Condition variables and mutexes should be global. Condition variables that are allocated in writable memory can synchronize threads among processes if they are shared by the cooperating processes (see mmap(2)) and are initialized for this purpose.

The scope of a condition variable is either intra-process or inter-process. This is dependent upon whether the argument is passed implicitly or explicitly to the initialization of that condition variable. A condition variable does not need to be explicitly initialized. A condition variable is initialized with all zeros, by default, and its scope is set to within the calling process. For inter-process synchronization, a condition variable must be initialized once, and only once, before use.

A condition variable must not be simultaneously initialized by multiple threads or re-initialized while in use by other threads.

Attributes of condition variables can be set to the default or customized at initialization.

The cond_init() function initializes the condition variable pointed to by cvp. A condition variable can have several different types of behavior, specified by type. No current type uses arg although a future type may specify additional behavior parameters with arg. The type argument c take one of the following values:

USYNC_THREAD

The condition variable can synchronize threads only in this process. This is the default.

USYNC_PROCESS

The condition variable can synchronize threads in this process and other processes. Only one process should initialize the condition variable. The object initialized with this attribute must be allocated in memory shared between processes, either in System V shared memory (see shmop(2)) or in memory mapped to a file (see mmap(2)). It is illegal to initialize the object this way and to not allocate it in such shared memory.

Initializing condition variables can also be accomplished by allocating in zeroed memory, in which case, a type of USYNC_THREAD is assumed.

If default condition variable attributes are used, statically allocated condition variables can be initialized by the macro DEFAULTCV.

Default condition variable initialization (intra-process):

cond_t cvp;

cond_init(&cvp, NULL, NULL); /*initialize condition variable with default*/

or

cond_init(&cvp, USYNC_THREAD, NULL);

or

cond_t  cond  = DEFAULTCV;

Customized condition variable initialization (inter-process):

cond_init(&cvp, USYNC_PROCESS, NULL); /* initialize cv with inter-process scope */

The condition wait interface allows a thread to wait for a condition and atomically release the associated mutex that it needs to hold to check the condition. The thread waits for another thread to make the condition true and that thread's resulting call to signal and wakeup the waiting thread.

The cond_wait() function atomically releases the mutex pointed to by mp and causes the calling thread to block on the condition variable pointed to by cvp. The blocked thread may be awakened by cond_signal(), cond_broadcast(), or when interrupted by delivery of a UNIX signal or a fork().

The cond_wait(), cond_timedwait(), and cond_reltimedwait() functions always return with the mutex locked and owned by the calling thread even when returning an error, except when the mutex is of USYNC_PROCESS_ROBUST type and has been left irrecoverable by the mutex's last owner. The cond_wait(), cond_timedwait(), and cond_reltimedwait() functions return the appropriate error value if they fail to internally reacquire the mutex.

A condition signal allows a thread to unblock a single thread waiting on the condition variable, whereas a condition broadcast allows a thread to unblock all threads waiting on the condition variable.

The cond_signal() function unblocks one thread that is blocked on the condition variable pointed to by cvp.

The cond_broadcast() function unblocks all threads that are blocked on the condition variable pointed to by cvp.

If no threads are blocked on the condition variable, then cond_signal() and cond_broadcast() have no effect.

Both functions should be called under the protection of the same mutex that is used with the condition variable being signaled. Otherwise, the condition variable may be signaled between the test of the associated condition and blocking in cond_wait(). This can cause an infinite wait.

The condition destroy functions destroy any state, but not the space, associated with the condition variable.

The cond_destroy() function destroys any state associated with the condition variable pointed to by cvp. The space for storing the condition variable is not freed.

The cond_wait() functin is normally used in a loop testing some condition, as follows:

(void) mutex_lock(mp);
while (cond == FALSE) {
     (void) cond_wait(cvp, mp);
}
(void) mutex_unlock(mp);

The cond_timedwait() function is normally used in a loop testing some condition. It uses an absolute timeout value as follows:

timestruc_t to;
...
(void) mutex_lock(mp);
to.tv_sec = time(NULL) + TIMEOUT;
to.tv_nsec = 0;
while (cond == FALSE) {
      err = cond_timedwait(cvp, mp, &to);
      if (err == ETIME) {
            /* timeout, do something */
            break;
      }
}
(void) mutex_unlock(mp);

The cond_reltimedwait() function is normally used in a loop testing in some condition. It uses a relative timeout value as follows:

timestruc_t to;
...
(void) mutex_lock(mp);
while (cond == FALSE) {
     to.tv_sec = TIMEOUT;
     to.tv_nsec = 0;
     err = cond_reltimedwait(cvp, mp, &to);
     if (err == ETIME) {
          /* timeout, do something */
          break;
     }
}
(void) mutex_unlock(mp);