Segmentation fault when moving data to register

I’m trying to write a function that adds two numbers in assembly, and calling it through C. I am getting a segmentation fault when i mov eax, [ebp+8]. I know this because GDB says:

Program received signal SIGSEGV, Segmentation fault.
add () at add_two_numbers.s:12
12          mov eax, [ebp+8]

Here is my Assembly code:

    .code32
    .intel_syntax noprefix
    .global add

    .section .text
add:
    push ebp
    mov ebp, esp

    sub esp, 4

    mov eax, [ebp+8]
    add eax, [ebp+12]

    mov esp, ebp
    pop ebp
    ret

and my C Driver code:

#include <stdio.h>

int add(int, int);

int main(void)
{
    int x = 4, y = 10;
    int z = add(x, y);
    printf("%d", z);

    return 0;
}

I Compile my code using gcc -Wall -o main driver.c add_two_numbers.s

>Solution :

After reading the manual I am now really certain of my answer.

The problem is on a modern Linux, the default output of gcc is a 64 bit binary. While .code32 tells the assembler to create 32 bit code, it doesn’t change the output file format to mark it as a 32 bit .o finally so you don’t actually get the link error you would want.

32 and 64 bit code are close enough that particular code decoded correctly; but it can’t possibly work. The problem is rsp is much higher than the 4GB barrier so mov ebp, esp didn’t put the top of the stack into it, but a sheared-off pointer that points into nilspace.

The straightforward fix is to just go ahead and write 64 bit assembly.

    .code
    .intel_syntax noprefix
    .global add

    .section .text
add:
    push rbp
    mov rbp, rsp

    sub rsp, 16  # the stack frames are 16 byte aligned so if you're going to reserve stack space, do it right!

    # Arguments are in registers rdi, rsi, rdx, rcx, r8, r9
    mov eax, edi
    add eax, esi

    mov rsp, rbp
    pop rbp
    ret