A semaphore is a non-negative integer count and is generally used to coordinate access to resources. The initial semaphore count is set to the number of free resources, then threads slowly increment and decrement the count as resources are added and removed. If the semaphore count drops to 0, which means no available resources, threads attempting to decrement the semaphore will block until the count is greater than 0.

Semaphores can synchronize threads in this process and other processes if they are allocated in writable memory and shared among the cooperating processes (see mmap(2)), and have been initialized for this purpose.

Semaphores must be initialized before use; semaphores pointed to by sp to count are initialized by sema_init(). The type argument can assign several different types of behavior to a semaphore. No current type uses arg, although it may be used in the future.

The type argument may be one of the following:

USYNC_PROCESS

The semaphore can synchronize threads in this process and other processes. Initializing the semaphore should be done by only one process. A semaphore initialized with this type must be allocated in memory shared between processes, either in Sys 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 not allocate it in such shared memory. arg is ignored.

USYNC_THREAD

The semaphore can synchronize threads only in this process. The arg argument is ignored. USYNC_THREAD does not support multiple mappings to the same logical synch object. If you need to mmap() a synch object to different locations within the same address space, then the synch object should be initialized as a shared object USYNC_PROCESS for Solaris threads and PTHREAD_PROCESS_PRIVATE for POSIX threads.

A semaphore must not be simultaneously initialized by multiple threads, nor re-initialized while in use by other threads.

Default semaphore initialization (intra-process):

sema_t sp; int count = 1; sema_init(&sp, count, NULL, NULL);

or

sema_init(&sp, count, USYNC_THREAD, NULL);

Customized semaphore initialization (inter-process):

sema_t sp; int count = 1; sema_init(&sp, count, USYNC_PROCESS, NULL);

The sema_destroy() function destroys any state related to the semaphore pointed to by sp. The semaphore storage space is not released.

The sema_wait() function blocks the calling thread until the semaphore count pointed to by sp is greater than 0, and then it atomically decrements the count.

The sema_trywait() function atomically decrements the semaphore count pointed to by sp, if the count is greater than 0; otherwise, it returns an error.

The sema_post() function atomically increments the semaphore count pointed to by sp. If there are any threads blocked on the semaphore, one will be unblocked.

The semaphore functionality described on this man page is for the Solaris threads implementation. For the POSIX-compliant semaphore interface documentation, see sem_open(3RT), sem_init(3RT), sem_wait(3RT), sem_post(3RT), sem_getvalue(3RT), sem_unlink(3RT), sem_close(3RT), sem_destroy(3RT)).

Upon successful completion, 0 is returned; otherwise, a non-zero value indicates an error.

These functions will fail if:

EINVAL

The sp argument does not refer to a valid semaphore.

EFAULT

Either the sp or arg argument points to an illegal address.

The sema_wait() function will fail if:

EINTR

The wait was interrupted by a signal or fork().

The sema_trywait() function will fail if:

EBUSY

The semaphore pointed to by sp has a 0 count.

The sema_post() function will fail if:

EOVERFLOW

The semaphore value pointed to by sp exceeds SEM_VALUE_MAX.

See attributes(5) for descriptions of the following attributes:

mmap(2), shmop(2), sem_close(3RT), sem_destroy(3RT), sem_getvalue(3RT), sem_init(3RT), sem_open(3RT), sem_post(3RT), sem_unlink(3RT), sem_wait(3RT), attributes(5), standards(5)

These functions are also available by way of:

#include <thread.h>

By default, there is no defined order of unblocking for multiple threads waiting for a semaphore.