Managed Vs. Unmanaged Methods/Transitions

In .Net Framework CLR provides execution of instruction targeted for CLR and the instructions that are not targeted for CLR. The instructions targeted for CLR are called as managed code and other type of code is called unmanaged code. After going through this topic you will know the following things mentioned below

Difference between Managed Method and Unmanaged Method.

Difference between Managed Type and Unmanaged Type.

How to call unmanaged methods in managed methods.

How to use unmanaged types.

When an application is launched for execution, first request is given to the Operating System, the OS will load the executable file in memory and starts executing the instruction from the entry point function in the executable file. Where in .NET executable file contains four main components the CLR header, the Metadata, the MSIL Code, and Native code.

CLR header will be used by the managed code in the module which will have the version number of the CLR on which the module is built and entry point method of the module in the executable.

Metadata describes the types used in the managed code, combination of CLR header and MSIL Code is the compiled format of a .Net language on a .Net Compiler, which will not have the instruction in targeted machine instruction format, which will again get compiled by the JIT Compiler

Native Code contains the machine instruction, which will be directly executed by the OS. Not all the .NET PE will have the Native code. PE of type EXE’s will be having a native method like main() called as “unmanaged stub” which will be an entry point for the OS to execute code, that function will jump to _CorExeMain function located in MSCoree.dll which will be executed by the OS to initialize CLR and attach the running .NET module to CLR. Once CLR is initialized and loaded CLR will start executing the assembly by executing the managed entry point function specified in the CLR header of the file.

Managed Code
Machine instructions in MSIL format and located in Assemblies will be executed by the CLR, will have the following intrinsic advantages,
Memory management to prevent memory leaks in the program code,

• Thread execution,
• Code safety verification,
• Compilation,

Executed on many platforms like Windows 95, Windows 98, Windows 2000,and other system services.
Managed methods will be marked as “cil” in MSIL code. Methods marked with “cil” will be compiled by mscorjit.dll before execution. C# and VB.NET will generate only managed code. (Managed C++ can generate managed code by specifying “#pragma managed”)

Unmanaged Code

Unmanaged codes are the instructions, which are targeted for specific platforms.
Unmanaged code will exist in any of the format,
A code instructions in managed C++ with “#pragma unmanaged”
COM/COM+ Components
Win32 Dlls/System Dlls

As these codes are in native format of OS, these instructions will be executed faster compared with JIT compilation and execution of Managed code.

Managed and Unmanaged transitions

As we get more benefits from managed code, still we may need to use unmanaged code. .Net provides many ways to access unmanaged code in managed code. Managed- Unmanaged transitions are achieved in .Net by set of services called Platform Invocation Services (P/Invoke) and IJW(It Just Works).

P/Invoke services are targeted for unmanaged code, which exists as COM/COM+ components and Win32 DLLs. COM/COM+ components will accessed by the concept called COM Interop - is a mechanism in which existing COM components will be accessed through a wrapper class called COM Callable Wrapper (CCW) in managed code without modifying the existing COM components. Using P/Invoke mechanism we can call Windows DLL functions in managed code.

IJW(It Just Works) targets code instructions built on C++ managed extensions, This mechanism is only for the code in Managed C++. In this way we can call the unmanaged methods directly by the managed code.

For example following code calls the MessageBox function in User32.dll(VB.NET).

CODES:

Declare is the statement to state the Win32 API functions located in the Win32 DLLs. And the respective arguments declared with CLR data type. In C#, we need to use the extern keyword with the attribute DLL Import to specify the Win32 DLL and the function should be declared as static function.

CODES:

In the above example MessageBox function is accessed from user32.dll by using the attribute DLL Import, and declared as static function Managed Types and Unmanaged Types We have seen how to call an unmanaged code in a managed code, now the question is how unmanaged code understands the managed data type and vice the versa. We will see how string will be sent from managed code and returned back to managed code. When passing a string value as an input argument to an unmanaged code CLR will take care of converting that to a native string type. When we try to call a function, which returns string then the managed code, has to allocate memory and send to the function in unmanaged code. For example if we want to retrieve the OS Temp path then we can call GetTempPath API function of “kernel32.dll”, Following code snippet shows how to call the function with string as an out argument with VB.NET.

CODES:

Following code snippet shows how to call the function with string as an out argument with C#.

CODES:

The above code uses StringBuilder class of System.Text namespace to allocate string with 255 characters (Just think for this as a fixed string we used to have in VB). Some of the functions in managed code will be accepting arguments as structures, structures in unmanaged code will be having set of member fields located in memory as the order in which it has been declared, that is the layout of member variable location is fixed. But in .Net, structures will have fields of managed data type and these member fields will automatically change the memory location of the structure. CLR will automatically move the data members to improve memory usage and performance. When an unmanaged method, which expects an argument as a structure, then managed code has to declare the structure so that it can be accessed by the unmanaged code. But .NET structures will have auto memory layout of data members, so to pass structures from managed code to unmanaged code has to be declared with an attribute StructLayout in System.Runtime.InteropServices namespace. StructLayout is used with an enumerated value LayoutKind with following options given below:

• Auto – default option which changes the member field memory layout.
• Sequential - specifies member variable should be placed in a sequential order as specified while declaring the type.
• Explicit- specifies the exact location of the member variable in the structure.