Counting Semaphore Interfaces

Semaphores. Semaphores are the basis for synchronization and mutual exclusion in NuttX. NuttX supports POSIX semaphores.

Semaphores are the preferred mechanism for gaining exclusive access to a resource. sched_lock() and sched_unlock() can also be used for this purpose. However, sched_lock() and sched_unlock() have other undesirable side-effects in the operation of the system: sched_lock() also prevents higher-priority tasks from running that do not depend upon the semaphore-managed resource and, as a result, can adversely affect system response times.

Priority Inversion. Proper use of semaphores avoids the issues of sched_lock(). However, consider the following example:

  1. Some low-priority task, Task C, acquires a semaphore in order to get exclusive access to a protected resource.

  2. Task C is suspended to allow some high-priority task,

  3. Task A attempts to acquire the semaphore held by Task C and gets blocked until Task C relinquishes the semaphore.

  4. Task C is allowed to execute again, but gets suspended by some medium-priority Task B.

At this point, the high-priority Task A cannot execute until Task B (and possibly other medium-priority tasks) completes and until Task C relinquishes the semaphore. In effect, the high-priority task, Task A behaves as though it were lower in priority than the low-priority task, Task C! This phenomenon is called priority inversion.

Some operating systems avoid priority inversion by automatically increasing the priority of the low-priority Task C (the operable buzz-word for this behavior is priority inheritance). NuttX supports this behavior, but only if CONFIG_PRIORITY_INHERITANCE is defined in your OS configuration file. If CONFIG_PRIORITY_INHERITANCE is not defined, then it is left to the designer to provide implementations that will not suffer from priority inversion. The designer may, as examples:

  • Implement all tasks that need the semaphore-managed resources at the same priority level,

  • Boost the priority of the low-priority task before the semaphore is acquired, or

  • Use sched_lock() in the low-priority task.

Priority Inheritance. As mentioned, NuttX does support priority inheritance provided that CONFIG_PRIORITY_INHERITANCE is defined in your OS configuration file. However, the implementation and configuration of the priority inheritance feature is sufficiently complex that more needs to be said. How can a feature that can be described by a single, simple sentence require such a complex implementation:

  • CONFIG_SEM_PREALLOCHOLDERS. First of all, in NuttX priority inheritance is implement on POSIX counting semaphores. The reason for this is that these semaphores are the most primitive waiting mechanism in NuttX; Most other waiting facilities are based on semaphores. So if priority inheritance is implemented for POSIX counting semaphores, then most NuttX waiting mechanisms will have this capability.

    Complexity arises because counting semaphores can have numerous holders of semaphore counts. Therefore, in order to implement priority inheritance across all holders, then internal data structures must be allocated to manage the various holders associated with a semaphore. The setting CONFIG_SEM_PREALLOCHOLDERS defines the maximum number of different threads (minus one per semaphore instance) that can take counts on a semaphore with priority inheritance support. This setting defines the size of a single pool of pre-allocated structures. It may be set to zero if priority inheritance is disabled OR if you are only using semaphores as mutexes (only one holder) OR if no more than two threads participate using a counting semaphore.

    The cost associated with setting CONFIG_SEM_PREALLOCHOLDERS is slightly increased code size and around 6-12 bytes times the value of CONFIG_SEM_PREALLOCHOLDERS.

  • CONFIG_SEM_NNESTPRIO: In addition, there may be multiple threads of various priorities that need to wait for a count from the semaphore. These, the lower priority thread holding the semaphore may have to be boosted numerous times and, to make things more complex, will have to keep track of all of the boost priorities values in order to correctly restore the priorities after a count has been handed out to the higher priority thread. The CONFIG_SEM_NNESTPRIO defines the size of an array, one array per active thread. This setting is the maximum number of higher priority threads (minus 1) than can be waiting for another thread to release a count on a semaphore. This value may be set to zero if no more than one thread is expected to wait for a semaphore.

    The cost associated with setting CONFIG_SEM_NNESTPRIO is slightly increased code size and (CONFIG_SEM_PREALLOCHOLDERS + 1) times the maximum number of active threads.

