Question

I’ve wrote a MIPS subroutine that implements the merge sort algorithm (the code is in the end of the post). It receives the pointer to the array, and its size. Sort it, and don’t return anything.

I’ve been debugging it, fixed some errors, and since the subroutine seems to work well until it reaches the base case (array of size 1), now I’m using gdb and this C code to debug it for that specific situation:

#include <unistd.h>
#include <stdlib.h>
#include <stdio.h>
#include <string.h>

extern void merge_sort(char*, int);

int main(int argc,char **argv){

    char* vector;
    vector = (char*)malloc(sizeof(char));
    if(vector == NULL) printf("error");

    size_t vector_size = 1;
    memcpy(vector,"5",vector_size);

    merge_sort(vector, vector_size);

    int i;
    for(i = 0; i < vector_size; i++){
            printf("%c ", vector[i]);
    }

    printf("\n");
    return 0;
}

My problem is that when I’m stepping through the subroutine lines, when I get to the penultimate line (jr ra), it crashes and exit with code 060, and gdb show this message:

Warning: GDB can't find the start of the function at 0x400730.
    GDB is unable to find the start of the function at 0x400730
    and thus can't determine the size of that function's stack 
    frame. This means that GDB may be unable to access that stack
    frame, or the frames below it.

    This problem is most likely caused by an invalid program 
    counter or stack pointer.  However, if you think GDB should
    simply search farther back from 0x400730 for code which looks
    like the beginning of a function, you can increase the range 
    of the search using the `set heuristic-fence-post' command. 
0x004008a0 in _start

I can’t figure out why that happens. I think it could be either a problem with the address stored in ra, a bad stack allocation, or wrong linkage between the C and MIPS code, but I have no idea about what I’m doing wrong.

Here is the MIPS code of merge_sort:
For the base case, the code runs up to the line beq t1, zero, SALIDA, then jumps to SALIDA, and finishes. So I just paste that parts of code (SALIDA means exit in spanish :P).

#include <mips/regdef.h>
#include <sys/syscall.h>

#define     SSIZE   (56)
#define     O_RA    (48)
#define     O_FP    (44)
#define     O_GP    (40)
#define     O_S3    (36)
#define     O_S2    (32)
#define     O_S1    (28)
#define     O_S0    (24)
#define     O_ARG0  (SSIZE)
#define     O_ARG1  ((SSIZE) + 4)

.text
.align  2
.globl  merge_sort
.ent    merge_sort

merge_sort:
.frame $fp, SSIZE, ra
.set noreorder
.cpload t9
.set reorder     

subu    sp,sp,SSIZE

sw  s0, O_S0(sp)
sw  s1, O_S1(sp)
sw  s2, O_S2(sp)
sw  s3, O_S3(sp)

sw  gp, O_GP(sp)
sw  $fp, O_FP(sp)
sw  ra, O_RA(sp) 
move    $fp, sp

sw  a0, O_ARG0($fp)
sw  a1, O_ARG1($fp)

lw  t0, O_ARG1($fp)
addi    t1, t0, -1

beq     t1, zero, SALIDA

SALIDA:     
lw  s0, O_S0($fp)
lw  s1, O_S1($fp)
lw  s2, O_S2($fp)
lw  s3, O_S3($fp)

move    sp, $fp
lw  gp, O_GP($fp)
lw  $fp, O_FP($fp)
lw  ra, O_RA($fp) 

addiu   sp, sp, SSIZE

jr  ra

.end    merge_sort

Answer 1

I think your problem is the following near the end of the subroutine:

lw  $fp, O_FP($fp)    // restores $fp
lw  ra, O_RA($fp)     // restore ra, but $fp isn't pointing to our frame anymore

I think it might work better if you reverse those instructions.