10. Semaphore Manager¶
10.1. Introduction¶
The semaphore manager provides functions to allocate, delete, and control semaphores. This manager is based on the POSIX 1003.1 standard.
The directives provided by the semaphore manager are:
- sem_init - Initialize an unnamed semaphore
- sem_destroy - Destroy an unnamed semaphore
- sem_open - Open a named semaphore
- sem_close - Close a named semaphore
- sem_unlink - Remove a named semaphore
- sem_wait - Lock a semaphore
- sem_trywait - Lock a semaphore
- sem_timedwait - Wait on a Semaphore for a Specified Time
- sem_post - Unlock a semaphore
- sem_getvalue - Get the value of a semeaphore
10.2. Background¶
10.2.1. Theory¶
Semaphores are used for synchronization and mutual exclusion by indicating the availability and number of resources. The task (the task which is returning resources) notifying other tasks of an event increases the number of resources held by the semaphore by one. The task (the task which will obtain resources) waiting for the event decreases the number of resources held by the semaphore by one. If the number of resources held by a semaphore is insufficient (namely 0), the task requiring resources will wait until the next time resources are returned to the semaphore. If there is more than one task waiting for a semaphore, the tasks will be placed in the queue.
10.2.2. “sem_t” Structure¶
The sem_t
structure is used to represent semaphores. It is passed as an
argument to the semaphore directives and is defined as follows:
typedef int sem_t;
10.2.3. Building a Semaphore Attribute Set¶
10.3. Operations¶
10.3.1. Using as a Binary Semaphore¶
Although POSIX supports mutexes, they are only visible between threads. To work between processes, a binary semaphore must be used.
Creating a semaphore with a limit on the count of 1 effectively restricts the semaphore to being a binary semaphore. When the binary semaphore is available, the count is 1. When the binary semaphore is unavailable, the count is 0.
Since this does not result in a true binary semaphore, advanced binary features like the Priority Inheritance and Priority Ceiling Protocols are not available.
There is currently no text in this section.
10.4. Directives¶
This section details the semaphore manager’s directives. A subsection is dedicated to each of this manager’s directives and describes the calling sequence, related constants, usage, and status codes.
10.4.1. sem_init - Initialize an unnamed semaphore¶
CALLING SEQUENCE:
int sem_init(
sem_t *sem,
int pshared,
unsigned int value
);
STATUS CODES:
EINVAL |
The value argument exceeds SEM_VALUE_MAX |
ENOSPC |
A resource required to initialize the semaphore has been exhausted The
limit on semaphores (SEM_VALUE_MAX ) has been reached |
ENOSYS |
The function sem_init is not supported by this implementation |
EPERM |
The process lacks appropriate privileges to initialize the semaphore |
DESCRIPTION:
The sem_init
function is used to initialize the unnamed semaphore referred
to by sem
. The value of the initialized semaphore is the parameter
value
. The semaphore remains valid until it is destroyed.
NOTES:
If the functions completes successfully, it shall return a value of zero.
otherwise, it shall return a value of -1 and set errno
to specify the error
that occurred.
Multiprocessing is currently not supported in this implementation.
10.4.2. sem_destroy - Destroy an unnamed semaphore¶
CALLING SEQUENCE:
int sem_destroy(
sem_t *sem
);
STATUS CODES:
EINVAL |
The value argument exceeds SEM_VALUE_MAX |
ENOSYS |
The function sem_init is not supported by this implementation |
EBUSY |
There are currently processes blocked on the semaphore |
DESCRIPTION:
The sem_destroy
function is used to destroy an unnamed semaphore refered to
by sem
. sem_destroy
can only be used on a semaphore that was created
using sem_init.
NOTES:
If the functions completes successfully, it shall return a value of zero.
Otherwise, it shall return a value of -1 and set errno
to specify the error
that occurred.
Multiprocessing is currently not supported in this implementation.