  • Increased Susceptibility to Bad Thread Behavior. These various structures tie the semaphore implementation more tightly to the behavior of the implementation. For examples, if a thread executes while holding counts on a semaphore, or if a thread exits without call sem_destroy() then. Or what if the thread with the boosted priority re-prioritizes itself? The NuttX implement of priority inheritance attempts to handle all of these types of corner cases, but it is very likely that some are missed. The worst case result is that memory could by stranded within the priority inheritance logic.

Locking versus Signaling Semaphores. Semaphores (and mutexes) may be used for many different purposes. One typical use is for mutual exclusion and locking of resources: In this usage, the thread that needs exclusive access to a resources takes the semaphore to get access to the resource. The same thread subsequently releases the semaphore count when it no longer needs exclusive access. Priority inheritance is intended just for this usage case.

In a different usage case, a semaphore may to be used to signal an event: One thread A waits on a semaphore for an event to occur. When the event occurs, another thread B will post the semaphore waking the waiting thread A. This is a completely different usage model; notice that in the mutual exclusion case, the same thread takes and posts the semaphore. In the signaling case, one thread takes the semaphore and a different thread posts the semaphore. Priority inheritance should never be used in this signaling case. Subtle, strange behaviors may result.

When priority inheritance is enabled with CONFIG_PRIORITY_INHERITANCE, the default protocol for the semaphore will be to use priority inheritance. For signaling semaphores, priority inheritance must be explicitly disabled by calling `sem_setprotocol <#semsetprotocol>`__ with SEM_PRIO_NONE. For the case of pthread mutexes, `pthread_mutexattr_setprotocol <#pthreadmutexattrsetprotocol>`__ with PTHREAD_PRIO_NONE.

This is discussed in much more detail on this Wiki page.

POSIX semaphore interfaces:

int sem_init(sem_t *sem, int pshared, unsigned int value)

Initializes the UN-NAMED semaphore sem. Following a successful call to sem_init(), the semaphore may be used in subsequent calls to sem_wait(), sem_post(), and sem_trywait(). The semaphore remains usable until it is destroyed.

Only sem itself may be used for performing synchronization. The result of referring to copies of sem in calls to sem_wait(), sem_trywait(), sem_post(), and sem_destroy(), is not defined.

Parameters
  • sem – Semaphore to be initialized

  • pshared – Process sharing (not used)

  • value – Semaphore initialization value

Returns

0 (OK), or -1 (ERROR) if unsuccessful.

POSIX Compatibility: Comparable to the POSIX interface of the same name. Differences from the full POSIX implementation include:

  • pshared is not used.

int sem_destroy(sem_t *sem)

Used to destroy the un-named semaphore indicated by sem. Only a semaphore that was created using sem_init() may be destroyed using sem_destroy(). The effect of calling sem_destroy() with a named semaphore is undefined. The effect of subsequent use of the semaphore sem is undefined until sem is re-initialized by another call to sem_init().

The effect of destroying a semaphore upon which other tasks are currently blocked is undefined.

Parameters
  • sem – Semaphore to be destroyed.

Returns

0 (OK), or -1 (ERROR) if unsuccessful.

POSIX Compatibility: Comparable to the POSIX interface of the same name.

sem_t *sem_open(const char *name, int oflag, ...)

Establishes a connection between named semaphores and a task. Following a call to sem_open() with the semaphore name, the task may reference the semaphore associated with name using the address returned by this call. The semaphore may be used in subsequent calls to sem_wait(), sem_trywait(), and sem_post(). The semaphore remains usable until the semaphore is closed by a successful call to sem_close().

If a task makes multiple calls to sem_open() with the same name, then the same semaphore address is returned (provided there have been no calls to sem_unlink()).

Input Parameters:

Parameters
  • name – Semaphore name

  • oflag

    Semaphore creation options. This may one of the following bit settings:

    • oflag = 0: Connect to the semaphore only if it already exists.

    • oflag = O_CREAT: Connect to the semaphore if it exists, otherwise create the semaphore.

    • oflag = O_CREAT with O_EXCL (O_CREAT|O_EXCL): Create a new semaphore unless one of this name already exists.

  • ...

    Optional parameters. NOTE: When the O_CREAT flag is specified, POSIX requires that a third and fourth parameter be supplied:

    • mode. The mode parameter is of type mode_t. This parameter is required but not used in the present implementation.

