Use umtx to implement process sharable semaphore, to make this work,
now type sema_t is a structure which can be put in a shared memory area,
and multiple processes can operate it concurrently.
User can either use mmap(MAP_SHARED) + sem_init(pshared=1) or use sem_open()
to initialize a shared semaphore.
Named semaphore uses file system and is located in /tmp directory, and its
file name is prefixed with 'SEMD', so now it is chroot or jail friendly.
In simplist cases, both for named and un-named semaphore, userland code
does not have to enter kernel to reduce/increase semaphore's count.
The semaphore is designed to be crash-safe, it means even if an application
is crashed in the middle of operating semaphore, the semaphore state is
still safely recovered by later use, there is no waiter counter maintained
by userland code.
The main semaphore code is in libc and libthr only has some necessary stubs,
this makes it possible that a non-threaded application can use semaphore
without linking to thread library.
Old semaphore implementation is kept libc to maintain binary compatibility.
The kernel ksem API is no longer used in the new implemenation.
Discussed on: threads@
2010-01-05 02:37:59 +00:00
|
|
|
/*
|
|
|
|
* Copyright (C) 2010 David Xu <davidxu@freebsd.org>.
|
|
|
|
* All rights reserved.
|
|
|
|
*
|
|
|
|
* Redistribution and use in source and binary forms, with or without
|
|
|
|
* modification, are permitted provided that the following conditions
|
|
|
|
* are met:
|
|
|
|
* 1. Redistributions of source code must retain the above copyright
|
|
|
|
* notice(s), this list of conditions and the following disclaimer as
|
|
|
|
* the first lines of this file unmodified other than the possible
|
|
|
|
* addition of one or more copyright notices.
|
|
|
|
* 2. Redistributions in binary form must reproduce the above copyright
|
|
|
|
* notice(s), this list of conditions and the following disclaimer in
|
|
|
|
* the documentation and/or other materials provided with the
|
|
|
|
* distribution.
|
|
|
|
*
|
|
|
|
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY
|
|
|
|
* EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
|
|
|
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
|
|
|
|
* PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER(S) BE
|
|
|
|
* LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
|
|
|
|
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
|
|
|
|
* SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
|
|
|
|
* BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
|
|
|
|
* WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
|
|
|
|
* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
|
|
|
|
* EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
|
|
|
|
*
|
|
|
|
* $FreeBSD$
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include "namespace.h"
|
|
|
|
#include <sys/types.h>
|
|
|
|
#include <sys/queue.h>
|
|
|
|
#include <sys/mman.h>
|
|
|
|
#include <sys/stat.h>
|
|
|
|
#include <errno.h>
|
|
|
|
#include <machine/atomic.h>
|
|
|
|
#include <sys/umtx.h>
|
|
|
|
#include <limits.h>
|
|
|
|
#include <fcntl.h>
|
|
|
|
#include <pthread.h>
|
|
|
|
#include <stdarg.h>
|
|
|
|
#include <stdlib.h>
|
|
|
|
#include <string.h>
|
|
|
|
#include <time.h>
|
|
|
|
#include <semaphore.h>
|
|
|
|
#include <unistd.h>
|
|
|
|
#include "un-namespace.h"
|
2010-09-25 01:57:47 +00:00
|
|
|
#include "libc_private.h"
|
Use umtx to implement process sharable semaphore, to make this work,
now type sema_t is a structure which can be put in a shared memory area,
and multiple processes can operate it concurrently.
User can either use mmap(MAP_SHARED) + sem_init(pshared=1) or use sem_open()
to initialize a shared semaphore.
Named semaphore uses file system and is located in /tmp directory, and its
file name is prefixed with 'SEMD', so now it is chroot or jail friendly.
In simplist cases, both for named and un-named semaphore, userland code
does not have to enter kernel to reduce/increase semaphore's count.
The semaphore is designed to be crash-safe, it means even if an application
is crashed in the middle of operating semaphore, the semaphore state is
still safely recovered by later use, there is no waiter counter maintained
by userland code.
The main semaphore code is in libc and libthr only has some necessary stubs,
this makes it possible that a non-threaded application can use semaphore
without linking to thread library.
Old semaphore implementation is kept libc to maintain binary compatibility.
The kernel ksem API is no longer used in the new implemenation.
Discussed on: threads@
2010-01-05 02:37:59 +00:00
|
|
|
|
2010-01-05 06:40:27 +00:00
|
|
|
__weak_reference(_sem_close, sem_close);
|
|
|
|
__weak_reference(_sem_destroy, sem_destroy);
|
|
|
|
__weak_reference(_sem_getvalue, sem_getvalue);
|
|
|
|
__weak_reference(_sem_init, sem_init);
|
|
|
|
__weak_reference(_sem_open, sem_open);
|
|
|
|
__weak_reference(_sem_post, sem_post);
|
|
|
|
__weak_reference(_sem_timedwait, sem_timedwait);
|
|
|
|
__weak_reference(_sem_trywait, sem_trywait);
|
|
|
|
__weak_reference(_sem_unlink, sem_unlink);
|
|
|
|
__weak_reference(_sem_wait, sem_wait);
|
Use umtx to implement process sharable semaphore, to make this work,
now type sema_t is a structure which can be put in a shared memory area,
and multiple processes can operate it concurrently.
User can either use mmap(MAP_SHARED) + sem_init(pshared=1) or use sem_open()
to initialize a shared semaphore.
Named semaphore uses file system and is located in /tmp directory, and its
file name is prefixed with 'SEMD', so now it is chroot or jail friendly.
In simplist cases, both for named and un-named semaphore, userland code
does not have to enter kernel to reduce/increase semaphore's count.
The semaphore is designed to be crash-safe, it means even if an application
is crashed in the middle of operating semaphore, the semaphore state is
still safely recovered by later use, there is no waiter counter maintained
by userland code.
The main semaphore code is in libc and libthr only has some necessary stubs,
this makes it possible that a non-threaded application can use semaphore
without linking to thread library.
Old semaphore implementation is kept libc to maintain binary compatibility.
The kernel ksem API is no longer used in the new implemenation.
