POSIX pthread programming

Question

I have to code a multithreaded(say 2 threads) program where each of these threads do a different task. Also, these threads must keep running infinitely in the background once started. Here is what I have done. Can somebody please give me some feedback if the method is good and if you see some problems. Also, I would like to know how to shut the threads in a systematic way once I terminate the execution say with Ctrl+C.

The main function creates two threads and let them run infinitely as below.

Here is the skeleton:

void    *func1();
void    *func2();

int main(int argc, char *argv[])
{   

    pthread_t th1,th2;
    pthread_create(&th1, NULL, func1, NULL);
    pthread_create(&th2, NULL, func2, NULL);

    fflush (stdout);
    for(;;){
    }
    exit(0); //never reached
}

void *func1()
{
    while(1){
    //do something
    }
}

void *func2()
{
    while(1){
    //do something
    }
} 

Thanks.

Edited code using inputs from the answers: Am I exiting the threads properly?

#include <stdlib.h>     /*  exit() */
#include <stdio.h>      /* standard in and output*/
#include <pthread.h>
#include <unistd.h>
#include <time.h>
#include <sys/time.h>
#include <sys/types.h>
#include <signal.h>
#include <semaphore.h>

sem_t end;

void    *func1();
void    *func2();

void ThreadTermHandler(int signo){
    if (signo == SIGINT) {
        printf("Ctrl+C detected !!! \n");
        sem_post(&end);
        }
}
void *func1()
{
    int value;
    for(;;){
        sem_getvalue(&end, &value);
        while(!value){
            printf("in thread 1 \n");
        }
    }
    return 0;
}

void *func2()
{
    int value;
    for(;;){
        sem_getvalue(&end, &value);
        while(!value){
            printf("value = %d\n", value);
        }
    }
    return 0;
}

int main(int argc, char *argv[])
{


    sem_init(&end, 0, 0);
    pthread_t th1,th2;
    int value  = -2;
    pthread_create(&th1, NULL, func1, NULL);
    pthread_create(&th2, NULL, func2, NULL);

    struct sigaction sa;
    sigemptyset(&sa.sa_mask);
    sa.sa_flags = SA_SIGINFO;
    sa.sa_sigaction = ThreadTermHandler;
    // Establish a handler to catch CTRL+c and use it for exiting.
    if (sigaction(SIGINT, &sa, NULL) == -1) {
        perror("sigaction for Thread Termination failed");
        exit( EXIT_FAILURE );
    }

    /* Wait for SIGINT. */
    while (sem_wait(&end)!=0){}
    //{
        printf("Terminating Threads.. \n");
        sem_post(&end);
                sem_getvalue(&end, &value);
        /* SIGINT received, cancel threads. */
        pthread_cancel(th1);
        pthread_cancel(th2);
        /* Join threads. */
        pthread_join(th1, NULL);
        pthread_join(th2, NULL);
    //}
    exit(0);
}

Answer 1

There are mainly two approaches for thread termination.

• Use a cancellation point. The thread will terminate when requested to cancel and it reaches a cancellation point, thus ending execution in a controlled fashion;
• Use a signal. Have the threads install a signal handler which provides a mechanism for termination (setting a flag and reacting to EINTR).

Both approaches has caveats. Refer to Kill Thread in Pthread Library for more details.

In your case, it seems a good opportunity to use cancellation points. I will work with a commented example. The error-checking has been omitted for clarity.

#define _POSIX_C_SOURCE 200809L
#include <pthread.h>
#include <signal.h>
#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>

void sigint(int signo) {
    (void)signo;
}

void *thread(void *argument) {
    (void)argument;
    for (;;) {
        // Do something useful.
        printf("Thread %u running.\n", *(unsigned int*)argument);

        // sleep() is a cancellation point in this example.
        sleep(1);
    }
    return NULL;
}

int main(void) {
    // Block the SIGINT signal. The threads will inherit the signal mask.
    // This will avoid them catching SIGINT instead of this thread.
    sigset_t sigset, oldset;
    sigemptyset(&sigset);
    sigaddset(&sigset, SIGINT);
    pthread_sigmask(SIG_BLOCK, &sigset, &oldset);

