[mono] Use correct cast_class for IntPtr[] by steveisok · Pull Request #103841 · dotnet/runtime

steveisok · 2024-06-21T21:45:04Z

This change fixes the behavior to be like modern .NET instead of .NET Framework.

Fixes#97145

This change fixes the behavior to be like modern .NET instead of .NET Framework. Fixesdotnet#97145

kg

Looks right, but I'm not 100% certain.

lambdageek · 2024-06-21T21:52:12Z

I wonder if there's any tests that capture the modern behavior that we previously disabled in Mono.

kg · 2024-06-21T21:52:14Z

cc @BrzVlad is this going to cause any problems for interp?

steveisok · 2024-06-21T23:14:34Z

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.

jkotas · 2024-06-22T01:57:49Z

 break;
-	case MONO_TYPE_U4:
-#if TARGET_SIZEOF_VOID_P == 4
 	case MONO_TYPE_I:


MONO_TYPE_BOOLEAN -> byte_class normalization above looks suspect as well. I do not think regular CoreCLR does normalization like that anywhere.

@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&)

steveisok · 2024-06-24T17:36:03Z

Weird, local device runs are fine.

steveisok · 2024-06-24T20:12:54Z

Appears there's some infra issues with the tvos and ios device legs. I validated system.runtime tests on a local device.

steveisok · 2024-06-21T21:45:04Z

This change fixes the behavior to be like modern .NET instead of .NET Framework.

Fixes#97145

This change fixes the behavior to be like modern .NET instead of .NET Framework. Fixesdotnet#97145

kg

Looks right, but I'm not 100% certain.

lambdageek · 2024-06-21T21:52:12Z

I wonder if there's any tests that capture the modern behavior that we previously disabled in Mono.

kg · 2024-06-21T21:52:14Z

cc @BrzVlad is this going to cause any problems for interp?

steveisok · 2024-06-21T23:14:34Z

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.

jkotas · 2024-06-22T01:57:49Z

 break;
-	case MONO_TYPE_U4:
-#if TARGET_SIZEOF_VOID_P == 4
 	case MONO_TYPE_I:


MONO_TYPE_BOOLEAN -> byte_class normalization above looks suspect as well. I do not think regular CoreCLR does normalization like that anywhere.

@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&)

steveisok · 2024-06-24T17:36:03Z

Weird, local device runs are fine.

steveisok · 2024-06-24T20:12:54Z

Appears there's some infra issues with the tvos and ios device legs. I validated system.runtime tests on a local device.

steveisok · 2024-06-21T21:45:04Z

This change fixes the behavior to be like modern .NET instead of .NET Framework.

Fixes#97145

This change fixes the behavior to be like modern .NET instead of .NET Framework. Fixesdotnet#97145

kg

Looks right, but I'm not 100% certain.

lambdageek · 2024-06-21T21:52:12Z

I wonder if there's any tests that capture the modern behavior that we previously disabled in Mono.

kg · 2024-06-21T21:52:14Z

cc @BrzVlad is this going to cause any problems for interp?

steveisok · 2024-06-21T23:14:34Z

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.

jkotas · 2024-06-22T01:57:49Z

 break;
-	case MONO_TYPE_U4:
-#if TARGET_SIZEOF_VOID_P == 4
 	case MONO_TYPE_I:


MONO_TYPE_BOOLEAN -> byte_class normalization above looks suspect as well. I do not think regular CoreCLR does normalization like that anywhere.

@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&)

steveisok · 2024-06-24T17:36:03Z

Weird, local device runs are fine.

steveisok · 2024-06-24T20:12:54Z

Appears there's some infra issues with the tvos and ios device legs. I validated system.runtime tests on a local device.

steveisok · 2024-06-21T21:45:04Z

This change fixes the behavior to be like modern .NET instead of .NET Framework.

