#include <sys/types.h> #include <sys/ipc.h> #include <sys/sem.h> int semop(int semid, struct sembuf *sops, size_t nsops); int semtimedop(int semid, struct sembuf *sops, size_t nsops, const struct timespec *timeout);
Feature Test Macro Requirements for glibc (see feature_test_macros(7)):
unsigned short semval; /* semaphore value */ unsigned short semzcnt; /* # waiting for zero */ unsigned short semncnt; /* # waiting for increase */ pid_t sempid; /* PID of process that last
semop() performs operations on selected semaphores in the set indicated by semid. Each of the nsops elements in the array pointed to by sops is a structure that specifies an operation to be performed on a single semaphore. The elements of this structure are of type struct sembuf, containing the following members:
unsigned short sem_num; /* semaphore number */ short sem_op; /* semaphore operation */ short sem_flg; /* operation flags */
Flags recognized in sem_flg are IPC_NOWAIT and SEM_UNDO. If an operation specifies SEM_UNDO, it will be automatically undone when the process terminates.
The set of operations contained in sops is performed in array order, and atomically, that is, the operations are performed either as a complete unit, or not at all. The behavior of the system call if not all operations can be performed immediately depends on the presence of the IPC_NOWAIT flag in the individual sem_flg fields, as noted below.
Each operation is performed on the sem_num-th semaphore of the semaphore set, where the first semaphore of the set is numbered 0. There are three types of operation, distinguished by the value of sem_op.
If sem_op is a positive integer, the operation adds this value to the semaphore value (semval). Furthermore, if SEM_UNDO is specified for this operation, the system subtracts the value sem_op from the semaphore adjustment (semadj) value for this semaphore. This operation can always proceed---it never forces a thread to wait. The calling process must have alter permission on the semaphore set.
If sem_op is zero, the process must have read permission on the semaphore set. This is a "wait-for-zero" operation: if semval is zero, the operation can immediately proceed. Otherwise, if IPC_NOWAIT is specified in sem_flg, semop() fails with errno set to EAGAIN (and none of the operations in sops is performed). Otherwise, semzcnt (the count of threads waiting until this semaphore's value becomes zero) is incremented by one and the thread sleeps until one of the following occurs:
If sem_op is less than zero, the process must have alter permission on the semaphore set. If semval is greater than or equal to the absolute value of sem_op, the operation can proceed immediately: the absolute value of sem_op is subtracted from semval, and, if SEM_UNDO is specified for this operation, the system adds the absolute value of sem_op to the semaphore adjustment (semadj) value for this semaphore. If the absolute value of sem_op is greater than semval, and IPC_NOWAIT is specified in sem_flg, semop() fails, with errno set to EAGAIN (and none of the operations in sops is performed). Otherwise, semncnt (the counter of threads waiting for this semaphore's value to increase) is incremented by one and the thread sleeps until one of the following occurs:
On successful completion, the sempid value for each semaphore specified in the array pointed to by sops is set to the caller's process ID. In addition, the sem_otime is set to the current time.
Note that if semtimedop() is interrupted by a signal, causing the call to fail with the error EINTR, the contents of timeout are left unchanged.
The sem_undo structures of a process aren't inherited by the child produced by fork(2), but they are inherited across an execve(2) system call.
semop() is never automatically restarted after being interrupted by a signal handler, regardless of the setting of the SA_RESTART flag when establishing a signal handler.
A semaphore adjustment (semadj) value is a per-process, per-semaphore integer that is the negated sum of all operations performed on a semaphore specifying the SEM_UNDO flag. Each process has a list of semadj values---one value for each semaphore on which it has operated using SEM_UNDO. When a process terminates, each of its per-semaphore semadj values is added to the corresponding semaphore, thus undoing the effect of that process's operations on the semaphore (but see BUGS below). When a semaphore's value is directly set using the SETVAL or SETALL request to semctl(2), the corresponding semadj values in all processes are cleared. The clone(2) CLONE_SYSVSEM flag allows more than one process to share a semadj list; see clone(2) for details.
The semval, sempid, semzcnt, and semnct values for a semaphore can all be retrieved using appropriate semctl(2) calls.
The implementation has no intrinsic limits for the adjust on exit maximum value (SEMAEM), the system wide maximum number of undo structures (SEMMNU) and the per-process maximum number of undo entries system parameters.
In kernels 2.6.x, x <= 10, there is a bug that in some circumstances prevents a thread that is waiting for a semaphore value to become zero from being woken up when the value does actually become zero. This bug is fixed in kernel 2.6.11.
struct sembuf sops[2]; int semid;
/* Code to set semid omitted */
sops[0].sem_num = 0; /* Operate on semaphore 0 */ sops[0].sem_op = 0; /* Wait for value to equal 0 */ sops[0].sem_flg = 0;
sops[1].sem_num = 0; /* Operate on semaphore 0 */ sops[1].sem_op = 1; /* Increment value by one */ sops[1].sem_flg = 0;
if (semop(semid, sops, 2) == -1) {
perror("semop");
exit(EXIT_FAILURE);
}
A further example of the use of semop() can be found in shmop(2).