fast_array_write.c

#include <setjmp.h>
#include <signal.h>
#include <stdio.h>
#include <stdlib.h> // EXIT_SUCCESS
#include <unistd.h> // sbrk

struct datum {
  int v1;
  int v2;
  int v3;
};

static jmp_buf env;
struct datum *data;

void on_segfault(int signum) {
  puts("signal rec'd");
  longjmp(env, 1);
}

int main(void) {
  struct sigaction action;

  action.sa_handler = on_segfault;
  sigemptyset(&action.sa_mask);
  action.sa_flags = 0;
  sigaction(SIGSEGV, &action, NULL);

  if (!setjmp(env)) {
    data = sbrk(sizeof(struct datum) * 1000);
    if (data == -1) {
      puts("sbrk failed");
      exit(EXIT_FAILURE);
    }
    struct datum *it = data;
    // Write data to array
    for (;; it++) {
      it->v1 = 1;
      it->v2 = 2;
      it->v3 = 3;
    }
  }
  puts("Success, printing data:");
  for (struct datum *it = data; it < (data + 1000); it++) {
    printf("%d %d %d", it->v1, it->v2, it->v3);
  }
  return EXIT_SUCCESS;
}

Fills data buffer with integers 1, 2, 3, 1, 2, 3 ...
Uses segfault signal instead of loop termination condition.

loool