Calculating the size in bytes of a C++ function

[evlncrn8]

exports, is a quick and dirty way...

make an export above your proc 'proc start'
make an export below your proc 'proc end'
end - start = size

tons of ways to do it really

Above and below have very relative meanings ...
Above and below where? In source code? And what are you suggesting, to export a function before and one after and calculate the difference to get the size?

How about sharing someone else of those *tons of ways*?


yaa

[Darren]

void main()
{
do whatever main function does
}

void dummy()
{
}

#define mainsize ((DWORD)dummy-(DWORD)main)


This will calculate size of main function

But this is not true. It depends entirely on how your compiler and linker lay out your code in the binary. Without doing anything special and without touching optimization flags, taking your example, I even got to the point that dummy's code was placed BEFORE main's code in the compiled binary!

yaa

[Darren]

Originally Posted by yaa

But this is not true. It depends entirely on how your compiler and linker lay out your code in the binary. Without doing anything special and without touching optimization flags, taking your example, I even got to the point that dummy's code was placed BEFORE main's code in the compiled binary!

yaa

Maybe this is correct when dealing with the main function, but I've used many times this example below and found it to work fine, but i do always compile it from a seperate .cpp this method can be useful for when adding complicated hooks or patches to a piece of code and placing the new function in a cave or new section

void funct1()
{
    funct1code...
}

void funct2()
{
    funct2code...
}

void funct3()
{
    funct3code...
}

void dummy()
{
}

#define funct1size ((DWORD)funct2-(DWORD)funct1))
#define funct2size ((DWORD)funct3-(DWORD)funct2))
#define funct3size ((DWORD)dummy-(DWORD)funct3))

[!k-0t1c!]

Your best bet, in my opinion, would be emitting different recognizeable byte sequences using

__asm { emit BYTE0; emit BYTE1; ...}

at the beginning and at the end of the function
This will however turn out to be a bit complicated for non-voids, as if you'll insert the sequence below the return instruction it'll get ignored, and if you'll insert it above the return you won't account for that...

Alternatively, if you don't need to dynamically get the value from time to time but you need it for processing after compilation of an executable, you can consider generating a mapfile for your compiled code and parse that.


Regards