    • value. The value parameter is type unsigned int. The semaphore is created with an initial value of value. Valid initial values for semaphores must be less than or equal to SEM_VALUE_MAX (defined in include/limits.h).

Returns

A pointer to sem_t or SEM_FAILED if unsuccessful.

POSIX Compatibility: Comparable to the POSIX interface of the same name. Differences from the full POSIX implementation include:

  • Treatment of links/connections is highly simplified. It is just a counting semaphore.

int sem_close(sem_t *sem)

This function is called to indicate that the calling task is finished with the specified named semaphore, sem. The sem_close() deallocates any system resources allocated by the system for this named semaphore.

If the semaphore has not been removed with a call to sem_unlink(), then sem_close() has no effect on the named semaphore. However, when the named semaphore has been fully unlinked, the semaphore will vanish when the last task closes it.

Care must be taken to avoid risking the deletion of a semaphore that another calling task has already locked.

Parameters
  • sem – Semaphore descriptor

Returns

0 (OK), or -1 (ERROR) if unsuccessful.

Assumptions/Limitations:

  • Care must be taken to avoid deletion of a semaphore that another task has already locked.

  • sem_close() must not be called with an un-named semaphore.

POSIX Compatibility: Comparable to the POSIX interface of the same name.

This function will remove the semaphore named by the input name parameter. If one or more tasks have the semaphore named by name open when sem_unlink() is called, destruction of the semaphore will be postponed until all references have been destroyed by calls to sem_close().

Parameters
  • name – Semaphore name

Returns

0 (OK), or -1 (ERROR) if unsuccessful.

Assumptions/Limitations:

  • Care must be taken to avoid deletion of a semaphore that another task has already locked.

  • sem_unlink() must not be called with an un-named semaphore.

POSIX Compatibility: Comparable to the POSIX interface of the same name. Differences from the full POSIX implementation include:

  • Treatment of links/connections is highly simplified. It is just a counting semaphore.

  • Calls to sem_open() to re-create or re-connect to the semaphore may refer to the same semaphore; POSIX specifies that a new semaphore with the same name should be created after sem_unlink() is called.

int sem_wait(sem_t *sem)

This function attempts to lock the semaphore referenced by sem. If the semaphore as already locked by another task, the calling task will not return until it either successfully acquires the lock or the call is interrupted by a signal.

Parameters
  • sem – Semaphore descriptor.

Returns

0 (OK), or -1 (ERROR) is unsuccessful

If sem_wait returns -1 (ERROR) then the cause of the failure will be indicated by the thread-specific `errno <#ErrnoAccess>`__. The following lists the possible values for `errno <#ErrnoAccess>`__:

  • EINVAL: Indicates that the sem input parameter is not valid.

  • EINTR: Indicates that the wait was interrupt by a signal received by this task. In this case, the semaphore has not be acquired.

POSIX Compatibility: Comparable to the POSIX interface of the same name.

int sem_timedwait(sem_t *sem, const struct timespec *abstime)

This function will lock the semaphore referenced by sem as in the sem_wait() function. However, if the semaphore cannot be locked without waiting for another process or thread to unlock the semaphore by performing a sem_post() function, this wait will be terminated when the specified timeout expires.

The timeout will expire when the absolute time specified by abstime passes, as measured by the clock on which timeouts are based (that is, when the value of that clock equals or exceeds abstime), or if the absolute time specified by abstime has already been passed at the time of the call. This function attempts to lock the semaphore referenced by sem. If the semaphore is already locked by another task, the calling task will not return until it either successfully acquires the lock or the call is interrupted by a signal.

Input Parameters:

Parameters
  • sem – Semaphore descriptor.

  • abstime – The absolute time to wait until a timeout is declared.

Returns

0 (OK), or -1 (ERROR) is unsuccessful

If sem_timedwait returns -1 (ERROR) then the cause of the failure will be indicated by the thread-specific `errno <#ErrnoAccess>`__. The following lists the possible values for `errno <#ErrnoAccess>`__:

EINVAL: Indicates that the sem input parameter is not valid or the thread would have blocked, and the abstime parameter specified a nanoseconds field value less than zero or greater than or equal to 1000 million.