Fixes#97145

This change fixes the behavior to be like modern .NET instead of .NET Framework. Fixesdotnet#97145

kg

Looks right, but I'm not 100% certain.

lambdageek · 2024-06-21T21:52:12Z

I wonder if there's any tests that capture the modern behavior that we previously disabled in Mono.

kg · 2024-06-21T21:52:14Z

cc @BrzVlad is this going to cause any problems for interp?

steveisok · 2024-06-21T23:14:34Z

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.

jkotas · 2024-06-22T01:57:49Z

 break;
-	case MONO_TYPE_U4:
-#if TARGET_SIZEOF_VOID_P == 4
 	case MONO_TYPE_I:


MONO_TYPE_BOOLEAN -> byte_class normalization above looks suspect as well. I do not think regular CoreCLR does normalization like that anywhere.

@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&)

steveisok · 2024-06-24T17:36:03Z

Weird, local device runs are fine.

steveisok · 2024-06-24T20:12:54Z

Appears there's some infra issues with the tvos and ios device legs. I validated system.runtime tests on a local device.

steveisok · 2024-06-21T21:45:04Z

This change fixes the behavior to be like modern .NET instead of .NET Framework.

Fixes#97145

This change fixes the behavior to be like modern .NET instead of .NET Framework. Fixesdotnet#97145

kg

Looks right, but I'm not 100% certain.

lambdageek · 2024-06-21T21:52:12Z

I wonder if there's any tests that capture the modern behavior that we previously disabled in Mono.

kg · 2024-06-21T21:52:14Z

cc @BrzVlad is this going to cause any problems for interp?

steveisok · 2024-06-21T23:14:34Z

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.

jkotas · 2024-06-22T01:57:49Z

 break;
-	case MONO_TYPE_U4:
-#if TARGET_SIZEOF_VOID_P == 4
 	case MONO_TYPE_I:


MONO_TYPE_BOOLEAN -> byte_class normalization above looks suspect as well. I do not think regular CoreCLR does normalization like that anywhere.

@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&)

steveisok · 2024-06-24T17:36:03Z

Weird, local device runs are fine.

steveisok · 2024-06-24T20:12:54Z

Appears there's some infra issues with the tvos and ios device legs. I validated system.runtime tests on a local device.

steveisok · 2024-06-21T21:45:04Z

This change fixes the behavior to be like modern .NET instead of .NET Framework.

Fixes#97145

This change fixes the behavior to be like modern .NET instead of .NET Framework. Fixesdotnet#97145

kg

Looks right, but I'm not 100% certain.

lambdageek · 2024-06-21T21:52:12Z

I wonder if there's any tests that capture the modern behavior that we previously disabled in Mono.

kg · 2024-06-21T21:52:14Z

cc @BrzVlad is this going to cause any problems for interp?

steveisok · 2024-06-21T23:14:34Z

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.

jkotas · 2024-06-22T01:57:49Z

 break;
-	case MONO_TYPE_U4:
-#if TARGET_SIZEOF_VOID_P == 4
 	case MONO_TYPE_I:


MONO_TYPE_BOOLEAN -> byte_class normalization above looks suspect as well. I do not think regular CoreCLR does normalization like that anywhere.

@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&)

steveisok · 2024-06-24T17:36:03Z

Weird, local device runs are fine.

steveisok · 2024-06-24T20:12:54Z

Appears there's some infra issues with the tvos and ios device legs. I validated system.runtime tests on a local device.

steveisok · 2024-06-21T21:45:04Z

This change fixes the behavior to be like modern .NET instead of .NET Framework.

Fixes#97145

This change fixes the behavior to be like modern .NET instead of .NET Framework. Fixesdotnet#97145

kg

Looks right, but I'm not 100% certain.

lambdageek · 2024-06-21T21:52:12Z

I wonder if there's any tests that capture the modern behavior that we previously disabled in Mono.

kg · 2024-06-21T21:52:14Z

cc @BrzVlad is this going to cause any problems for interp?

steveisok · 2024-06-21T23:14:34Z

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.

jkotas · 2024-06-22T01:57:49Z

 break;
-	case MONO_TYPE_U4:
-#if TARGET_SIZEOF_VOID_P == 4
 	case MONO_TYPE_I:


MONO_TYPE_BOOLEAN -> byte_class normalization above looks suspect as well. I do not think regular CoreCLR does normalization like that anywhere.

