CODEWITHSUNDEEP
assemblyx86-64nasm
Ian Macpherson
Ian Macpherson

Reputation: 11

NASM: segmentation fault(core dumped)

I'm new to assembly and I've been trying to learn through random problems. Im using nasm 64 bit on a linux machine. I've been trying to generate the fibonacci sequence. However I get a segmentation fault(core dumped) error when I run my executable.

section .data

    input db 2

section .bss

    fib resb 128

section .text

_start:

    mov rax, 1
    mov rcx, 1
    mov rdx, fib
    mov rdx, 1
    inc rdx
    mov rbx, 0

    call _fibLoop
    call _fibPrint

    mov rax, 60
    mov rdi, 0
    syscall

_fibLoop:

    mov [rdx], rax
    inc rdx
    add rcx, rax
    push rcx
    mov rcx, [rax]
    pop rax
    inc rbx
    cmp rbx, [input]
    jne _fibLoop
    ret

I know the call to _fibPrint is not the issue cause it does virtually nothing. I assume the way I am writing to my reserved memory is flawed. However Ive been able to do this similarly in the past so I'm lost on whats wrong.

Upvotes: 0

Views: 1185

Answers (1)

1201ProgramAlarm
1201ProgramAlarm

Reputation: 32727

As part of your initialization, you have this sequence:

mov rdx, fib
mov rdx, 1
inc rdx

This will leave rdx with the value 2, and not the offset of a buffer to hold your numbers. Then, at the start of _fibLoop, you write to it with

mov [rdx], rax

This will try to access memory that you cannot access, resulting in the segmentation fault.

I think removing the extra two lines after mov rdx,fib will fix that crash, letting you move on to debugging the other bugs using GDB or whatever other debugger you like.

(Like that inc rdx in _fibLoop only advances the pointer by 1 byte, but you're doing 8 byte stores. And that input is also only 1 byte, but you're doing an 8-byte load there, too.)

Upvotes: 3

Related Questions