C# internals

Question

Hi

At the lowest layer im aware that every CPU has an Instruction set ie a list of Opcodes written in either Binary or Hex to be able to carry out various tasks.

Ive had a look at the MSDN documentation for the additional Opcodes within the Reflection.Emit class and they look like they are programmed in Binary or Hex ie

Conv_I4 format or value 69

Conv_I format or value 0xD3

Would this be all that is needed to be able to compile to native code for the CPU to understand as it obviously Hex and Binary?

Regards

Steven

Vulpes · Answer

Those are in fact MSIL (Microsoft Intermediate Language) opcodes. They are what the C# compiler (or any other .NET compiler) produces. They are converted 'on the fly' by the CLR's JIT (Just in Time) compiler to native code for whatever platform the application is running on.

Vulpes · Answer

Compiler writing is very specialized and you need a certain sort of mind to do it well. I've always been a high level programmer though I do have some interest in what goes on 'under the hood' :)

S · Answer

Thanks Vulpes have you ever done any programming this low level given your C++ background?

Vulpes · Answer

Well the Visual C# compiler and the CLR (including the JIT) are written in C++ so yes you're right that the actual code which converts C# to MSIL and the MSIL to native code is written in C++.

S · Answer

So the actual code which does the converting or translating from C# to native code which a CPU understands is programmed in C++ is that correct?

Vulpes · Answer

Well the values of these opcodes aren't the same as the values of the native instructions that will be generated by the JIT. A single MSIL opcode may need to be converted to several native instructions as MSIL is a relatively high level assembler. The CPU itself only understands the native instructions it was designed to cater for.

S · Answer

And its the values in Hex or Binary which each of these Opcodes have to allow the conversion to native code that the CPU understands?