@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&)

steveisok · 2024-06-24T17:36:03Z

Weird, local device runs are fine.

steveisok · 2024-06-24T20:12:54Z

Appears there's some infra issues with the tvos and ios device legs. I validated system.runtime tests on a local device.

steveisok · 2024-06-21T21:45:04Z

This change fixes the behavior to be like modern .NET instead of .NET Framework.

Fixes#97145

This change fixes the behavior to be like modern .NET instead of .NET Framework. Fixesdotnet#97145

kg

Looks right, but I'm not 100% certain.

lambdageek · 2024-06-21T21:52:12Z

I wonder if there's any tests that capture the modern behavior that we previously disabled in Mono.

kg · 2024-06-21T21:52:14Z

cc @BrzVlad is this going to cause any problems for interp?

steveisok · 2024-06-21T23:14:34Z

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.

jkotas · 2024-06-22T01:57:49Z

 break;
-	case MONO_TYPE_U4:
-#if TARGET_SIZEOF_VOID_P == 4
 	case MONO_TYPE_I:


MONO_TYPE_BOOLEAN -> byte_class normalization above looks suspect as well. I do not think regular CoreCLR does normalization like that anywhere.

@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&)

steveisok · 2024-06-24T17:36:03Z

Weird, local device runs are fine.

steveisok · 2024-06-24T20:12:54Z

Appears there's some infra issues with the tvos and ios device legs. I validated system.runtime tests on a local device.

[mono] Use correct cast_class for IntPtr[]
8d3cbd0
This change fixes the behavior to be like modern .NET instead of .NET Framework. Fixesdotnet#97145

steveisok requested review from lambdageek and thaystg as code owners June 21, 2024 21:45

ghost added the area-VM-meta-mono label Jun 21, 2024

dotnet-policy-serviceBot assigned steveisok Jun 21, 2024

steveisok requested review from fanyang-mono and kg June 21, 2024 21:45

lambdageek approved these changes Jun 21, 2024
View reviewed changes

kg approved these changes Jun 21, 2024
View reviewed changes

jkotas reviewed Jun 22, 2024
View reviewed changes

fanyang-mono reviewed Jun 24, 2024
View reviewed changes

Comment threadsrc/mono/mono/metadata/class-init.c

Merge remote-tracking branch 'upstream/main' into fix-intptr-cast
4cabf4c

fanyang-mono approved these changes Jun 24, 2024
View reviewed changes

Add test
0cb529d

jkotas reviewed Jun 25, 2024
View reviewed changes

Comment threadsrc/tests/Loader/classloader/Casting/Normalization.csproj Outdated

steveisok added 4 commits June 25, 2024 12:33

Fix hasty stuff
45323c5

Losing it a bit ;-)
ae4a5e9

it's x0 instead of x
ea60e34

I guess test methods need to be static
4facff1

build-analysisBot mentioned this pull request Jun 26, 2024
Failed test: baseservices/exceptions/unhandled/unhandledTester/unhandledTester #100495
Closed

This was referenced Jun 26, 2024
The Operation will be canceled. The next steps may not contain expected logs. dotnet/dnceng#3008
Open
The job running on agent NetCore-Public ran longer than the maximum time #104044
Closed

steveisok merged commit 5a0eb6e into dotnet:mainJun 26, 2024

steveisok deleted the fix-intptr-cast branch June 26, 2024 20:47

jkotas mentioned this pull request Jun 28, 2024
Convert Array.IsSimpleCopy and CanAssignArray type to managed #104103
Merged

github-actionsBot locked and limited conversation to collaborators Jul 27, 2024