10.4.3. sem_open - Open a named semaphore¶
CALLING SEQUENCE:
int sem_open(
const char *name,
int oflag
);
ARGUMENTS:
The following flag bit may be set in oflag:
O_CREAT |
Creates the semaphore if it does not already exist. If O_CREAT is set
and the semaphore already exists then O_CREAT has no
effect. Otherwise, sem_open() creates a semaphore. The O_CREAT
flag requires the third and fourth argument: mode and value of type
mode_t and unsigned int , respectively. |
O_EXCL |
If O_EXCL and O_CREAT are set, all call to sem_open() shall
fail if the semaphore name exists |
STATUS CODES:
EACCES |
Valid name specified but oflag permissions are denied, or the semaphore name specified does not exist and permission to create the named semaphore is denied. |
EEXIST |
O_CREAT and O_EXCL are set and the named semaphore already exists. |
EINTR |
The sem_open() operation was interrupted by a signal. |
EINVAL |
The sem_open() operation is not supported for the given name. |
EMFILE |
Too many semaphore descriptors or file descriptors in use by this process. |
ENAMETOOLONG |
The length of the name exceed PATH_MAX or name component is longer
than NAME_MAX while POSIX_NO_TRUNC is in effect. |
ENOENT |
O_CREAT is not set and the named semaphore does not exist. |
ENOSPC |
There is insufficient space for the creation of a new named semaphore. |
ENOSYS |
The function sem_open() is not supported by this implementation. |
DESCRIPTION:
The sem_open()
function establishes a connection between a specified
semaphore and a process. After a call to sem_open with a specified semaphore
name, a process can reference to semaphore by the associated name using the
address returned by the call. The oflag arguments listed above control the
state of the semaphore by determining if the semaphore is created or accessed
by a call to sem_open()
.
NOTES:
10.4.4. sem_close - Close a named semaphore¶
CALLING SEQUENCE:
int sem_close(
sem_t *sem_close
);
STATUS CODES:
EACCES |
The semaphore argument is not a valid semaphore descriptor. |
ENOSYS |
The function sem_close is not supported by this implementation. |
DESCRIPTION:
The sem_close()
function is used to indicate that the calling process is
finished using the named semaphore indicated by sem
. The function
sem_close
deallocates any system resources that were previously allocated
by a sem_open
system call. If sem_close()
completes successfully it
returns a 1, otherwise a value of -1 is return and errno
is set.
NOTES:
10.4.5. sem_unlink - Unlink a semaphore¶
CALLING SEQUENCE:
int sem_unlink(
const char *name
);
STATUS CODES:
EACCESS |
Permission is denied to unlink a semaphore. |
ENAMETOOLONG |
The length of the strong name exceed NAME_MAX while POSIX_NO_TRUNC
is in effect. |
ENOENT |
The name of the semaphore does not exist. |
ENOSPC |
There is insufficient space for the creation of a new named semaphore. |
ENOSYS |
The function sem_unlink is not supported by this implementation. |
DESCRIPTION:
The sem_unlink()
function shall remove the semaphore name by the string
name. If a process is currently accessing the name semaphore, the
sem_unlink
command has no effect. If one or more processes have the
semaphore open when the sem_unlink
function is called, the destruction of
semaphores shall be postponed until all reference to semaphore are destroyed by
calls to sem_close
, _exit()
, or exec
. After all references have
been destroyed, it returns immediately.
If the termination is successful, the function shall return 0. Otherwise, a -1
is returned and the errno
is set.
NOTES:
10.4.6. sem_wait - Wait on a Semaphore¶
CALLING SEQUENCE:
int sem_wait(
sem_t *sem
);
STATUS CODES:
EINVAL |
The sem argument does not refer to a valid semaphore |
DESCRIPTION:
This function attempts to lock a semaphore specified by sem
. If the
semaphore is available, then the semaphore is locked (i.e., the semaphore
value is decremented). If the semaphore is unavailable (i.e., the semaphore
value is zero), then the function will block until the semaphore becomes
available. It will then successfully lock the semaphore. The semaphore
remains locked until released by a sem_post()
call.
