Strange MinGw Windows runtime error

Attachment 15778I have this program:

Code:

---

#include <stdio.h>
#include <stdlib.h>
int main(void)
{
    int (*a)[5][5] = calloc(25, 1);
    *a[0][0] = 5;

    size_t i, j;
    for(i =0; i < 5; i++)
    {
        for(j = 0; j < 5; j++)
        {
            *a[i][j] = 25;
        }
    }
    printf("%d\n", *a[0][4]);
    free(a);

    return EXIT_SUCCESS;
}

---

When run, it just does nothing and exits. I used MinGW gdb as shown in the attached screenshot and I'm getting a SIGTRAP ntdll!RtlIsNonEmptyDirectoryReparsePointAllowed () error. At first I thought maybe some memory corruption was going on but when I examined the array, it is successfully populated and this crash happens AFTER the loop when the printf statement is run.

Does anyone have experience with this error or know what it may be?

Well you're not allocating anywhere near enough memory for a start.

Try
int (*a)[5][5] = calloc(25, sizeof(int));

Besides that, your repeated usage of *a[0][0] and the like is incorrect. It should be (*a)[0][0], and similiar. Your compiler should have warned you, and if it didn't, try compiling at a higher warning level.

Gah. Well strangely even with -Wall it still doesn't flag anything and builds just fine.

But that was the problem. Made those 2 changes and it actually builds and runs way faster, and works correctly now.

Oh, my bad. I forgot that that would look fine to the compiler: array index notation is ultimately syntactic sugar for pointer notation. It will just do the index computations as per normal, and as long as you happen to remain within the bounds, it's okay. It's just that the computed index isn't going to be what you expect, e.g., *a[i][j] is equivalent to **(*(a + i) + j), which means that for i > 0, you exceed the bounds of the array (this array isn't the array of 5 arrays of 5 ints; it's a hypothetical array of an unknown number of arrays of 5 arrays of 5 ints, of which this array you allocated is the "first element").

Put it another way: if we "flatten" this hypothetical 3D array into a 1D array, then *a[i][j] i.e., **(*(a + i) + j) would be equivalent to a[25 * i + 5 * j + 0], whereas you presumably want the equivalent of a[5 * i + j].

Therefore, my advice would be to not to mess with pointers to entire arrays unless you're really traversing across a multi-dimensional array using that pointer. If not for this pointer to a 2D array itself, we wouldn't have to consider this notion of a hypothetical 3D array.

Gotcha, thanks. Yeah sounds like a recipe for difficult-to-find bugs!

On another note, does anyone know more about

ntdll!RtlIsNonEmptyDirectoryReparsePointAllowed? I'm going to go and look into it but this isn't the first time I've gotten that. It always fascinates me when I get an error like that instead of a segfault. Basically it seems that I caused a segfault in a Windows library.