Strange segmentation fault

Question

I've defined a random function (int random(int sup, int seed)) which returns a value between 0 and sup-1. I've defined a struct, point, of which pos_parents and population are 2-dimensional arrays. The swap functions swap elements of the v array, which is a array of "indexes". All is done in order to sort out par_n members into pos_parents out of population members without sorting twice the same member.

This gives segmentation fault.

If I replace the variable r inside population[v[r]][j] with an explicit value, then it all functions. How is this possible? I've tried the random function and it doesn't seem to have any problem.

In addition, when segmentation fault occours, printf won't even activate for the first loop.

point population[pop_size][array_size];
point pos_parents[4*par_n][array_size];
int v[pop_size];

for (i=0; i<4*par_n;i++)
    v[i]=i;

for(t=0;t<time_limit;t++) //The cycle of life
{
    for(i=0;i<4*par_n;i++)
    {
        r=random(pop_size-i,i);
        printf("%d\t",r);

        for(j=0;j<array_size;j++)
        {
            pos_parents[i][j]=population[v[r]][j];
        }
        swap(&(v[r]),&(v[pop_size-1-i]));
    }

When executing i type 3(route locations-array size), 8(pop_size), 1(time limit), 1 (par_n)

This is the entire code (less than 150 lines), always insert 1 to time_limit, because I haven't still completed the cycle. https://docs.google.com/open?id=0ByylOngTmkJddVZqbGs1cS1IZkE

P.S. I'm trying to write an evolutionary algorithm, for route optimization

The first thing you should do when getting a segmentation fault, or other crash, is to run your program in a debugger. This will help you locate the place of the crash, as well as let you examine variables to help you see what might have caused the crash. — Some programmer dude
In your case you are probably accessing indexes outside one of the arrays. — Some programmer dude
Right. In the above code, I'd guess that the size of v[] is less than 4*par_n. So the first for-loop is overwriting something important in the stack frame. — Barmar
Beyond the debugger suggesion of @JoachimPileborg, you have not posted enough code to determine what might be happening. It's almost certain you are touching an array location that doesn't exist, but with no declarations (or even better a runnable example), we can't help. — Gene
If you've got par_n members, why are you using 4 * par_n for the upper bounds on your arrays? That could be a major part of your trouble. The printf() typically won't send anything to the screen until a newline is printed, or until the print buffer is full. For debugging, either print to stderr, or print a newline (and optionally use fflush(0) to flush the output. I worry about you providing the seed to the random function each time. For the rest, please consider providing a SSCCE — a Short, Self-Contained, Correct (Compiling) Example that reproduces the problem. — Jonathan Leffler

Kerrek SB Kerrek SB · Accepted Answer · 2012-08-18T17:33:21

The loop with v[i] = i; goes from 0 to 4 * par_n, but v is an array of size pop_size. That looks like an out-of-bounds problem waiting to strike. And the same again for the counter i in r = random(pop_size - i, i);, since i is used in v[i].

Strange segmentation fault

1 Answers