ReadyToRun and Unicode class names

[JensNordenbro]

Description

<Project Sdk="Microsoft.NET.Sdk">

  <PropertyGroup>
    <OutputType>Exe</OutputType>
    <TargetFramework>net6.0</TargetFramework>
    <ImplicitUsings>enable</ImplicitUsings>
    <Nullable>enable</Nullable>
    **<PublishReadyToRun>true</PublishReadyToRun>**
  </PropertyGroup>
</Project>

Console.WriteLine("Hello, World!");

var type = Type.GetType("AotProblem." + "_测试数据记录仪_Iiİı_åäö_Controller_DataLogger1_log_all_", false);

Console.WriteLine(type?.FullName??"null");

var obj = Activator.CreateInstance(type);

Console.WriteLine(obj?.GetType().Name);

Console.ReadLine();


namespace AotProblem
{
    public class _测试数据记录仪_Iiİı_åäö_Controller_DataLogger1_log_all_
    {

    }
}

$ dotnet --list-sdks
5.0.400 [C:\Program Files\dotnet\sdk]
6.0.100-rc.2.21505.57 [C:\Program Files\dotnet\sdk]

$ dotnet --list-runtimes
Microsoft.AspNetCore.All 2.1.29 [C:\Program Files\dotnet\shared\Microsoft.AspNetCore.All]
Microsoft.AspNetCore.App 2.1.29 [C:\Program Files\dotnet\shared\Microsoft.AspNetCore.App]
Microsoft.AspNetCore.App 3.1.18 [C:\Program Files\dotnet\shared\Microsoft.AspNetCore.App]
Microsoft.AspNetCore.App 5.0.9 [C:\Program Files\dotnet\shared\Microsoft.AspNetCore.App]
Microsoft.AspNetCore.App 6.0.0-rc.2.21480.10 [C:\Program Files\dotnet\shared\Microsoft.AspNetCore.App]
Microsoft.NETCore.App 2.1.29 [C:\Program Files\dotnet\shared\Microsoft.NETCore.App]
Microsoft.NETCore.App 3.1.18 [C:\Program Files\dotnet\shared\Microsoft.NETCore.App]
Microsoft.NETCore.App 5.0.9 [C:\Program Files\dotnet\shared\Microsoft.NETCore.App]
Microsoft.NETCore.App 6.0.0-rc.2.21480.5 [C:\Program Files\dotnet\shared\Microsoft.NETCore.App]
Microsoft.WindowsDesktop.App 3.1.18 [C:\Program Files\dotnet\shared\Microsoft.WindowsDesktop.App]
Microsoft.WindowsDesktop.App 5.0.9 [C:\Program Files\dotnet\shared\Microsoft.WindowsDesktop.App]
Microsoft.WindowsDesktop.App 6.0.0-rc.2.21501.6 [C:\Program Files\dotnet\shared\Microsoft.WindowsDesktop.App]

Found on Windows 10 21H1 19043.1320.

Reproduction Steps

Consider code + csproj above. (dotnet new console with true|false</..> added to it. )

The code below will work when PublishReadyToRun is false, but not when true.

Changing class name to "K" makes it work in all cases however that will not be possible for many customers.

Expected behavior

Code should work regardless of PublishReadyToRun bool value.

Actual behavior

Does work on PublishReadyToRun=false
Does NOT work on PublishReadyToRun=true

Regression?

No response

Known Workarounds

No response

Configuration

No response

Other information

No response

[janvorli]

cc: @dotnet/crossgen-contrib

[mangod9]

@trylek / @AntonLapounov would you be able to take a look at this?

[trylek]

I'll try to repro this locally and we can discuss it in more detail either in today technical sync meeting or in tomorrow Crossgen2 sync-up meeting.

[AntonLapounov]

Do we know how big the set of characters affected by this issue?

[trylek]

I have already figured out the cause - the runtime hash calculation in

runtime/src/coreclr/vm/typehashingalgorithms.h

Line 28 in 0266f03

hash1 = (hash1 + _rotl(hash1, 5)) ^ src[i];

sign-extends the individual UTF8 bytes whereas Crossgen2 implementation in

runtime/src/coreclr/tools/aot/ILCompiler.ReadyToRun/Compiler/ReadyToRunHashCode.cs

Line 37 in 0266f03

hash1 = unchecked(hash1 + RotateLeft(hash1, 5)) ^ src[i];

zero-extends them. I have just sent out the fix for PR. I guess it also answers @AntonLapounov's question - it basically affects all non-ASCII characters in type names.