assembly toy problem

less.asm

global main
extern printf

SECTION .text

small_enough:
  mov rax, 1
  add rsi, 1
  ;rdi array, rsi array length, rdx limit
loop:
  ;compare current element of array to limit
  mov r8, [rdi]
  ;move rdi to point to next element
  add rdi, 8

  ;r8 is current element
  cmp r8, rdx
  jg fail
  ;check loop condition
  sub rsi, 1
  cmp rsi, 1
  jne loop

  ;unless you jump to fail, we should skip it
  jmp return
fail:
  mov rax, 0
return:
  ret

main:
  mov rdi, array
  mov rsi, 2
  mov rdx, 200
  call small_enough
  push rax
  mov rdi, format
  mov rsi, [rsp]
  xor rax, rax
  call printf
  pop rax
  ret

format: db "%d", 10, 0
array: dq 66, 101

This works on my machine (prints "1") but on CodeWars, I fail the test. No idea why. Here is the problem on CodeWars: https://www.codewars.com/kata/57cc981a58da9e302a000214/train/nasm

One possibility is that in my local code I recreate the test data using the data section. But in the test, it's an array literal which would be on the stack. Maybe those work differently. Also they define the type as int. My local numbers are 64 bit, but I don't know if that's true of C ints. In fact, I believe this is the problem.

Based on this StackOverflow discussion: https://stackoverflow.com/questions/11438794/is-the-size-of-c-int-2-bytes-or-4-bytes

The size of a C int is almost certainly not 8 bytes on CodeWars. So it'll be 2 or (more likely) 8 bytes. But changing the iteration amount won't be enough, I'll also have to use 4 byte registers to store these 4 byte numbers in, which explains what some other people on CodeWars were doing. Interesting!