ETIMEDOUT: The semaphore could not be locked before the specified timeout expired.

EDEADLK: A deadlock condition was detected.

EINTR: Indicates that the wait was interrupt by a signal received by this task. In this case, the semaphore has not be acquired.

POSIX Compatibility: Derived from IEEE Std 1003.1d-1999.

int sem_trywait(sem_t *sem)

This function locks the specified semaphore only if the semaphore is currently not locked. In any event, the call returns without blocking.

Parameters
  • sem – The semaphore descriptor

Returns

0 (OK) or -1 (ERROR) if unsuccessful

If sem_trywait returns -1 (ERROR) then the cause of the failure will be indicated by the thread-specific `errno <#ErrnoAccess>`__. The following lists the possible values for `errno <#ErrnoAccess>`__:

  • EINVAL: Indicates that the sem input parameter is not valid.

  • EAGAIN: Indicates that the semaphore was not acquired.

POSIX Compatibility: Comparable to the POSIX interface of the same name.

int sem_post(sem_t *sem)

When a task has finished with a semaphore, it will call sem_post(). This function unlocks the semaphore referenced by sem by performing the semaphore unlock operation.

If the semaphore value resulting from this operation is positive, then no tasks were blocked waiting for the semaphore to become unlocked; The semaphore value is simply incremented.

If the value of the semaphore resulting from this operation is zero, then on of the tasks blocked waiting for the semaphore will be allowed to return successfully from its call to sem_wait().

Note

sem_post() may be called from an interrupt handler.

Parameters
  • sem – Semaphore descriptor

Returns

0 (OK) or -1 (ERROR) if unsuccessful.

Assumptions/Limitations:. When called from an interrupt handler, it will appear as though the interrupt task is the one that is performing the unlock.

POSIX Compatibility: Comparable to the POSIX interface of the same name.

int sem_getvalue(sem_t *sem, int *sval)

This function updates the location referenced by sval argument to have the value of the semaphore referenced by sem without effecting the state of the semaphore. The updated value represents the actual semaphore value that occurred at some unspecified time during the call, but may not reflect the actual value of the semaphore when it is returned to the calling task.

If sem is locked, the value return by sem_getvalue() will either be zero or a negative number whose absolute value represents the number of tasks waiting for the semaphore.

Parameters
  • sem – Semaphore descriptor

  • sval – Buffer by which the value is returned

Returns

0 (OK) or -1 (ERROR) if unsuccessful.

POSIX Compatibility: Comparable to the POSIX interface of the same name.

int sem_getprotocol(FAR const pthread_mutexattr_t *attr, FAR int *protocol)

Return the value of the semaphore protocol attribute.

Parameters
  • attr – A pointer to the semaphore to be queried

  • protocol – The user provided location in which to store the protocol value. May be one of SEM_PRIO_NONE, or SEM_PRIO_INHERIT, SEM_PRIO_PROTECT.

Returns

If successful, the sem_getprotocol() function will return zero (OK). Otherwise, an -1 (ERROR) will be returned and the errno value will be set to indicate the nature of the error.

POSIX Compatibility: Non-standard NuttX interface. Should not be used in portable code. Analogous to pthread_muxtexattr_getprotocol().

int sem_setprotocol(FAR pthread_mutexattr_t *attr, int protocol)

Set semaphore protocol attribute. See the paragraph Locking versus Signaling Semaphores for some important information about the use of this interface.

Parameters
  • attr – A pointer to the semaphore to be modified

  • protocol – The new protocol to use. One of SEM_PRIO_NONE, or SEM_PRIO_INHERIT, SEM_PRIO_PROTECT. SEM_PRIO_INHERIT is supported only if CONFIG_PRIORITY_INHERITANCE is defined; SEM_PRIO_PROTECT is not currently supported in any configuration.

Returns

If successful, the sem_setprotocol() function will return zero (OK). Otherwise, an -1 (ERROR) will be returned and the errno value will be set to indicate the nature of the error.

POSIX Compatibility: Non-standard NuttX interface. Should not be used in portable code. Analogous to pthread_muxtexattr_setprotocol().