Exercise 34-1: Signal to process group race condition

A parent creates children (all in the same process group), makes itself immune to SIGUSR1, then sends SIGUSR1 to the group. What problem can arise, especially with shell pipelines?

Exercise 34-2: setpgid() before and after exec()

Write a program to verify that a parent can change the PGID of a child before exec(), but not after.

/* Exercise 34-2 outline */
#include <unistd.h>
#include <stdio.h>
#include <errno.h>
#include <string.h>

int main(void)
{
    pid_t child = fork();
    if (child == 0) {
        sleep(2);           /* Give parent time to test both cases */
        execl("/bin/sleep", "sleep", "30", NULL);
    }

    /* Test 1: setpgid BEFORE exec (should succeed) */
    if (setpgid(child, child) == 0)
        printf("setpgid before exec: SUCCESS\n");
    else
        printf("setpgid before exec: FAILED (%s)\n", strerror(errno));

    /* Wait for child to exec */
    sleep(3);

    /* Test 2: setpgid AFTER exec (should fail with EACCES) */
    if (setpgid(child, child) == 0)
        printf("setpgid after exec: SUCCESS (unexpected)\n");
    else
        printf("setpgid after exec: FAILED with %s (expected EACCES)\n",
               strerror(errno));
    return 0;
}

Exercise 34-3: setsid() from a process group leader fails

Verify that calling setsid() from a process group leader returns -1 with EPERM, but succeeds after fork().

/* Exercise 34-3 */
#define _XOPEN_SOURCE 500
#include <unistd.h>
#include <stdio.h>
#include <errno.h>
#include <string.h>

int main(void)
{
    printf("PID=%ld, PGID=%ld\n", (long)getpid(), (long)getpgrp());

    /* A process is a group leader if PID == PGID */
    if (getpid() == getpgrp()) {
        printf("We are a process group leader - setsid() should fail\n");
        if (setsid() == -1)
            printf("setsid() FAILED as expected: %s\n", strerror(errno));
        else
            printf("setsid() succeeded (unexpected!)\n");
    }

    /* Fork: child is NOT a group leader (child PID != parent PGID) */
    pid_t child = fork();
    if (child == 0) {
        printf("Child PID=%ld, PGID=%ld - calling setsid()\n",
               (long)getpid(), (long)getpgrp());
        pid_t sid = setsid();
        if (sid == -1)
            printf("setsid() failed: %s\n", strerror(errno));
        else
            printf("setsid() succeeded: new SID=%ld\n", (long)sid);
        _exit(0);
    }
    wait(NULL);
    return 0;
}
/* Output:
 * PID=1234, PGID=1234
 * We are a process group leader - setsid() should fail
 * setsid() FAILED as expected: Operation not permitted
 * Child PID=1235, PGID=1234 - calling setsid()
 * setsid() succeeded: new SID=1235
 */

Exercise 34-4: SIGHUP not sent to foreground group when controlling process survives

If SIGHUP does NOT terminate the controlling process (it handles the signal and continues), then the kernel does NOT send SIGHUP to the foreground process group. Verify this.

Exercise 34-5: Race condition if SIGTSTP unblock is moved to start of handler

In Listing 34-6, what race condition is created if sigprocmask(SIG_UNBLOCK) is placed at the very beginning of the handler?

Exercise 34-6: read() fails with EIO for orphaned group terminal access

Write a program to verify that when a member of an orphaned process group attempts read() from the controlling terminal, read() fails with EIO.

Exercise 34-7: SIGTTIN/SIGTTOU/SIGTSTP discarded vs delivered for orphaned group

Verify that these signals are silently discarded (no effect) when sent to an orphaned process group member with default disposition, but delivered normally when a handler is installed.