Hi I'm trying to do dynamic memory allocation of a large matrix in C but I'm running into the following error:

Exception thrown at 0x00007FF63A248571 in cdempd.exe: 0xC0000005: Access violation writing location 0x0000000000000000. occurred

sometimes it's Access violation writing location 0xFFFFFFFFB412E2A0.

```
double ndivx, ndivy, ndivz, nt, r, box, dx, totnode;
int main()
{
ndivx = 19.0;
ndivy = 19.0;
ndivz = 19.0;
int totnode = ndivx * ndivy * ndivz;
r = 0.005; //diameter of sphere
dx = 0.0025 / ndivx;
double dx = r / ndivx; // distance between points
int cols = 3;
int** coords;
coords = malloc(totnode * sizeof(int*));
for (int i = 0; i < totnode; i++) {
coords[i] = malloc(cols * sizeof(int));
}
//int* coord = (int*)malloc(totnode * cols * sizeof(int));
// int offset = i * cols + j;
// now mat[offset] corresponds to m(i, j)
//create a cube of equidistant points
int numm = 0;
for (int i = 1; i <= ndivx; i++)
{
for (int j = 1; j <= ndivy; j++)
{
for (int k = 1; k <= ndivz; k++)
{
coords[numm][0] = -1.0 / 2.0 * (r)+(dx / 2.0) + (i - 1.0) * dx;
coords[numm][1] = -1.0 / 2.0 * (r)+(dx / 2.0) + (j - 1.0) * dx;
coords[numm][2] = -1.0 / 2.0 * (r)+(dx / 2.0) + (k - 1.0) * dx;
numm = numm + 1;
}
}
}
}
```

pd.r is a double 0.005, dx is a double about 0.00026315, totnode is 6859.

I've tried two methods, the one that is there and the one commented out with //. Both give me the same error. I'm using visual studio 2019. I'm not so familiar with c and visual studio so forgive me if the question is silly. Any help would be appreciated thank you.

Aside from some of the other errors [after correction], *all* values of `coords`

are set to *zero*. This is because `coords`

is a pointer to `int`

and *not* (e.g.) `double`

and your equation uses `-1.0 / ...`

which will always produce a fraction.

Also, as David pointed out, you're indexing from 1 [vs. 0] in the `for`

loops. This could cause access violations/segfaults.

I've changed the `for`

loops to start from 0. And, I've adjusted the equation accordingly (using a macro).

You were defining some things like index variables or size variables as `double`

instead of `int`

(e.g.) `ndivx`

Also, I introduced a `typedef`

for the coordinate values.

Here's some cleaned up code that may help get you further:

```
#include <stdio.h>
#include <stdlib.h>
#if 0
double ndivx, ndivy, ndivz, nt, r, box, dx, totnode;
#endif
#if 0
typedef int coord_t;
#else
typedef double coord_t;
#endif
#define SETCOORD(_xidx,_var) \
do { \
coords[numm][_xidx] = -1.0 / 2.0 * r + (dx / 2.0) + (_var * dx); \
printf("coords[%d][%d]=%g\n",numm,_xidx,(double) coords[numm][_xidx]); \
} while (0)
int
main(void)
{
#if 1
int ndivx;
int ndivy;
int ndivz;
double r;
double dx;
#endif
ndivx = 19;
ndivy = 19;
ndivz = 19;
int totnode = ndivx * ndivy * ndivz;
r = 0.005; // diameter of sphere
dx = 0.0025 / ndivx;
#if 0
double dx = r / ndivx; // distance between points
#else
dx = r / ndivx; // distance between points
#endif
int cols = 3;
#if 0
int **coords;
#else
coord_t **coords;
#endif
coords = malloc(totnode * sizeof(coord_t *));
for (int i = 0; i < totnode; i++) {
coords[i] = malloc(cols * sizeof(coord_t));
}
// int* coord = (int*)malloc(totnode * cols * sizeof(int));
// int offset = i * cols + j;
// now mat[offset] corresponds to m(i, j)
// create a cube of equidistant points
int numm = 0;
for (int i = 0; i < ndivx; i++) {
for (int j = 0; j < ndivy; j++) {
for (int k = 0; k < ndivz; k++) {
SETCOORD(0,i);
SETCOORD(1,j);
SETCOORD(2,k);
numm = numm + 1;
}
}
}
return 0;
}
```

answered on Stack Overflow Mar 10, 2020 by Craig Estey

User contributions licensed under CC BY-SA 3.0