Issues while manually increasing address of a pointer with inline assembly (gcc)

Question

my apologies for the rookie question. I am diving into assembly a bit more and I am trying to use some inline asm on a small library I wrote to do linear algebra. Everything works fine but I am having issues increasing the address pointed by the pointers. I need this step to be able to multiply elements of a matrix.

void multiply(int n, int m, int* ptrM1, int* ptrM2, int* result)
{
    int counter = n*n*m;
    int index = 1;
    int inter = 0;
    while (index != (counter+1))
    {
        asm 
        (
            "movl %2, %%eax\n\t"
            "mull %3\n\t"
            "movl %%eax, %0\n\t"
            "addl %0, %1\n\t"
            "addq $0x04, (%2)\n\t"      //<-----Problematic lines
            "addq $0x04, (%3)\n\t"      //<-----Problematic lines
            :"+c"(inter),"+b"(*result)
            :"r"(*ptrM1), "r"(*ptrM2)
            :"%eax"
        );
        printf("%d  ", *result);
        //++ptrM1;                      //Want to do this in assembly
        //++ptrM2;                      //Same
        if ((index % m ) == 0){++result;}
        ++index;
    }
}

The program compiles as it is, but I get a core dumped (segmentation fault) error when I try to run it. I suspect it has to do with the syntax I am using in the lines I commented above. I am also suspicious about the constraints I gave to the two variables since they both perform as input and output but are only declared as inputs. The thing is, I literally copied how gcc handled the "++pointer_address" in the disassembled code. Anybody that can help me out? It would be much appreciated.

EDIT: Excellent tips in the comments so far, but still don't fix the problem. Here is how I implemented them:

void multiply(int n, int m, int* ptrM1, int* ptrM2, int* result)
{
    int counter = n*n*m;
    int index = 1;
    int inter = 0;
    while (index != (counter+1))
    {
        asm 
        (
            "movl %4, %%eax\n\t"
            "mull %5\n\t"
            "movl %%eax, %0\n\t"
            "addl %0, %1\n\t"
            "addq $0x04, %2\n\t" //Problematic lines
            "addq $0x04, %3\n\t" //problematic lines
            :"+c"(inter),"+b"(*result), "+rm"(ptrM1), "+rm"(ptrM2)
            :"r"(*ptrM1), "r"(*ptrM2)
            :"%eax"
        );
        printf("%d  ", *result);
        //++ptrM1;                  //Want to do this in assemlby
        //++ptrM2;                  //Same
        if ((index % m ) == 0){++result;}
        ++index;
    }
}

EDIT 2: And this is the original function written in C, for comparison.

void multiply(int n, int m, int* ptrM1, int* ptrM2, int* result)

{
    int counter = n*n*m;
    int index = 1;
    int inter = 0;
    while (index != (counter+1))
    {
        if ((index % m ) == 0)
        {
            inter = *ptrM1 * *ptrM2;
            *result += inter;
            ++ptrM1;
            ++ptrM2;
            ++result;
        }
        else
        {
            inter = *ptrM1 * *ptrM2;
            *result += inter;
            ++ptrM1;
            ++ptrM2;
        }
        ++index;
    }
}

Answer 1

So thanks to everybody who commented. The final working function is the following:

void multiply(int n, int m, int* ptrM1, int* ptrM2, int* result)
{
    int counter = n*n*m;
    int index = 1;
    int inter = 0;
    while (index != (counter+1))
    {
        asm 
        (
            "movl %4, %%eax\n\t"
            "mull %5\n\t"
            "movl %%eax, %0\n\t"
            "addl %0, %1\n\t"
            "addq $0x04, %2\n\t" 
            "addq $0x04, %3\n\t" 
            :"+c"(inter),"+b"(*result), "+rm"(ptrM1), "+rm"(ptrM2)
            :"r"(*ptrM1), "r"(*ptrM2)
            :"%eax", "%edx"
        );

        if ((index % m ) == 0){++result;}
        ++index;
    }
}

So to summarize, I was both using the pointer wrong and not clobbering a cluttered register. The solution consisted in adding the de-deferenced pointers into the outputs with "+rm" constrains, and clubbering both eax and edx since mull will store a 64 bit result in edx:eax .