multithreading - occasional segfault in matrix multiplication in c - c

Question

I'm working on a small program of multithreaded matrix multiplication. My first job is to fill the entry of matrices with a random integer. I met some segment faults after I tried to pass a function pointer to pthread_create. And I think the problem is in function pthread_join.
But there are two issues in general.
The first one is the segment fault does not happen every time. Sometimes the code works, but most of the times it doesn't. So it really confuses me.
The other one is when the code is working, there are always several entries still not initialized, especially for matrix[0][0], it is never initialized. And I don't quite know where to debug that one.
Here's my code:
#include <stdio.h>
#include <string.h>
#include <stdlib.h>
#include <time.h>
#include <unistd.h>
#include <pthread.h>
#define N 5
#define MAX 10
int A[N][N];
int B[N][N];
int C[N][N];
pthread_t pid[N][N];
typedef struct {
int row, col;
} Pos;
typedef void* (*thread_func)(void*);
void print_matrix(int M[][N]) {
int i, j;
for (i = 0; i < N; i++) {
for (j = 0; j < N; j++) {
printf("%3d", M[i][j]);
if (j < N - 1) {
printf(", ");
}
}
printf("\n");
}
}
void join_threads(void) {
int i, j;
for (i = 0; i < N; i++) {
for (j = 0; j < N; j++) {
pthread_join(pid[i][j], NULL);
}
}
}
void* fill_entry(void* arg) {
Pos* pos = (Pos*)arg;
A[pos->row][pos->col] = rand() % MAX;
B[pos->row][pos->col] = rand() % MAX;
return NULL;
}
void dispatch_jobs(thread_func job_func) {
int i, j;
for (i = 0; i < N; i++) {
for (j = 0; j < N; j++) {
Pos pos;
pos.row = i;
pos.col = j;
if (pthread_create(&pid[i][j], NULL, job_func, (void*)&pos)) {
perror("pthread_create");
exit(-1);
}
}
}
}
int main(void) {
srand(time(NULL));
dispatch_jobs(&fill_entry);
join_threads();
printf("Matrix A:\n");
print_matrix(A);
printf("Matrix B:\n");
print_matrix(B);
return 0;
}

Answer 1

Pos pos;
pos.row = i;
pos.col = j;
if (pthread_create(&pid[i][j], NULL, job_func, (void*)&pos)) {
perror("pthread_create");
exit(-1);
}
You are passing a pointer to a local variable to the threads. Once the thread tries to access the data, i.e. dereferences the pointer, the variable is long gone, reused, and contains garbage data.