Benchmarkdotnet: Disassembler fails with generic instance

[xoofx]

Hello! (Followup of #612)
So I wanted to try the disassembler with the 0.10.12 version but getting a crash on the following example:

```c#
[DisassemblyDiagnoser]
public class BenchFail where T : new()
{
[Benchmark] public void Bench() => new T();
}
// ...
// In the main
BenchmarkDotNet.Running.BenchmarkRunner.Run>();

And the error:

// AfterAll
Failed to disassemble with following exception:
Object reference not set to an instance of an object.
at BenchmarkDotNet.Disassembler.Program.Disassemble(Settings settings, ClrRuntime runtime, State state)
at BenchmarkDotNet.Disassembler.Program.Handle(Settings settings)
at BenchmarkDotNet.Disassembler.Program.Main(String[] args)

Unhandled Exception: System.InvalidOperationException: There is an error in XML document (0, 0). ---> System.Xml.XmlException: Root element is missing.
at System.Xml.XmlTextReaderImpl.Throw(Exception e)
at System.Xml.XmlTextReaderImpl.ParseDocumentContent()
at System.Xml.XmlReader.MoveToContent()
at Microsoft.Xml.Serialization.GeneratedAssembly.XmlSerializationReaderDisassemblyResult.Read9_DisassemblyResult()
--- End of inner exception stack trace ---
at System.Xml.Serialization.XmlSerializer.Deserialize(XmlReader xmlReader, String encodingStyle, XmlDeserializationEvents events)
at BenchmarkDotNet.Diagnosers.WindowsDisassembler.Dissasemble(DiagnoserActionParameters parameters)
at BenchmarkDotNet.Diagnosers.DisassemblyDiagnoser.Handle(HostSignal signal, DiagnoserActionParameters parameters)
at BenchmarkDotNet.Extensions.CommonExtensions.ForEachT
at BenchmarkDotNet.Loggers.SynchronousProcessOutputLoggerWithDiagnoser.ProcessInput()
at BenchmarkDotNet.Toolchains.Executor.Execute(Process process, Benchmark benchmark, SynchronousProcessOutputLoggerWithDiagnoser loggerWithDiagnoser, ILogger logger)
at BenchmarkDotNet.Toolchains.Executor.Execute(Benchmark benchmark, ILogger logger, String exePath, String workingDirectory, String args, IDiagnoser diagnoser, IResolver resolver, IConfig config)
at BenchmarkDotNet.Toolchains.Executor.Execute(ExecuteParameters executeParameters)
at BenchmarkDotNet.Running.BenchmarkRunnerCore.Execute(ILogger logger, Benchmark benchmark, IToolchain toolchain, BuildResult buildResult, IConfig config, IResolver resolver)
at BenchmarkDotNet.Running.BenchmarkRunnerCore.RunCore(Benchmark benchmark, ILogger logger, ReadOnlyConfig config, String rootArtifactsFolderPath, Func2 toolchainProvider, IResolver resolver, BuildResult buildResult) at BenchmarkDotNet.Running.BenchmarkRunnerCore.Run(BenchmarkRunInfo benchmarkRunInfo, ILogger logger, String title, String rootArtifactsFolderPath, Func2 toolchainProvider, IResolver resolver, List1 artifactsToCleanup) at BenchmarkDotNet.Running.BenchmarkRunnerCore.Run(BenchmarkRunInfo benchmarkRunInfo, Func2 toolchainProvider)
```

Likely something not supported/correctly escaped when using a generic instance? (I was actually glad to see that I could run some benchmark with this - without the print disassembly)

[AndreyAkinshin]

Hello @xoofx, thanks for the bug report!
@adamsitnik could you take a look?

[adamsitnik]

@xoofx I have made a little research and some debugging and it looks like it's currently impossible to get the asm of generic type method. It's a ClrMD limitation, which we can not overcome.

Example:

namespace ConsoleApp7
{
    class Program
    {
        static void Main(string[] args)
        {
            var _int = new Generic<int>().Create();
            var _short = new Generic<short>().Create();
            var _object = new Generic<object>().Create();

            if (_int + _short * _object.GetHashCode() == 123)
                throw new Exception();
        }
    }

    public class Generic<T> where T : new()
    {
        public T Create() => new T();
    }
}

When I attach to such simple program ClrMD reports that there is only one type called Generic, it has method Create but this method is not compiled (CompilationType == None) and there is no address of the implementation (NativeCode == ulong.MaxValue)

Of course, this is how the generic works, but I would expect that it's going to report generics of value types as separate types Generic<int> Generic<short> and that the Generic type methods are going to be compiled for reference types.

This problem is described in ClrMD docs

Of course, this is how the generic works, but I would expect that it's going to report generics of value types as separate types Generic<int> Generic<short> and that the Generic type methods are going to be compiled for reference types.

I am going to provide a nice error message and try to get the workaround suggested by @nietras in #631 working (separate type per

[adamsitnik]

@xoofx I found the way to get what we need.

@AndreyAkinshin I added an extra, non-inlinable method to our auto-generated type. It's executed only once for jitting, when the diagnoser starts disassembling it starts with this method. In case of generic types the method contains address of the compiled generic method, and once we have this address the ClrMD is able to recognize the type.

After this build is finished the new package on our CI will contain the fix.

[nietras]

Awesome!

On Feb 4, 2018 15:45, "Adam Sitnik" [e-mail] wrote:

@xoofx https://github.com/xoofx I found the way to get what we need.

@AndreyAkinshin https://github.com/andreyakinshin I added an extra,
non-inlinable method to our auto-generated type. It's executed only once
for jitting, when the diagnoser starts disassembling it starts with this
method. In case of generic types the method contains address of the
compiled generic method, and once we have this address the ClrMD is able to
recognize the type.

After this
https://ci.appveyor.com/project/dotnetfoundation/benchmarkdotnet/build/0.10.12.420
build is finished the new package on our CI will contain the fix.

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
https://github.com/dotnet/BenchmarkDotNet/issues/640#issuecomment-362911715,
or mute the thread
https://github.com/notifications/unsubscribe-auth/AKTG75oS82mUtQLcsMWjC_e4nBvMEAKTks5tRcKAgaJpZM4R2Pru
.

[xoofx]

@xoofx I found the way to get what we need.

thanks a lot!

[AndreyAkinshin]

@adamsitnik,

it's currently impossible to get the asm of generic type method

I found the way to get what we need.

It's always good to see when people are not afraid of impossible problems. =)

[adamsitnik]

@xoofx @nietras the build I referenced last time did not pass, but this one did, so 0.10.12.423 is the version for you

It's always good to see when people are not afraid of impossible problems. =)

I hate to give up, and this kind of problems make me very motivated to solve them.