .NET Framework.
Fixes #97145