Producer Consumer with Circular Buffer and Suspected Race Condition

This is a discussion on Producer Consumer with Circular Buffer and Suspected Race Condition within the C Programming forums, part of the General Programming Boards category; Hey all, I am relatively new to C programming and need to write a program that uses semaphores to control ...

  1. #1
    Registered User
    Join Date
    Nov 2010
    Posts
    2

    Producer Consumer with Circular Buffer and Suspected Race Condition

    Hey all,

    I am relatively new to C programming and need to write a program that uses semaphores to control access to shared memory that places items into a circular buffer. The problem I have is one where the consumer runs out of things to consume and the producer doesn't wake up to place anything new into the buffer. I have the following code, which gives the output below it. I have been looking at this now for days and just seem to be going around in circles. Ay help would be appreciated.

    Thanks,

    Terry

    Code:
    #define _SVID_SOURCE
    #include <stdio.h>
    #include <stdlib.h>
    #include <string.h>
    #include <semaphore.h>
    #include <fcntl.h>
    #include <sys/uio.h>
    #include <sys/shm.h>
    #include <sys/stat.h>
    #include <sys/types.h>
    #include <unistd.h>
    #include <pthread.h>
    
    #define BUFFER_SIZE 10
    #define ERROR_CHAR (-1)
    #define MAX_FILE_SIZE 256
    
    #define TRUE 1
    #define FALSE 0
    
    void buffer_char(char *c);
    char* unbuffer_char(void);
    
    typedef struct
    {
        char file_path[BUFFER_SIZE][MAX_FILE_SIZE];
        int head;
        int tail;
        sem_t full;
        sem_t empty;
        sem_t mutex;
    } Buffer;
    
    Buffer *buffer_ptr;
    
    int main(int argc, char** argv)
    {
        int shmid;
        pid_t prod_pid, con_pid;
    
        /* shmid is the id of the shared memory address for our buffer */
        shmid = shmget(IPC_PRIVATE, getpagesize(), IPC_CREAT | 0666);
    
        /* get a pointer to our buffer in shared memory */
        buffer_ptr = (Buffer*) shmat(shmid, NULL, 0);
    
        /* initialise the buffer */
        buffer_ptr->head = 0;
        buffer_ptr->tail = 0;
    
        /* initialise our semaphores (2nd param 1 means shared betweeen processes */
        sem_init(&buffer_ptr->empty, 1, BUFFER_SIZE);
        sem_init(&buffer_ptr->full, 1, 0);
        sem_init(&buffer_ptr->mutex, 1, 1);
    
        /* spawn our child processes */
        prod_pid = fork();
        if (prod_pid == 0)
        {
            /* this is the producer process */
            int i;
            char items[15][10] = {"File 1", "File 2", "File 3", "File 4", "File 5", "File 6", "File 7", "File 8", "File 9", "File 10", "File 11", "File 12", "File 13", "File 14", "File 15"};
    
            for (i = 0; i < 15; i++)
            {
                sem_wait(&buffer_ptr->empty);
                sem_wait(&buffer_ptr->mutex);
    
                printf("Buffering %s\n", items[i]);
                buffer_char(items[i]);
    
                sem_post(&buffer_ptr->mutex);
                sem_post(&buffer_ptr->full);
            }
            exit(0);
        }
    
        con_pid = fork();
        if (con_pid == 0)
        {
            /* this is the consumer process */
            while (TRUE)
            {
                sem_wait(&buffer_ptr->full);
                sem_wait(&buffer_ptr->mutex);
    
                printf("Unbeffering %s\n", unbuffer_char());
    
                sem_post(&buffer_ptr->mutex);
                sem_post(&buffer_ptr->empty);
            }
            exit(0);
        }
    
    
        if (prod_pid != 0 && con_pid != 0)
        {
            /* this is the main process so wait here for both processes to finish before closing out and finishing */
            wait(prod_pid);
            wait(con_pid);
            printf("************************************");
    
            /* detach the shared memory and deallocate the memory segment */
            shmdt(&buffer_ptr);
            shmctl(shmid, IPC_RMID, 0);
    