Discussed on: threads@
2010-01-05 02:37:59 +00:00
|
|
|
|
|
|
|
#define SEM_PREFIX "/tmp/SEMD"
|
|
|
|
#define SEM_MAGIC ((u_int32_t)0x73656d31)
|
|
|
|
|
|
|
|
struct sem_nameinfo {
|
|
|
|
int open_count;
|
|
|
|
char *name;
|
|
|
|
sem_t *sem;
|
|
|
|
LIST_ENTRY(sem_nameinfo) next;
|
|
|
|
};
|
|
|
|
|
|
|
|
static pthread_once_t once = PTHREAD_ONCE_INIT;
|
|
|
|
static pthread_mutex_t sem_llock;
|
|
|
|
static LIST_HEAD(,sem_nameinfo) sem_list = LIST_HEAD_INITIALIZER(sem_list);
|
|
|
|
|
|
|
|
static void
|
|
|
|
sem_prefork()
|
|
|
|
{
|
|
|
|
|
|
|
|
_pthread_mutex_lock(&sem_llock);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
sem_postfork()
|
|
|
|
{
|
|
|
|
_pthread_mutex_unlock(&sem_llock);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
sem_child_postfork()
|
|
|
|
{
|
|
|
|
_pthread_mutex_unlock(&sem_llock);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
sem_module_init(void)
|
|
|
|
{
|
|
|
|
pthread_mutexattr_t ma;
|
|
|
|
|
|
|
|
_pthread_mutexattr_init(&ma);
|
|
|
|
_pthread_mutexattr_settype(&ma, PTHREAD_MUTEX_RECURSIVE);
|
|
|
|
_pthread_mutex_init(&sem_llock, &ma);
|
|
|
|
_pthread_mutexattr_destroy(&ma);
|
|
|
|
_pthread_atfork(sem_prefork, sem_postfork, sem_child_postfork);
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline int
|
|
|
|
sem_check_validity(sem_t *sem)
|
|
|
|
{
|
|
|
|
|
|
|
|
if (sem->_magic == SEM_MAGIC)
|
|
|
|
return (0);
|
|
|
|
else {
|
|
|
|
errno = EINVAL;
|
|
|
|
return (-1);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
int
|
2010-01-05 06:40:27 +00:00
|
|
|
_sem_init(sem_t *sem, int pshared, unsigned int value)
|
Use umtx to implement process sharable semaphore, to make this work,
now type sema_t is a structure which can be put in a shared memory area,
and multiple processes can operate it concurrently.
User can either use mmap(MAP_SHARED) + sem_init(pshared=1) or use sem_open()
to initialize a shared semaphore.
Named semaphore uses file system and is located in /tmp directory, and its
file name is prefixed with 'SEMD', so now it is chroot or jail friendly.
In simplist cases, both for named and un-named semaphore, userland code
does not have to enter kernel to reduce/increase semaphore's count.
The semaphore is designed to be crash-safe, it means even if an application
is crashed in the middle of operating semaphore, the semaphore state is
still safely recovered by later use, there is no waiter counter maintained
by userland code.
The main semaphore code is in libc and libthr only has some necessary stubs,
this makes it possible that a non-threaded application can use semaphore
without linking to thread library.
Old semaphore implementation is kept libc to maintain binary compatibility.
The kernel ksem API is no longer used in the new implemenation.
Discussed on: threads@
2010-01-05 02:37:59 +00:00
|
|
|
{
|
|
|
|
|
|
|
|
if (value > SEM_VALUE_MAX) {
|
|
|
|
errno = EINVAL;
|
|
|
|
return (-1);
|
|
|
|
}
|
|
|
|
|
|
|
|
bzero(sem, sizeof(sem_t));
|
|
|
|
sem->_magic = SEM_MAGIC;
|
|
|
|
sem->_kern._count = (u_int32_t)value;
|
|
|
|
sem->_kern._has_waiters = 0;
|
|
|
|
sem->_kern._flags = pshared ? USYNC_PROCESS_SHARED : 0;
|
|
|
|
return (0);
|
|
|
|
}
|
|
|
|
|
|
|
|
sem_t *
|
2010-01-05 06:40:27 +00:00
|
|
|
_sem_open(const char *name, int flags, ...)
|
Use umtx to implement process sharable semaphore, to make this work,
now type sema_t is a structure which can be put in a shared memory area,
and multiple processes can operate it concurrently.
User can either use mmap(MAP_SHARED) + sem_init(pshared=1) or use sem_open()
to initialize a shared semaphore.
Named semaphore uses file system and is located in /tmp directory, and its
file name is prefixed with 'SEMD', so now it is chroot or jail friendly.
In simplist cases, both for named and un-named semaphore, userland code
does not have to enter kernel to reduce/increase semaphore's count.
The semaphore is designed to be crash-safe, it means even if an application
is crashed in the middle of operating semaphore, the semaphore state is
still safely recovered by later use, there is no waiter counter maintained
by userland code.
The main semaphore code is in libc and libthr only has some necessary stubs,
this makes it possible that a non-threaded application can use semaphore
without linking to thread library.
Old semaphore implementation is kept libc to maintain binary compatibility.
The kernel ksem API is no longer used in the new implemenation.
Discussed on: threads@
2010-01-05 02:37:59 +00:00
|
|
|
{
|
|
|
|
char path[PATH_MAX];
|
|
|
|
|
|
|
|
struct stat sb;
|
|
|
|
va_list ap;
|
|
|
|
struct sem_nameinfo *ni = NULL;
|
|
|
|
sem_t *sem = NULL;
|
2010-01-18 10:29:04 +00:00
|
|
|
int fd = -1, mode, len, errsave;
|
2010-01-07 04:15:49 +00:00
|
|
|
int value = 0;
|
Use umtx to implement process sharable semaphore, to make this work,
now type sema_t is a structure which can be put in a shared memory area,
and multiple processes can operate it concurrently.
User can either use mmap(MAP_SHARED) + sem_init(pshared=1) or use sem_open()
to initialize a shared semaphore.
Named semaphore uses file system and is located in /tmp directory, and its
file name is prefixed with 'SEMD', so now it is chroot or jail friendly.
In simplist cases, both for named and un-named semaphore, userland code
does not have to enter kernel to reduce/increase semaphore's count.
The semaphore is designed to be crash-safe, it means even if an application
is crashed in the middle of operating semaphore, the semaphore state is
still safely recovered by later use, there is no waiter counter maintained
by userland code.
The main semaphore code is in libc and libthr only has some necessary stubs,
this makes it possible that a non-threaded application can use semaphore
without linking to thread library.
Old semaphore implementation is kept libc to maintain binary compatibility.
The kernel ksem API is no longer used in the new implemenation.
Discussed on: threads@
2010-01-05 02:37:59 +00:00
|
|
|
|
|
|
|
if (name[0] != '/') {
|
|
|
|
errno = EINVAL;
|
2010-01-13 08:53:23 +00:00
|
|
|
return (SEM_FAILED);
|
Use umtx to implement process sharable semaphore, to make this work,
now type sema_t is a structure which can be put in a shared memory area,
and multiple processes can operate it concurrently.
User can either use mmap(MAP_SHARED) + sem_init(pshared=1) or use sem_open()
to initialize a shared semaphore.
