CODEWITHSUNDEEP
csegmentation-faultfork
Mehdi
Mehdi

Reputation: 2228

Segmentation fault when i run two process with fork

When I run this program :

#include <sys/types.h>
#include <pthread.h>
#include <sys/mman.h>
#include <stdio.h>
#include <fcntl.h>
#include <sys/stat.h>
#include <unistd.h>
#include <sys/wait.h>
#include <stdlib.h>

int main () {

    int stat;

    pthread_mutex_t *mutex = (pthread_mutex_t*)mmap (0, sizeof (pthread_mutex_t) + sizeof (long),PROT_READ | PROT_WRITE,
                                                     MAP_SHARED ,-1, 0);

    long *data = (long*)(&mutex[1]); /* map 'data' after mutex */
    int pid;

    pthread_mutexattr_t attr;
    pthread_mutexattr_init (&attr);
    pthread_mutexattr_setpshared (&attr, PTHREAD_PROCESS_SHARED);
    pthread_mutex_init (mutex, &attr);

    *data = 0;

    pid = fork ();

    pthread_mutex_lock (mutex);
    (*data)++;
    pthread_mutex_unlock (mutex);

    if (!pid) /* child exits */
        exit (0);
    else
        waitpid (pid, &stat, 0);
    printf ("data is %ld\n", *data);

    return 0;
}

I have segmentation fault ! What is problem please ? and what can i do to resolve this error ?

after debbuging with valgrind, i have this message : 

==28660== Process terminating with default action of signal 11 (SIGSEGV)
==28660==  Access not within mapped region at address 0xF
==28660==    at 0x4E35C10: pthread_mutex_init (pthread_mutex_init.c:83)
==28660==    by 0x4008F0: main (in /home/program)
==28660==  If you believe this happened as a result of a stack
==28660==  overflow in your program's main thread (unlikely but
==28660==  possible), you can try to increase the size of the
==28660==  main thread stack using the --main-stacksize= flag.
==28660==  The main thread stack size used in this run was 8388608.

Upvotes: 0

Views: 1273

Answers (3)

Gautham Kantharaju
Gautham Kantharaju

Reputation: 1763

I tried compiling and running your program. Even I was facing the same problem, segmentation fault.

Solution: If you are passing fd with value "-1", make it a habit of using "MAP_ANON" flag i.e. for example (MAP_SHARED | MAP_ANON).

Hope this would solve your problem :)

Upvotes: 1

Alvin Wong
Alvin Wong

Reputation: 12420

If you are trying to call mmap without a file descriptor (not backed by a file or device), you should set the MAP_ANONYMOUS flag. Without MAP_ANONYMOUS, you will receive a EBADF error.

fd is not a valid file descriptor (and MAP_ANONYMOUS was not set).

Try to use MAP_SHARED | MAP_ANONYMOUS as the flag if it is not backed by a file or device.

i.e.:

pthread_mutex_t *mutex = (pthread_mutex_t*)mmap (0, sizeof (pthread_mutex_t) + sizeof (long),PROT_READ | PROT_WRITE,
                                                 MAP_SHARED | MAP_ANONYMOUS ,-1, 0);

Upvotes: 3

kimhyunkang
kimhyunkang

Reputation: 155

You are calling mmap(void *start, size_t length, int prot , int flags, int fd, off_t offset); with fd -1. As far as I know, calling mmap with -1 is not valid.

Check if (mutex == MAP_FAILED) to find out if mmap succeeds, and if it fails, find out which error mmap produced by checking errno

Upvotes: 2

Related Questions