If the call is unsuccessful, then the function returns -1 and sets errno
to
the appropriate error code.
NOTES:
Multiprocessing is not supported in this implementation.
10.4.7. sem_trywait - Non-blocking Wait on a Semaphore¶
CALLING SEQUENCE:
int sem_trywait(
sem_t *sem
);
STATUS CODES:
EAGAIN |
The semaphore is not available (i.e., the semaphore value is zero), so the semaphore could not be locked. |
EINVAL |
The sem argument does not refewr to a valid semaphore |
DESCRIPTION:
This function attempts to lock a semaphore specified by sem
. If the
semaphore is available, then the semaphore is locked (i.e., the semaphore value
is decremented) and the function returns a value of 0. The semaphore remains
locked until released by a sem_post()
call. If the semaphore is unavailable
(i.e., the semaphore value is zero), then the function will return a value
of -1 immediately and set errno
to EAGAIN
.
If the call is unsuccessful, then the function returns -1 and sets errno
to
the appropriate error code.
NOTES:
Multiprocessing is not supported in this implementation.
10.4.8. sem_timedwait - Wait on a Semaphore for a Specified Time¶
CALLING SEQUENCE:
int sem_timedwait(
sem_t *sem,
const struct timespec *abstime
);
STATUS CODES:
EAGAIN |
The semaphore is not available (i.e., the semaphore value is zero), so the semaphore could not be locked. |
EINVAL |
The sem argument does not refewr to a valid semaphore |
DESCRIPTION:
This function attemtps to lock a semaphore specified by sem
, and will wait
for the semaphore until the absolute time specified by abstime
. If the
semaphore is available, then the semaphore is locked (i.e., the semaphore value
is decremented) and the function returns a value of 0. The semaphore remains
locked until released by a sem_post()
call. If the semaphore is
unavailable, then the function will wait for the semaphore to become available
for the amount of time specified by timeout
.
If the semaphore does not become available within the interval specified by
timeout
, then the function returns -1 and sets errno
to EAGAIN
. If
any other error occurs, the function returns -1 and sets errno
to the
appropriate error code.
NOTES:
Multiprocessing is not supported in this implementation.
10.4.9. sem_post - Unlock a Semaphore¶
CALLING SEQUENCE:
int sem_post(
sem_t *sem
);
STATUS CODES:
EINVAL |
The sem argument does not refer to a valid semaphore |
DESCRIPTION:
This function attempts to release the semaphore specified by sem
. If other
tasks are waiting on the semaphore, then one of those tasks (which one depends
on the scheduler being used) is allowed to lock the semaphore and return from
its sem_wait()
, sem_trywait()
, or sem_timedwait()
call. If there
are no other tasks waiting on the semaphore, then the semaphore value is simply
incremented. sem_post()
returns 0 upon successful completion.
If an error occurs, the function returns -1 and sets errno
to the
appropriate error code.
NOTES:
Multiprocessing is not supported in this implementation.
10.4.10. sem_getvalue - Get the value of a semaphore¶
CALLING SEQUENCE:
int sem_getvalue(
sem_t *sem,
int *sval
);
STATUS CODES:
EINVAL |
The sem argument does not refer to a valid semaphore |
ENOSYS |
The function sem_getvalue is not supported by this implementation |
DESCRIPTION:
The sem_getvalue
functions sets the location referenced by the sval
argument to the value of the semaphore without affecting the state of the
semaphore. The updated value represents a semaphore value that occurred at some
point during the call, but is not necessarily the actual value of the semaphore
when it returns to the calling process.
If sem
is locked, the value returned by sem_getvalue
will be zero or a
negative number whose absolute value is the number of processes waiting for the
semaphore at some point during the call.
NOTES:
If the functions completes successfully, it shall return a value of zero.
Otherwise, it shall return a value of -1 and set errno
to specify the error
that occurred.