Named semaphore uses file system and is located in /tmp directory, and its
file name is prefixed with 'SEMD', so now it is chroot or jail friendly.
In simplist cases, both for named and un-named semaphore, userland code
does not have to enter kernel to reduce/increase semaphore's count.
The semaphore is designed to be crash-safe, it means even if an application
is crashed in the middle of operating semaphore, the semaphore state is
still safely recovered by later use, there is no waiter counter maintained
by userland code.
The main semaphore code is in libc and libthr only has some necessary stubs,
this makes it possible that a non-threaded application can use semaphore
without linking to thread library.
Old semaphore implementation is kept libc to maintain binary compatibility.
The kernel ksem API is no longer used in the new implemenation.
Discussed on: threads@
2010-01-05 02:37:59 +00:00
|
|
|
}
|
|
|
|
name++;
|
|
|
|
|
|
|
|
if (flags & ~(O_CREAT|O_EXCL)) {
|
|
|
|
errno = EINVAL;
|
2010-01-13 08:53:23 +00:00
|
|
|
return (SEM_FAILED);
|
Use umtx to implement process sharable semaphore, to make this work,
now type sema_t is a structure which can be put in a shared memory area,
and multiple processes can operate it concurrently.
User can either use mmap(MAP_SHARED) + sem_init(pshared=1) or use sem_open()
to initialize a shared semaphore.
Named semaphore uses file system and is located in /tmp directory, and its
file name is prefixed with 'SEMD', so now it is chroot or jail friendly.
In simplist cases, both for named and un-named semaphore, userland code
does not have to enter kernel to reduce/increase semaphore's count.
The semaphore is designed to be crash-safe, it means even if an application
is crashed in the middle of operating semaphore, the semaphore state is
still safely recovered by later use, there is no waiter counter maintained
by userland code.
The main semaphore code is in libc and libthr only has some necessary stubs,
this makes it possible that a non-threaded application can use semaphore
without linking to thread library.
Old semaphore implementation is kept libc to maintain binary compatibility.
The kernel ksem API is no longer used in the new implemenation.
Discussed on: threads@
2010-01-05 02:37:59 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
_pthread_once(&once, sem_module_init);
|
|
|
|
|
|
|
|
_pthread_mutex_lock(&sem_llock);
|
|
|
|
LIST_FOREACH(ni, &sem_list, next) {
|
|
|
|
if (strcmp(name, ni->name) == 0) {
|
|
|
|
ni->open_count++;
|
|
|
|
sem = ni->sem;
|
|
|
|
_pthread_mutex_unlock(&sem_llock);
|
|
|
|
return (sem);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
if (flags & O_CREAT) {
|
|
|
|
va_start(ap, flags);
|
|
|
|
mode = va_arg(ap, int);
|
2010-01-07 04:15:49 +00:00
|
|
|
value = va_arg(ap, int);
|
Use umtx to implement process sharable semaphore, to make this work,
now type sema_t is a structure which can be put in a shared memory area,
and multiple processes can operate it concurrently.
User can either use mmap(MAP_SHARED) + sem_init(pshared=1) or use sem_open()
to initialize a shared semaphore.
Named semaphore uses file system and is located in /tmp directory, and its
file name is prefixed with 'SEMD', so now it is chroot or jail friendly.
In simplist cases, both for named and un-named semaphore, userland code
does not have to enter kernel to reduce/increase semaphore's count.
The semaphore is designed to be crash-safe, it means even if an application
is crashed in the middle of operating semaphore, the semaphore state is
still safely recovered by later use, there is no waiter counter maintained
by userland code.
The main semaphore code is in libc and libthr only has some necessary stubs,
this makes it possible that a non-threaded application can use semaphore
without linking to thread library.
Old semaphore implementation is kept libc to maintain binary compatibility.
The kernel ksem API is no longer used in the new implemenation.
Discussed on: threads@
2010-01-05 02:37:59 +00:00
|
|
|
va_end(ap);
|
|
|
|
}
|
|
|
|
|
|
|
|
len = sizeof(*ni) + strlen(name) + 1;
|
|
|
|
ni = (struct sem_nameinfo *)malloc(len);
|
|
|
|
if (ni == NULL) {
|
|
|
|
errno = ENOSPC;
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
|
|
|
|
ni->name = (char *)(ni+1);
|
|
|
|
strcpy(ni->name, name);
|
|
|
|
|
|
|
|
strcpy(path, SEM_PREFIX);
|
|
|
|
if (strlcat(path, name, sizeof(path)) >= sizeof(path)) {
|
|
|
|
errno = ENAMETOOLONG;
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
|
|
|
|
fd = _open(path, flags|O_RDWR, mode);
|
|
|
|
if (fd == -1)
|
|
|
|
goto error;
|
|
|
|
if (flock(fd, LOCK_EX) == -1)
|
|
|
|
goto error;
|
|
|
|
if (_fstat(fd, &sb)) {
|
|
|
|
flock(fd, LOCK_UN);
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
if (sb.st_size < sizeof(sem_t)) {
|
|
|
|
sem_t tmp;
|
|
|
|
|
|
|
|
tmp._magic = SEM_MAGIC;
|
|
|
|
tmp._kern._has_waiters = 0;
|
2010-01-07 04:15:49 +00:00
|
|
|
tmp._kern._count = value;
|
Use umtx to implement process sharable semaphore, to make this work,
now type sema_t is a structure which can be put in a shared memory area,
and multiple processes can operate it concurrently.
User can either use mmap(MAP_SHARED) + sem_init(pshared=1) or use sem_open()
to initialize a shared semaphore.
Named semaphore uses file system and is located in /tmp directory, and its
file name is prefixed with 'SEMD', so now it is chroot or jail friendly.
In simplist cases, both for named and un-named semaphore, userland code
does not have to enter kernel to reduce/increase semaphore's count.
The semaphore is designed to be crash-safe, it means even if an application
is crashed in the middle of operating semaphore, the semaphore state is
still safely recovered by later use, there is no waiter counter maintained
by userland code.
The main semaphore code is in libc and libthr only has some necessary stubs,
this makes it possible that a non-threaded application can use semaphore
without linking to thread library.
Old semaphore implementation is kept libc to maintain binary compatibility.
The kernel ksem API is no longer used in the new implemenation.
