I came across the following code on msdn:
unsafe static void SquarePtrParam (int* p)
{
*p *= *p;
}
unsafe static void Main()
{
Point pt = new Point();
pt.x = 5;
pt.y = 6;
// Pin pt in place:
fixed (int* p = &pt.x)
{
SquarePtrParam (p);
}
// pt now unpinned.
Console.WriteLine ("{0} {1}", pt.x, pt.y);
}
I am just wondering, we are directly accessing pointer in SquarePtrParam
function, does it inherit information that array is fixed from calling method?
Why don't we need to explicitly set it to fixed
locally in SquarePtrParam
.
I guess I could use some elaborations about this fixed
statement.
Fixed statement implement unpin memory area in the same manner as "using" statement close opened files in using(FileStream stream = new FileStream(..)) construction. Memory will be pinned until you left the fixed code block.
In the IL code it will create dummy PINNED local variable and store a pointer into it. This will not allow GC move memory area contains this pointer. After you will left fixed block it will store zero into this PINNED variable. Just like this:
public static unsafe void TestInternal(byte* pointer)
{
Console.WriteLine((IntPtr)pointer);
}
public static void FixedDemo()
{
Byte[] newArray = new Byte[1024];
unsafe
{
fixed (Byte* pointer = &newArray[0])
{
TestInternal(pointer);
}
}
Console.WriteLine("Test Complete");
}
So FixedDemo in IL Code:
.method public hidebysig static void FixedDemo() cil managed
{
// Code size 47 (0x2f)
.maxstack 2
.locals init ([0] uint8[] newArray,
[1] uint8& pinned pointer)
IL_0000: nop
IL_0001: ldc.i4 0x400 // Put 1024 on the stack
IL_0006: newarr [mscorlib]System.Byte // allocate new array of 1024 length
IL_000b: stloc.0 // Store it in local variable 0
IL_000c: nop
IL_000d: ldloc.0 // Put local variable 0 on the stack
IL_000e: ldc.i4.0 // Put zero on the stack
IL_000f: ldelema [mscorlib]System.Byte // Load address of zero index from array
IL_0014: stloc.1 // !!! Here we pin memory by storing it in pinned variable
IL_0015: nop
IL_0016: ldloc.1 // Load function argument
IL_0017: conv.i // Perform conversion
IL_0018: call void FinMath.Tests.Program::TestInternal(uint8*)
IL_001d: nop
IL_001e: nop
IL_001f: ldc.i4.0 // Load zero on the stack
IL_0020: conv.u // Perform conversion
IL_0021: stloc.1 // !!!! Here we unpin memory
IL_0022: nop
IL_0023: ldstr "Test Complete" // Load string
IL_0028: call void [mscorlib]System.Console::WriteLine(string) // Out message
IL_002d: nop
IL_002e: ret
} // end of method Program::FixedDemo
For more information visit:
MSDN
MSDN Magazine: Garbage Collection: Automatic Memory Management in the Microsoft .NET Framework
Common language Runtime standard ECMA 335 Partition III, 1.1.4.2 Managed Pointer (type &)
does it inherit information that array is fixed from calling method?
No, it doesn't need that information. All it knows is that it's been passed a pointer. One way of obtaining a pointer is via the fixed
statement, but there are other ways (e.g. by converting an IntPtr), and any such compatible pointers could also be passed to SquarePtrParam
.