Linux C++中的阻塞队列实现方式

原文：http://stackoverflow.com/questions/5716437/condition-variable

 

Well, conditional variables allow you to wait for certain condition to occur. In practice your thread may sleep on conditional variable and other thread wakes it up.

Conditional variable also usually comes with mutex. This allows you to solve following synchronisation problem: how can you check state of some mutex protected data structure and then wait until it's state changes to something else. I.e.

/* in thread 1 */
pthread_mutex_lock(mx); /* protecting state access */
if (state == GOOD) {
    pthread_mutex_unlock(mx);
    wait_for_event();
    pthread_mutex_lock(mx);
}
pthread_mutex_unlock(mx);


/* in thread 2 */
pthread_mutex_lock(mx); /* protecting state access */
state = GOOD;
pthread_mutex_unlock(mx);
signal_event(); /* expecting to wake thread 1 up */

 

 

This pseudocode sample carries a bug. What happens if scheduler decides to switch context from thread 1 to thread 2 after pthread_mutex_unlock(mx), but before wait_for_event(). In this case, thread 2 will not wake thread 1 and thread 1 will continue sleeping, possibly forever.

Conditional variable solves this by atomically unlocking the mutex before sleeping and atomically locking it after waking up. Code that works looks like this:

/* in thread 1 */
pthread_mutex_lock(mx); /* protecting state access */
if (state == GOOD) {
    pthread_cond_wait(cond, mx); /* unlocks the mutex and sleeps, then locks it back */
}
pthread_mutex_unlock(mx);

/* in thread 2 */
pthread_mutex_lock(mx); /* protecting state access */
state = GOOD;
pthread_cond_signal(cond); /* expecting to wake thread 1 up */
pthread_mutex_unlock(mx);