Discussed on: threads@
2010-01-05 02:37:59 +00:00
|
|
|
tmp._kern._flags = USYNC_PROCESS_SHARED | SEM_NAMED;
|
|
|
|
if (_write(fd, &tmp, sizeof(tmp)) != sizeof(tmp)) {
|
|
|
|
flock(fd, LOCK_UN);
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
flock(fd, LOCK_UN);
|
|
|
|
sem = (sem_t *)mmap(NULL, sizeof(sem_t), PROT_READ|PROT_WRITE,
|
|
|
|
MAP_SHARED|MAP_NOSYNC, fd, 0);
|
|
|
|
if (sem == MAP_FAILED) {
|
|
|
|
sem = NULL;
|
|
|
|
if (errno == ENOMEM)
|
|
|
|
errno = ENOSPC;
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
if (sem->_magic != SEM_MAGIC) {
|
|
|
|
errno = EINVAL;
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
ni->open_count = 1;
|
|
|
|
ni->sem = sem;
|
|
|
|
LIST_INSERT_HEAD(&sem_list, ni, next);
|
|
|
|
_pthread_mutex_unlock(&sem_llock);
|
|
|
|
_close(fd);
|
|
|
|
return (sem);
|
|
|
|
|
|
|
|
error:
|
2010-01-18 10:29:04 +00:00
|
|
|
errsave = errno;
|
Use umtx to implement process sharable semaphore, to make this work,
now type sema_t is a structure which can be put in a shared memory area,
and multiple processes can operate it concurrently.
User can either use mmap(MAP_SHARED) + sem_init(pshared=1) or use sem_open()
to initialize a shared semaphore.
Named semaphore uses file system and is located in /tmp directory, and its
file name is prefixed with 'SEMD', so now it is chroot or jail friendly.
In simplist cases, both for named and un-named semaphore, userland code
does not have to enter kernel to reduce/increase semaphore's count.
The semaphore is designed to be crash-safe, it means even if an application
is crashed in the middle of operating semaphore, the semaphore state is
still safely recovered by later use, there is no waiter counter maintained
by userland code.
The main semaphore code is in libc and libthr only has some necessary stubs,
this makes it possible that a non-threaded application can use semaphore
without linking to thread library.
Old semaphore implementation is kept libc to maintain binary compatibility.
The kernel ksem API is no longer used in the new implemenation.
Discussed on: threads@
2010-01-05 02:37:59 +00:00
|
|
|
_pthread_mutex_unlock(&sem_llock);
|
|
|
|
if (fd != -1)
|
|
|
|
_close(fd);
|
|
|
|
if (sem != NULL)
|
|
|
|
munmap(sem, sizeof(sem_t));
|
|
|
|
free(ni);
|
2010-01-18 10:29:04 +00:00
|
|
|
errno = errsave;
|
Use umtx to implement process sharable semaphore, to make this work,
now type sema_t is a structure which can be put in a shared memory area,
and multiple processes can operate it concurrently.
User can either use mmap(MAP_SHARED) + sem_init(pshared=1) or use sem_open()
to initialize a shared semaphore.
Named semaphore uses file system and is located in /tmp directory, and its
file name is prefixed with 'SEMD', so now it is chroot or jail friendly.
In simplist cases, both for named and un-named semaphore, userland code
does not have to enter kernel to reduce/increase semaphore's count.
The semaphore is designed to be crash-safe, it means even if an application
is crashed in the middle of operating semaphore, the semaphore state is
still safely recovered by later use, there is no waiter counter maintained
by userland code.
The main semaphore code is in libc and libthr only has some necessary stubs,
this makes it possible that a non-threaded application can use semaphore
without linking to thread library.
Old semaphore implementation is kept libc to maintain binary compatibility.
The kernel ksem API is no longer used in the new implemenation.
Discussed on: threads@
2010-01-05 02:37:59 +00:00
|
|
|
return (SEM_FAILED);
|
|
|
|
}
|
|
|
|
|
|
|
|
int
|
2010-01-05 06:40:27 +00:00
|
|
|
_sem_close(sem_t *sem)
|
Use umtx to implement process sharable semaphore, to make this work,
now type sema_t is a structure which can be put in a shared memory area,
and multiple processes can operate it concurrently.
User can either use mmap(MAP_SHARED) + sem_init(pshared=1) or use sem_open()
to initialize a shared semaphore.
Named semaphore uses file system and is located in /tmp directory, and its
file name is prefixed with 'SEMD', so now it is chroot or jail friendly.
In simplist cases, both for named and un-named semaphore, userland code
does not have to enter kernel to reduce/increase semaphore's count.
The semaphore is designed to be crash-safe, it means even if an application
is crashed in the middle of operating semaphore, the semaphore state is
still safely recovered by later use, there is no waiter counter maintained
by userland code.
The main semaphore code is in libc and libthr only has some necessary stubs,
this makes it possible that a non-threaded application can use semaphore
without linking to thread library.
Old semaphore implementation is kept libc to maintain binary compatibility.
The kernel ksem API is no longer used in the new implemenation.
Discussed on: threads@
2010-01-05 02:37:59 +00:00
|
|
|
{
|
|
|
|
struct sem_nameinfo *ni;
|
|
|
|
|
|
|
|
if (sem_check_validity(sem) != 0)
|
|
|
|
return (-1);
|
|
|
|
|
|
|
|
if (!(sem->_kern._flags & SEM_NAMED)) {
|
|
|
|
errno = EINVAL;
|
|
|
|
return (-1);
|
|
|
|
}
|
|
|
|
|
2010-01-15 01:19:58 +00:00
|
|
|
_pthread_once(&once, sem_module_init);
|
|
|
|
|
Use umtx to implement process sharable semaphore, to make this work,
now type sema_t is a structure which can be put in a shared memory area,
and multiple processes can operate it concurrently.
User can either use mmap(MAP_SHARED) + sem_init(pshared=1) or use sem_open()
to initialize a shared semaphore.
Named semaphore uses file system and is located in /tmp directory, and its
file name is prefixed with 'SEMD', so now it is chroot or jail friendly.
In simplist cases, both for named and un-named semaphore, userland code
does not have to enter kernel to reduce/increase semaphore's count.
The semaphore is designed to be crash-safe, it means even if an application
is crashed in the middle of operating semaphore, the semaphore state is
still safely recovered by later use, there is no waiter counter maintained
by userland code.
The main semaphore code is in libc and libthr only has some necessary stubs,
this makes it possible that a non-threaded application can use semaphore
without linking to thread library.
Old semaphore implementation is kept libc to maintain binary compatibility.
The kernel ksem API is no longer used in the new implemenation.