            /* finally, close the signature file */
    
            return (EXIT_SUCCESS);
        }
    }
    
    void buffer_char(char* c)
    {
        /* Use modulo as a trick to wrap around the end of the buffer back to the beginning */
        if ((buffer_ptr->tail + 1) % BUFFER_SIZE != buffer_ptr->head)
        {
            strcpy(buffer_ptr->file_path[buffer_ptr->tail], c);
            buffer_ptr->tail = (buffer_ptr->tail + 1) % BUFFER_SIZE;
        }
    }
    
    char* unbuffer_char(void)
    {
        if (buffer_ptr->tail != buffer_ptr->head)
        {
            char *temp = buffer_ptr->file_path[buffer_ptr->head];
            buffer_ptr->head = (buffer_ptr->head + 1) % BUFFER_SIZE;
    
            printf("Head: %d\n", buffer_ptr->head);
            printf("Tail: %d\n", buffer_ptr->tail);
    
            return (temp);
        }
        else
        {
            printf("Problem is here!\n");
            printf("Head: %d\n", buffer_ptr->head);
            printf("Tail: %d\n", buffer_ptr->tail);
        }
    }
    Output:

    Code:
    Buffering File 1
    Buffering File 2
    Buffering File 3
    Buffering File 4
    Buffering File 5
    Buffering File 6
    Buffering File 7
    Buffering File 8
    Buffering File 9
    Buffering File 10
    Head: 1
    Tail: 9
    Unbeffering File 1
    Head: 2
    Tail: 9
    Unbeffering File 2
    Head: 3
    Tail: 9
    Unbeffering File 3
    Head: 4
    Tail: 9
    Unbeffering File 4
    Head: 5
    Tail: 9
    Unbeffering File 5
    Head: 6
    Tail: 9
    Unbeffering File 6
    Head: 7
    Tail: 9
    Unbeffering File 7
    Head: 8
    Tail: 9
    Unbeffering File 8
    Head: 9
    Tail: 9
    Unbeffering File 9
    Problem is here!
    Head: 9
    Tail: 9

  2. #2
    and the hat of wrongness Salem's Avatar
    Join Date
    Aug 2001
    Location
    The edge of the known universe
    Posts
    32,671
    Does it have anything to do with falling off the end of a function which expects to return a result?

    Code:
    $ gcc -Wall -c foo.c
    foo.c: In function ‘main’:
    foo.c:42: warning: implicit declaration of function ‘getpagesize’
    foo.c:99: warning: implicit declaration of function ‘wait’
    foo.c: In function ‘unbuffer_char’:
    foo.c:141: warning: control reaches end of non-void function
    foo.c: In function ‘main’:
    foo.c:111: warning: control reaches end of non-void function
    You'll get a garbage answer, which you then try and "print", and no doubt, it blows up.
    If you dance barefoot on the broken glass of undefined behaviour, you've got to expect the occasional cut.
    If at first you don't succeed, try writing your phone number on the exam paper.
    I support http://www.ukip.org/ as the first necessary step to a free Europe.

  3. #3
    Registered User Codeplug's Avatar
    Join Date
    Mar 2003
    Posts
    4,676
    You should try to understand the code that you're copying:
    http://en.wikipedia.org/wiki/Circula..._One_Slot_Open

    You have a 9 slot queue, and a 10 count semaphore.

    getpagesize() is no longer part of Posix. "sizeof(Buffer)" is sufficient.

    "wait()" doesn't do what you think it does. You should include <sys/wait.h> and call waitpid() in a loop.
    http://www.opengroup.org/onlinepubs/...s/waitpid.html

    unbuffer_char() should return 0 if the queue is empty.
    buffer_char() should log and return an error also if the queue is full.

    Normally, the semaphore's would prevent dequeue-on-empty and enque-on-full. However, error checking and logging provides the first indication that things aren't "normal".

    You mostly have a Posix-compliant source - so instead of "_SVID_SOURCE", I would use "-D_XOPEN_SOURCE=700" when compiling instead.

    Here's how I would compile that code using gcc:
    gcc foo.c -Wall -pedantic -std=c99 -D_XOPEN_SOURCE=700 -pthread

    Even though you're not using pthreads, "-pthread" will link the libs needed for semaphores as well as any necessary platform-specific defines and libraries used in mult-threading. Alternatively, "-lrt" will provide the semaphore routines.

    gg

  4. #4
    Registered User
    Join Date
    Nov 2010
    Posts
    2
    Thanks for the feedback guys.

    T.

Popular pages Recent additions subscribe to a feed

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21