Advice regarding what this assembler is doing MOVs for?

Question

This C++:

int my_func(int y, int t){
    int m = 0;
    m= y*t;
    return m;
}

int main(){
    int h = my_func(1,2);
    return 1;
}

Producing this x86:

int main(){
push        ebp  
mov         ebp,esp  
sub         esp,8  
    int m = my_func(1,2);
mov         dword ptr [ebp-4],0  
mov         eax,1
imul        eax,eax,2  
mov         dword ptr [ebp-4],eax     ;Why?
mov         ecx,dword ptr [ebp-4]     ;Why?
mov         dword ptr [m],ecx         ;Why?
    return 1;
mov         eax,1  
}
mov         esp,ebp  
pop         ebp  
ret

I don't understand why three movs are required when the value of m is already contained in register eax, after the imul?

Is it because eax has to be used for the return 1 line and eax is therefore a special register?

Also, what does dword ptr [ebp-4] refer to exactly? Is it the 1 or the 2?

Answer 1

[ebp-4] is where my_func's local copy of m is stored on the stack. [m] is where main's local copy of m is stored. What's missing here is the definition for main's copy of m. I would have expected it to also be an offset from ebp. Using VS2005, I compiled this example, using volatile to prevent optimizing the muliplty away.

__inline int myfun(int a, int b)
{
int m;
    m = a*b;
    return (m);
}

int main (void)
{
volatile int a, b, m;
    a = 1;
    b = 2;
    m = myfun(a, b);
    return(0);
}

and got this assembly code. Note that m and a share the same memory location: -4[ebp+8]:

_b$ = -8                                                ; size = 4
_m$ = -4                                                ; size = 4
_a$ = -4                                                ; size = 4
_main   PROC
        sub     esp, 8
        mov     DWORD PTR _a$[esp+8], 1
        mov     DWORD PTR _b$[esp+8], 2
        mov     ecx, DWORD PTR _b$[esp+8]
        mov     eax, DWORD PTR _a$[esp+8]
        imul    eax, ecx
        mov     DWORD PTR _m$[esp+8], eax
        xor     eax, eax
        add     esp, 8
        ret     0