Discussed on: threads@
2010-01-05 02:37:59 +00:00
|
|
|
_pthread_mutex_lock(&sem_llock);
|
|
|
|
LIST_FOREACH(ni, &sem_list, next) {
|
|
|
|
if (sem == ni->sem) {
|
|
|
|
if (--ni->open_count > 0) {
|
|
|
|
_pthread_mutex_unlock(&sem_llock);
|
|
|
|
return (0);
|
|
|
|
}
|
|
|
|
else
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
if (ni) {
|
|
|
|
LIST_REMOVE(ni, next);
|
|
|
|
_pthread_mutex_unlock(&sem_llock);
|
|
|
|
munmap(sem, sizeof(*sem));
|
|
|
|
free(ni);
|
|
|
|
return (0);
|
|
|
|
}
|
|
|
|
_pthread_mutex_unlock(&sem_llock);
|
2010-01-13 08:53:23 +00:00
|
|
|
errno = EINVAL;
|
Use umtx to implement process sharable semaphore, to make this work,
now type sema_t is a structure which can be put in a shared memory area,
and multiple processes can operate it concurrently.
User can either use mmap(MAP_SHARED) + sem_init(pshared=1) or use sem_open()
to initialize a shared semaphore.
Named semaphore uses file system and is located in /tmp directory, and its
file name is prefixed with 'SEMD', so now it is chroot or jail friendly.
In simplist cases, both for named and un-named semaphore, userland code
does not have to enter kernel to reduce/increase semaphore's count.
The semaphore is designed to be crash-safe, it means even if an application
is crashed in the middle of operating semaphore, the semaphore state is
still safely recovered by later use, there is no waiter counter maintained
by userland code.
The main semaphore code is in libc and libthr only has some necessary stubs,
this makes it possible that a non-threaded application can use semaphore
without linking to thread library.
Old semaphore implementation is kept libc to maintain binary compatibility.
The kernel ksem API is no longer used in the new implemenation.
Discussed on: threads@
2010-01-05 02:37:59 +00:00
|
|
|
return (-1);
|
|
|
|
}
|
|
|
|
|
|
|
|
int
|
2010-01-05 06:40:27 +00:00
|
|
|
_sem_unlink(const char *name)
|
Use umtx to implement process sharable semaphore, to make this work,
now type sema_t is a structure which can be put in a shared memory area,
and multiple processes can operate it concurrently.
User can either use mmap(MAP_SHARED) + sem_init(pshared=1) or use sem_open()
to initialize a shared semaphore.
Named semaphore uses file system and is located in /tmp directory, and its
file name is prefixed with 'SEMD', so now it is chroot or jail friendly.
In simplist cases, both for named and un-named semaphore, userland code
does not have to enter kernel to reduce/increase semaphore's count.
The semaphore is designed to be crash-safe, it means even if an application
is crashed in the middle of operating semaphore, the semaphore state is
still safely recovered by later use, there is no waiter counter maintained
by userland code.
The main semaphore code is in libc and libthr only has some necessary stubs,
this makes it possible that a non-threaded application can use semaphore
without linking to thread library.
Old semaphore implementation is kept libc to maintain binary compatibility.
The kernel ksem API is no longer used in the new implemenation.
Discussed on: threads@
2010-01-05 02:37:59 +00:00
|
|
|
{
|
|
|
|
char path[PATH_MAX];
|
|
|
|
|
|
|
|
if (name[0] != '/') {
|
|
|
|
errno = ENOENT;
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
name++;
|
|
|
|
|
|
|
|
strcpy(path, SEM_PREFIX);
|
|
|
|
if (strlcat(path, name, sizeof(path)) >= sizeof(path)) {
|
|
|
|
errno = ENAMETOOLONG;
|
|
|
|
return (-1);
|
|
|
|
}
|
|
|
|
return unlink(path);
|
|
|
|
}
|
|
|
|
|
|
|
|
int
|
2010-01-05 06:40:27 +00:00
|
|
|
_sem_destroy(sem_t *sem)
|
Use umtx to implement process sharable semaphore, to make this work,
now type sema_t is a structure which can be put in a shared memory area,
and multiple processes can operate it concurrently.
User can either use mmap(MAP_SHARED) + sem_init(pshared=1) or use sem_open()
to initialize a shared semaphore.
Named semaphore uses file system and is located in /tmp directory, and its
file name is prefixed with 'SEMD', so now it is chroot or jail friendly.
In simplist cases, both for named and un-named semaphore, userland code
does not have to enter kernel to reduce/increase semaphore's count.
The semaphore is designed to be crash-safe, it means even if an application
is crashed in the middle of operating semaphore, the semaphore state is
still safely recovered by later use, there is no waiter counter maintained
by userland code.
The main semaphore code is in libc and libthr only has some necessary stubs,
this makes it possible that a non-threaded application can use semaphore
without linking to thread library.
Old semaphore implementation is kept libc to maintain binary compatibility.
The kernel ksem API is no longer used in the new implemenation.
Discussed on: threads@
2010-01-05 02:37:59 +00:00
|
|
|
{
|
|
|
|
|
|
|
|
if (sem_check_validity(sem) != 0)
|
|
|
|
return (-1);
|
|
|
|
|
|
|
|
if (sem->_kern._flags & SEM_NAMED) {
|
|
|
|
errno = EINVAL;
|
|
|
|
return (-1);
|
|
|
|
}
|
|
|
|
sem->_magic = 0;
|
|
|
|
return (0);
|
|
|
|
}
|
|
|
|
|
|
|
|
int
|
2010-01-05 06:40:27 +00:00
|
|
|
_sem_getvalue(sem_t * __restrict sem, int * __restrict sval)
|
Use umtx to implement process sharable semaphore, to make this work,
now type sema_t is a structure which can be put in a shared memory area,
and multiple processes can operate it concurrently.
User can either use mmap(MAP_SHARED) + sem_init(pshared=1) or use sem_open()
to initialize a shared semaphore.
Named semaphore uses file system and is located in /tmp directory, and its
file name is prefixed with 'SEMD', so now it is chroot or jail friendly.
In simplist cases, both for named and un-named semaphore, userland code
does not have to enter kernel to reduce/increase semaphore's count.
The semaphore is designed to be crash-safe, it means even if an application
is crashed in the middle of operating semaphore, the semaphore state is
still safely recovered by later use, there is no waiter counter maintained
by userland code.
The main semaphore code is in libc and libthr only has some necessary stubs,
this makes it possible that a non-threaded application can use semaphore
without linking to thread library.
Old semaphore implementation is kept libc to maintain binary compatibility.
The kernel ksem API is no longer used in the new implemenation.
Discussed on: threads@
2010-01-05 02:37:59 +00:00
|
|
|
{
|
|
|
|
|
|
|
|
if (sem_check_validity(sem) != 0)
|
|
|
|
return (-1);
|
|
|
|
|
|
|
|
*sval = (int)sem->_kern._count;
|
|
|
|
return (0);
|
|
|
|
}
|
|
|
|
|
2010-01-05 03:39:31 +00:00
|
|
|
static __inline int
|
Use umtx to implement process sharable semaphore, to make this work,
now type sema_t is a structure which can be put in a shared memory area,
and multiple processes can operate it concurrently.
