When a parent process forks, we understand that the child is created; it is a copy of the parent. This will include the VAS, and, thus, the data and stack segments. Keeping this fact in mind, check out the following code snippet (ch10/fork5.c):

static int g=7;
[...]
int main(int argc, char **argv)
    [...]
    int loc=8;
    switch((ret = fork())) {
    case -1 : FATAL("fork failed, aborting!
");
    case 0 : /* Child */
          printf("Child process, PID %d:
", getpid());
          loc ++;
          g --;
          printf( " loc=%d g=%d
", loc, g);
          printf("Child (%d) done, exiting ...
", getpid());
          exit(EXIT_SUCCESS);
    default : /* Parent */
    #if 1
          sleep(2); /* let the child run first */
    #endif
          printf("Parent process, PID %d:
", getpid());
          loc --;
          g ++;
          printf( " loc=%d g=%d
", loc, g);
    }
    printf("Parent (%d) will exit now...
", getpid());
    exit(EXIT_SUCCESS);

The preceding program (ch10/fork5) has an initialized global variable g and an initialized local variable loc. The parent process, after fork, sleeps for two seconds thus more-or-less guaranteeing that the child process runs first (this kind of synchronization is incorrect in production quality code; we shall address this point in detail later in this chapter). Both the child and parent processes work on the global and local variables; the key question here is this: will the data get corrupted?

Let's just run it and see:

$ ./fork5
Child process, PID 17271:
 loc=9 g=6
Child (17271) done, exiting ...
Parent process, PID 17270:               << after 2 sec >>
 loc=7 g=8
Parent (17270) will exit now...
$ 

Well, the data variables are not corrupted. Again, the key point here is this: as the child has a copy of the parent's variables, all goes well. They change independently of one another; they do not step on each other's toes. So, consider this:

Fork rule #4: Data is copied across the fork, not shared.