    // Spawn the two threads.
    pthread_t thread1, thread2;
    pthread_create(&thread1, NULL, thread, &(unsigned int){1});
    pthread_create(&thread2, NULL, thread, &(unsigned int){2});

    // Install the signal handler for SIGINT.
    struct sigaction s;
    s.sa_handler = sigint;
    sigemptyset(&s.sa_mask);
    s.sa_flags = 0;
    sigaction(SIGINT, &s, NULL);

    // Restore the old signal mask only for this thread.
    pthread_sigmask(SIG_SETMASK, &oldset, NULL);

    // Wait for SIGINT to arrive.
    pause();

    // Cancel both threads.
    pthread_cancel(thread1);
    pthread_cancel(thread2);

    // Join both threads.
    pthread_join(thread1, NULL);
    pthread_join(thread2, NULL);

    // Done.
    puts("Terminated.");
    return EXIT_SUCCESS;
}

The need for blocking/unblocking signals is that if you send SIGINT to the process, any thread may be able to catch it. You do so before spawning the threads to avoid having them doing it by themselves and needing to synchronize with the parent. After the threads are created, you restore the mask and install a handler.

Cancellation points can be tricky if the threads allocates a lot of resources; in that case, you will have to use pthread_cleanup_push() and pthread_cleanup_pop(), which are a mess. But the approach is feasible and rather elegant if used properly.

Answer 2

The answer depends a lot on what you want to do when the user presses CtrlC.

If your worker threads are not modifying data that needs to be saved on exit, you don't need to do anything. The default action of SIGINT is to terminate the process, and that includes all threads that make up the process.

If your threads do need to perform cleanup, however, you've got some work to do. There are two separate issues you need to consider:

• How you handle the signal and get the message to threads that they need to terminate.
• How your threads receive and handle the request to terminate.

First of all, signal handlers are a pain. Unless you're very careful, you have to assume most library functions are not legal to call from a signal handler. Fortunately, sem_post is specified to be async-signal-safe, and can meet your requirements perfectly:

1. At the beginning of your program, initialize a semaphore with sem_init(&exit_sem, 0, 0);
2. Install a signal handler for SIGINT (and any other termination signals you want to handle, like SIGTERM) that performs sem_post(&exit_sem); and returns.
3. Replace the for(;;); in the main thread with while (sem_wait(&exit_sem)!=0).
4. After sem_wait succeeds, the main thread should inform all other threads that they should exit, then wait for them all to exit.

The above can also be accomplished without semaphores using signal masks and sigwaitinfo, but I prefer the semaphore approach because it doesn't require you to learn lots of complicated signal semantics.

Now, there are several ways you could handle informing the worker threads that it's time to quit. Some options I see:

1. Having them check sem_getvalue(&exit_sem) periodically and cleanup and exit if it returns a nonzero value. Note however that this will not work if the thread is blocked indefinitely, for example in a call to read or write.
2. Use pthread_cancel, and carefully place cancellation handlers (pthread_cleanup_push) all over the place.
3. Use pthread_cancel, but also use pthread_setcancelstate to disable cancellation during most of your code, and only re-enable it when you're going to perform blocking IO operations. This way you need only put the cleanup handlers just in the places where cancellation is enabled.
4. Learn advanced signal semantics, and setup an additional signal and interrupting signal handler which you send to all threads via pthread_kill which will cause blocking syscalls to return with an EINTR error. Then your threads can act on this and exit the normal C way via a string of failure returns all the way back up the the start function.

I would not recommend approach 4 for beginners, because it's hard to get right, but for advanced C programmers it may be the best because it allows you to use the existing C idiom of reporting exceptional conditions via return values rather than "exceptions".

Also note that with pthread_cancel, you will need to periodically call pthread_testcancel if you are not calling any other functions which are cancellation points. Otherwise the cancellation request will never be acted upon.

Answer 3

This is a bad idea:

for(;;){
}

because your main thread will execute unnecessary CPU instructions.

If you need to wait in the main thread, use pthread_join as answered in this question: Multiple threads in C program