User can either use mmap(MAP_SHARED) + sem_init(pshared=1) or use sem_open()
to initialize a shared semaphore.
Named semaphore uses file system and is located in /tmp directory, and its
file name is prefixed with 'SEMD', so now it is chroot or jail friendly.
In simplist cases, both for named and un-named semaphore, userland code
does not have to enter kernel to reduce/increase semaphore's count.
The semaphore is designed to be crash-safe, it means even if an application
is crashed in the middle of operating semaphore, the semaphore state is
still safely recovered by later use, there is no waiter counter maintained
by userland code.
The main semaphore code is in libc and libthr only has some necessary stubs,
this makes it possible that a non-threaded application can use semaphore
without linking to thread library.
Old semaphore implementation is kept libc to maintain binary compatibility.
The kernel ksem API is no longer used in the new implemenation.
Discussed on: threads@
2010-01-05 02:37:59 +00:00
|
|
|
usem_wake(struct _usem *sem)
|
|
|
|
{
|
2012-01-16 06:15:14 +00:00
|
|
|
rmb();
|
Use umtx to implement process sharable semaphore, to make this work,
now type sema_t is a structure which can be put in a shared memory area,
and multiple processes can operate it concurrently.
User can either use mmap(MAP_SHARED) + sem_init(pshared=1) or use sem_open()
to initialize a shared semaphore.
Named semaphore uses file system and is located in /tmp directory, and its
file name is prefixed with 'SEMD', so now it is chroot or jail friendly.
In simplist cases, both for named and un-named semaphore, userland code
does not have to enter kernel to reduce/increase semaphore's count.
The semaphore is designed to be crash-safe, it means even if an application
is crashed in the middle of operating semaphore, the semaphore state is
still safely recovered by later use, there is no waiter counter maintained
by userland code.
The main semaphore code is in libc and libthr only has some necessary stubs,
this makes it possible that a non-threaded application can use semaphore
without linking to thread library.
Old semaphore implementation is kept libc to maintain binary compatibility.
The kernel ksem API is no longer used in the new implemenation.
Discussed on: threads@
2010-01-05 02:37:59 +00:00
|
|
|
if (!sem->_has_waiters)
|
|
|
|
return (0);
|
|
|
|
return _umtx_op(sem, UMTX_OP_SEM_WAKE, 0, NULL, NULL);
|
|
|
|
}
|
|
|
|
|
2010-01-05 03:39:31 +00:00
|
|
|
static __inline int
|
Use umtx to implement process sharable semaphore, to make this work,
now type sema_t is a structure which can be put in a shared memory area,
and multiple processes can operate it concurrently.
User can either use mmap(MAP_SHARED) + sem_init(pshared=1) or use sem_open()
to initialize a shared semaphore.
Named semaphore uses file system and is located in /tmp directory, and its
file name is prefixed with 'SEMD', so now it is chroot or jail friendly.
In simplist cases, both for named and un-named semaphore, userland code
does not have to enter kernel to reduce/increase semaphore's count.
The semaphore is designed to be crash-safe, it means even if an application
is crashed in the middle of operating semaphore, the semaphore state is
still safely recovered by later use, there is no waiter counter maintained
by userland code.
The main semaphore code is in libc and libthr only has some necessary stubs,
this makes it possible that a non-threaded application can use semaphore
without linking to thread library.
Old semaphore implementation is kept libc to maintain binary compatibility.
The kernel ksem API is no longer used in the new implemenation.
Discussed on: threads@
2010-01-05 02:37:59 +00:00
|
|
|
usem_wait(struct _usem *sem, const struct timespec *timeout)
|
|
|
|
{
|
|
|
|
if (timeout && (timeout->tv_sec < 0 || (timeout->tv_sec == 0 &&
|
|
|
|
timeout->tv_nsec <= 0))) {
|
|
|
|
errno = ETIMEDOUT;
|
|
|
|
return (-1);
|
|
|
|
}
|
|
|
|
return _umtx_op(sem, UMTX_OP_SEM_WAIT, 0, NULL,
|
|
|
|
__DECONST(void*, timeout));
|
|
|
|
}
|
|
|
|
|
|
|
|
int
|
2010-01-05 06:40:27 +00:00
|
|
|
_sem_trywait(sem_t *sem)
|
Use umtx to implement process sharable semaphore, to make this work,
now type sema_t is a structure which can be put in a shared memory area,
and multiple processes can operate it concurrently.
User can either use mmap(MAP_SHARED) + sem_init(pshared=1) or use sem_open()
to initialize a shared semaphore.
Named semaphore uses file system and is located in /tmp directory, and its
file name is prefixed with 'SEMD', so now it is chroot or jail friendly.
In simplist cases, both for named and un-named semaphore, userland code
does not have to enter kernel to reduce/increase semaphore's count.
The semaphore is designed to be crash-safe, it means even if an application
is crashed in the middle of operating semaphore, the semaphore state is
still safely recovered by later use, there is no waiter counter maintained
by userland code.
The main semaphore code is in libc and libthr only has some necessary stubs,
this makes it possible that a non-threaded application can use semaphore
without linking to thread library.
Old semaphore implementation is kept libc to maintain binary compatibility.
The kernel ksem API is no longer used in the new implemenation.
Discussed on: threads@
2010-01-05 02:37:59 +00:00
|
|
|
{
|
|
|
|
int val;
|
|
|
|
|
|
|
|
if (sem_check_validity(sem) != 0)
|
|
|
|
return (-1);
|
|
|
|
|
|
|
|
while ((val = sem->_kern._count) > 0) {
|
|
|
|
if (atomic_cmpset_acq_int(&sem->_kern._count, val, val - 1))
|
|
|
|
return (0);
|
|
|
|
}
|
|
|
|
errno = EAGAIN;
|
|
|
|
return (-1);
|
|
|
|
}
|
|
|
|
|
|
|
|
#define TIMESPEC_SUB(dst, src, val) \
|
|
|
|
do { \
|
|
|
|
(dst)->tv_sec = (src)->tv_sec - (val)->tv_sec; \
|
|
|
|
(dst)->tv_nsec = (src)->tv_nsec - (val)->tv_nsec; \
|
|
|
|
if ((dst)->tv_nsec < 0) { \
|
|
|
|
(dst)->tv_sec--; \
|
|
|
|
(dst)->tv_nsec += 1000000000; \
|
|
|
|
} \
|
|
|
|
} while (0)
|
|
|
|
|
|
|
|
|
|
|
|
int
|
2010-01-05 06:40:27 +00:00
|
|
|
_sem_timedwait(sem_t * __restrict sem,
|
Use umtx to implement process sharable semaphore, to make this work,
now type sema_t is a structure which can be put in a shared memory area,
and multiple processes can operate it concurrently.
User can either use mmap(MAP_SHARED) + sem_init(pshared=1) or use sem_open()
to initialize a shared semaphore.
Named semaphore uses file system and is located in /tmp directory, and its
file name is prefixed with 'SEMD', so now it is chroot or jail friendly.
In simplist cases, both for named and un-named semaphore, userland code
does not have to enter kernel to reduce/increase semaphore's count.
The semaphore is designed to be crash-safe, it means even if an application
is crashed in the middle of operating semaphore, the semaphore state is
still safely recovered by later use, there is no waiter counter maintained
by userland code.
The main semaphore code is in libc and libthr only has some necessary stubs,
this makes it possible that a non-threaded application can use semaphore
without linking to thread library.
Old semaphore implementation is kept libc to maintain binary compatibility.
The kernel ksem API is no longer used in the new implemenation.
Discussed on: threads@
2010-01-05 02:37:59 +00:00
|
|
|
const struct timespec * __restrict abstime)
|
|
|
|
{
|
|
|
|
struct timespec ts, ts2;
|
2010-09-25 01:57:47 +00:00
|
|
|
int val, retval;
|
Use umtx to implement process sharable semaphore, to make this work,
now type sema_t is a structure which can be put in a shared memory area,
and multiple processes can operate it concurrently.
User can either use mmap(MAP_SHARED) + sem_init(pshared=1) or use sem_open()
to initialize a shared semaphore.
Named semaphore uses file system and is located in /tmp directory, and its
file name is prefixed with 'SEMD', so now it is chroot or jail friendly.
In simplist cases, both for named and un-named semaphore, userland code
does not have to enter kernel to reduce/increase semaphore's count.
The semaphore is designed to be crash-safe, it means even if an application
is crashed in the middle of operating semaphore, the semaphore state is
still safely recovered by later use, there is no waiter counter maintained
by userland code.
The main semaphore code is in libc and libthr only has some necessary stubs,
this makes it possible that a non-threaded application can use semaphore
without linking to thread library.
Old semaphore implementation is kept libc to maintain binary compatibility.
The kernel ksem API is no longer used in the new implemenation.
Discussed on: threads@
2010-01-05 02:37:59 +00:00
|
|
|
|
|
|
|
if (sem_check_validity(sem) != 0)
|
|
|
|
return (-1);
|
|
|
|
|
|
|
|
retval = 0;
|
|
|
|
for (;;) {
|
|
|
|
while ((val = sem->_kern._count) > 0) {
|
|
|
|
if (atomic_cmpset_acq_int(&sem->_kern._count, val, val - 1))
|
|
|
|
return (0);
|
|
|
|
}
|
|
|
|
|
2010-09-25 01:57:47 +00:00
|
|
|
if (retval) {
|
|
|
|
_pthread_testcancel();
|
Use umtx to implement process sharable semaphore, to make this work,
now type sema_t is a structure which can be put in a shared memory area,
and multiple processes can operate it concurrently.
User can either use mmap(MAP_SHARED) + sem_init(pshared=1) or use sem_open()
to initialize a shared semaphore.
Named semaphore uses file system and is located in /tmp directory, and its
file name is prefixed with 'SEMD', so now it is chroot or jail friendly.
In simplist cases, both for named and un-named semaphore, userland code
does not have to enter kernel to reduce/increase semaphore's count.
The semaphore is designed to be crash-safe, it means even if an application
is crashed in the middle of operating semaphore, the semaphore state is
still safely recovered by later use, there is no waiter counter maintained
by userland code.
The main semaphore code is in libc and libthr only has some necessary stubs,
this makes it possible that a non-threaded application can use semaphore
without linking to thread library.
Old semaphore implementation is kept libc to maintain binary compatibility.
The kernel ksem API is no longer used in the new implemenation.
Discussed on: threads@
2010-01-05 02:37:59 +00:00
|
|
|
break;
|
2010-09-25 01:57:47 +00:00
|
|
|
}
|
Use umtx to implement process sharable semaphore, to make this work,
now type sema_t is a structure which can be put in a shared memory area,
and multiple processes can operate it concurrently.
User can either use mmap(MAP_SHARED) + sem_init(pshared=1) or use sem_open()
to initialize a shared semaphore.
Named semaphore uses file system and is located in /tmp directory, and its
file name is prefixed with 'SEMD', so now it is chroot or jail friendly.
In simplist cases, both for named and un-named semaphore, userland code
does not have to enter kernel to reduce/increase semaphore's count.
The semaphore is designed to be crash-safe, it means even if an application
is crashed in the middle of operating semaphore, the semaphore state is
still safely recovered by later use, there is no waiter counter maintained
by userland code.
The main semaphore code is in libc and libthr only has some necessary stubs,
this makes it possible that a non-threaded application can use semaphore
without linking to thread library.
Old semaphore implementation is kept libc to maintain binary compatibility.
The kernel ksem API is no longer used in the new implemenation.
Discussed on: threads@
2010-01-05 02:37:59 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* The timeout argument is only supposed to
|
|
|
|
* be checked if the thread would have blocked.
|
|
|
|
*/
|
|
|
|
if (abstime != NULL) {
|
|
|
|
if (abstime->tv_nsec >= 1000000000 || abstime->tv_nsec < 0) {
|
|
|
|
errno = EINVAL;
|
|
|
|
return (-1);
|
|
|
|
}
|
|
|
|
clock_gettime(CLOCK_REALTIME, &ts);
|
|
|
|
TIMESPEC_SUB(&ts2, abstime, &ts);
|
|
|
|
}
|
2010-09-25 01:57:47 +00:00
|
|
|
_pthread_cancel_enter(1);
|
Use umtx to implement process sharable semaphore, to make this work,
now type sema_t is a structure which can be put in a shared memory area,
and multiple processes can operate it concurrently.
User can either use mmap(MAP_SHARED) + sem_init(pshared=1) or use sem_open()
to initialize a shared semaphore.
Named semaphore uses file system and is located in /tmp directory, and its
file name is prefixed with 'SEMD', so now it is chroot or jail friendly.
In simplist cases, both for named and un-named semaphore, userland code
does not have to enter kernel to reduce/increase semaphore's count.
The semaphore is designed to be crash-safe, it means even if an application
is crashed in the middle of operating semaphore, the semaphore state is
still safely recovered by later use, there is no waiter counter maintained
by userland code.
The main semaphore code is in libc and libthr only has some necessary stubs,
this makes it possible that a non-threaded application can use semaphore
without linking to thread library.
Old semaphore implementation is kept libc to maintain binary compatibility.
The kernel ksem API is no longer used in the new implemenation.
Discussed on: threads@
2010-01-05 02:37:59 +00:00
|
|
|
retval = usem_wait(&sem->_kern, abstime ? &ts2 : NULL);
|
2010-09-25 01:57:47 +00:00
|
|
|
_pthread_cancel_leave(0);
|
Use umtx to implement process sharable semaphore, to make this work,
now type sema_t is a structure which can be put in a shared memory area,
and multiple processes can operate it concurrently.
User can either use mmap(MAP_SHARED) + sem_init(pshared=1) or use sem_open()
to initialize a shared semaphore.
Named semaphore uses file system and is located in /tmp directory, and its
file name is prefixed with 'SEMD', so now it is chroot or jail friendly.
In simplist cases, both for named and un-named semaphore, userland code
does not have to enter kernel to reduce/increase semaphore's count.
The semaphore is designed to be crash-safe, it means even if an application
is crashed in the middle of operating semaphore, the semaphore state is
still safely recovered by later use, there is no waiter counter maintained
by userland code.
The main semaphore code is in libc and libthr only has some necessary stubs,
this makes it possible that a non-threaded application can use semaphore
without linking to thread library.
Old semaphore implementation is kept libc to maintain binary compatibility.
The kernel ksem API is no longer used in the new implemenation.
Discussed on: threads@
2010-01-05 02:37:59 +00:00
|
|
|
}
|
|
|
|
return (retval);
|
|
|
|
}
|
|
|
|
|
|
|
|
int
|
2010-01-05 06:40:27 +00:00
|
|
|
_sem_wait(sem_t *sem)
|
Use umtx to implement process sharable semaphore, to make this work,
now type sema_t is a structure which can be put in a shared memory area,
and multiple processes can operate it concurrently.
User can either use mmap(MAP_SHARED) + sem_init(pshared=1) or use sem_open()
to initialize a shared semaphore.
Named semaphore uses file system and is located in /tmp directory, and its
file name is prefixed with 'SEMD', so now it is chroot or jail friendly.
In simplist cases, both for named and un-named semaphore, userland code
does not have to enter kernel to reduce/increase semaphore's count.
The semaphore is designed to be crash-safe, it means even if an application
is crashed in the middle of operating semaphore, the semaphore state is
still safely recovered by later use, there is no waiter counter maintained
by userland code.
The main semaphore code is in libc and libthr only has some necessary stubs,
this makes it possible that a non-threaded application can use semaphore
without linking to thread library.
Old semaphore implementation is kept libc to maintain binary compatibility.
The kernel ksem API is no longer used in the new implemenation.
Discussed on: threads@
2010-01-05 02:37:59 +00:00
|
|
|
{
|
2010-01-05 06:40:27 +00:00
|
|
|
return _sem_timedwait(sem, NULL);
|
Use umtx to implement process sharable semaphore, to make this work,
now type sema_t is a structure which can be put in a shared memory area,
and multiple processes can operate it concurrently.
User can either use mmap(MAP_SHARED) + sem_init(pshared=1) or use sem_open()
to initialize a shared semaphore.
Named semaphore uses file system and is located in /tmp directory, and its
file name is prefixed with 'SEMD', so now it is chroot or jail friendly.
In simplist cases, both for named and un-named semaphore, userland code
does not have to enter kernel to reduce/increase semaphore's count.
The semaphore is designed to be crash-safe, it means even if an application
is crashed in the middle of operating semaphore, the semaphore state is
still safely recovered by later use, there is no waiter counter maintained
by userland code.
The main semaphore code is in libc and libthr only has some necessary stubs,
this makes it possible that a non-threaded application can use semaphore
without linking to thread library.
Old semaphore implementation is kept libc to maintain binary compatibility.
The kernel ksem API is no longer used in the new implemenation.
Discussed on: threads@
2010-01-05 02:37:59 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* POSIX:
|
|
|
|
* The sem_post() interface is reentrant with respect to signals and may be
|
|
|
|
* invoked from a signal-catching function.
|
|
|
|
* The implementation does not use lock, so it should be safe.
|
|
|
|
*/
|
|
|
|
int
|
2010-01-05 06:40:27 +00:00
|
|
|
_sem_post(sem_t *sem)
|
Use umtx to implement process sharable semaphore, to make this work,
now type sema_t is a structure which can be put in a shared memory area,
and multiple processes can operate it concurrently.
User can either use mmap(MAP_SHARED) + sem_init(pshared=1) or use sem_open()
to initialize a shared semaphore.
Named semaphore uses file system and is located in /tmp directory, and its
file name is prefixed with 'SEMD', so now it is chroot or jail friendly.
In simplist cases, both for named and un-named semaphore, userland code
does not have to enter kernel to reduce/increase semaphore's count.
The semaphore is designed to be crash-safe, it means even if an application
is crashed in the middle of operating semaphore, the semaphore state is
still safely recovered by later use, there is no waiter counter maintained
by userland code.
The main semaphore code is in libc and libthr only has some necessary stubs,
this makes it possible that a non-threaded application can use semaphore
without linking to thread library.
Old semaphore implementation is kept libc to maintain binary compatibility.
The kernel ksem API is no longer used in the new implemenation.
Discussed on: threads@
2010-01-05 02:37:59 +00:00
|
|
|
{
|
|
|
|
|
|
|
|
if (sem_check_validity(sem) != 0)
|
|
|
|
return (-1);
|
|
|
|
|
|
|
|
atomic_add_rel_int(&sem->_kern._count, 1);
|
2010-01-05 03:39:31 +00:00
|
|
|
return usem_wake(&sem->_kern);
|
Use umtx to implement process sharable semaphore, to make this work,
now type sema_t is a structure which can be put in a shared memory area,
and multiple processes can operate it concurrently.
User can either use mmap(MAP_SHARED) + sem_init(pshared=1) or use sem_open()
to initialize a shared semaphore.
Named semaphore uses file system and is located in /tmp directory, and its
file name is prefixed with 'SEMD', so now it is chroot or jail friendly.
In simplist cases, both for named and un-named semaphore, userland code
does not have to enter kernel to reduce/increase semaphore's count.
The semaphore is designed to be crash-safe, it means even if an application
is crashed in the middle of operating semaphore, the semaphore state is
still safely recovered by later use, there is no waiter counter maintained
by userland code.
The main semaphore code is in libc and libthr only has some necessary stubs,
this makes it possible that a non-threaded application can use semaphore
without linking to thread library.
Old semaphore implementation is kept libc to maintain binary compatibility.
The kernel ksem API is no longer used in the new implemenation.
Discussed on: threads@
2010-01-05 02:37:59 +00:00
|
|
|
}
|