Sun Microsystems, Inc. gratefully acknowledges The Open Group for
permission to reproduce portions of its copyrighted documentation.
Original documentation from The Open Group can be obtained online at
http://www.opengroup.org/bookstore/.

The Institute of Electrical and Electronics Engineers and The Open
Group, have given us permission to reprint portions of their
documentation.

In the following statement, the phrase ``this text'' refers to portions
of the system documentation.

Portions of this text are reprinted and reproduced in electronic form
in the SunOS Reference Manual, from IEEE Std 1003.1, 2004 Edition,
Standard for Information Technology -- Portable Operating System
Interface (POSIX), The Open Group Base Specifications Issue 6,
Copyright (C) 2001-2004 by the Institute of Electrical and Electronics
Engineers, Inc and The Open Group. In the event of any discrepancy
between these versions and the original IEEE and The Open Group
Standard, the original IEEE and The Open Group Standard is the referee
document. The original Standard can be obtained online at
http://www.opengroup.org/unix/online.html.

This notice shall appear on any product containing this material.

The contents of this file are subject to the terms of the
Common Development and Distribution License (the "License").
You may not use this file except in compliance with the License.

You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
or http://www.opensolaris.org/os/licensing.
See the License for the specific language governing permissions
and limitations under the License.

When distributing Covered Code, include this CDDL HEADER in each
file and include the License file at usr/src/OPENSOLARIS.LICENSE.
If applicable, add the following below this CDDL HEADER, with the
fields enclosed by brackets "[]" replaced with your own identifying
information: Portions Copyright [yyyy] [name of copyright owner]


Portions Copyright (c) 1995 IEEE. All Rights Reserved.
Copyright (c) 2001, The IEEE and The Open Group. All Rights Reserved.
Copyright (c) 2007, Sun Microsystems, Inc. All Rights Reserved.
Copyright 2014 Nexenta Systems, Inc. All rights reserved.

mutex_init, mutex_lock, mutex_trylock, mutex_unlock, mutex_consistent, mutex_destroy - mutual exclusion locks

cc -mt [ flag... ] file... [ library... ]
#include <thread.h>
#include <synch.h>

int mutex_init(mutex_t *mp, int type, void *arg);

int mutex_lock(mutex_t *mp);

int mutex_trylock(mutex_t *mp);

int mutex_unlock(mutex_t *mp);

int mutex_consistent(mutex_t *mp);

int mutex_destroy(mutex_t *mp);

Mutual exclusion locks (mutexes) prevent multiple threads from simultaneously executing critical sections of code that access shared data (that is, mutexes are used to serialize the execution of threads). All mutexes must be global. A successful call for a mutex lock by way of mutex_lock() will cause another thread that is also trying to lock the same mutex to block until the owner thread unlocks it by way of mutex_unlock(). Threads within the same process or within other processes can share mutexes.

Mutexes can synchronize threads within the same process or in other processes. Mutexes can be used to synchronize threads between processes if the mutexes are allocated in writable memory and shared among the cooperating processes (see mmap(2)), and have been initialized for this task.

Mutexes are either intra-process or inter-process, depending upon the argument passed implicitly or explicitly to the initialization of that mutex. A statically allocated mutex does not need to be explicitly initialized; by default, a statically allocated mutex is initialized with all zeros and its scope is set to be within the calling process.

For inter-process synchronization, a mutex needs to be allocated in memory shared between these processes. Since the memory for such a mutex must be allocated dynamically, the mutex needs to be explicitly initialized using mutex_init().

The mutex_init() function initializes the mutex referenced by mp with the type specified by type. Upon successful initialization the state of the mutex becomes initialized and unlocked. Only the attribute type LOCK_PRIO_PROTECT uses arg. The type argument must be one of the following: USYNC_THREAD

USYNC_PROCESS

The type argument can be augmented by the bitwise-inclusive-OR of zero or more of the following flags: LOCK_ROBUST

LOCK_RECURSIVE LOCK_ERRORCHECK LOCK_PRIO_INHERIT LOCK_PRIO_PROTECT

See pthread_mutexattr_getrobust(3C) for more information about robust mutexes. The LOCK_ROBUST attribute is the same as the POSIX PTHREAD_MUTEX_ROBUST attribute.

See pthread_mutexattr_settype(3C) for more information on recursive and error checking mutex types. The combination (LOCK_RECURSIVE | LOCK_ERRORCHECK) is the same as the POSIX PTHREAD_MUTEX_RECURSIVE type. By itself, LOCK_ERRORCHECK is the same as the POSIX PTHREAD_MUTEX_ERRORCHECK type.

The LOCK_PRIO_INHERIT attribute is the same as the POSIX PTHREAD_PRIO_INHERIT attribute. The LOCK_PRIO_PROTECT attribute is the same as the POSIX PTHREAD_PRIO_PROTECT attribute. See pthread_mutexattr_getprotocol(3C), pthread_mutexattr_getprioceiling(3C), and pthread_mutex_getprioceiling(3C) for a full discussion. The LOCK_PRIO_INHERIT and LOCK_PRIO_PROTECT attributes are mutually exclusive. Specifying both of these attributes causes mutex_init() to fail with EINVAL.

Initializing mutexes can also be accomplished by allocating in zeroed memory (default), in which case a type of USYNC_THREAD is assumed. In general, the following rules apply to mutex initialization:

These rules do not apply to LOCK_ROBUST mutexes. See the description for LOCK_ROBUST above. If default mutex attributes are used, the macro DEFAULTMUTEX can be used to initialize mutexes that are statically allocated.

Default mutex initialization (intra-process):

mutex_t mp;
mutex_init(&mp, USYNC_THREAD, NULL);

or

mutex_t mp = DEFAULTMUTEX;

Customized mutex initialization (inter-process):

mutex_init(&mp, USYNC_PROCESS, NULL);

Customized mutex initialization (inter-process robust):

mutex_init(&mp, USYNC_PROCESS | LOCK_ROBUST, NULL);

Statically allocated mutexes can also be initialized with macros specifying LOCK_RECURSIVE and/or LOCK_ERRORCHECK: mutex_t mp = RECURSIVEMUTEX;

mutex_t mp = ERRORCHECKMUTEX; mutex_t mp = RECURSIVE_ERRORCHECKMUTEX;

A critical section of code is enclosed by a the call to lock the mutex and the call to unlock the mutex to protect it from simultaneous access by multiple threads. Only one thread at a time may possess mutually exclusive access to the critical section of code that is enclosed by the mutex-locking call and the mutex-unlocking call, whether the mutex's scope is intra-process or inter-process. A thread calling to lock the mutex either gets exclusive access to the code starting from the successful locking until its call to unlock the mutex, or it waits until the mutex is unlocked by the thread that locked it.

Mutexes have ownership, unlike semaphores. Although any thread, within the scope of a mutex, can get an unlocked mutex and lock access to the same critical section of code, only the thread that locked a mutex should unlock it.

If a thread waiting for a mutex receives a signal, upon return from the signal handler, the thread resumes waiting for the mutex as if there was no interrupt. A mutex protects code, not data; therefore, strongly bind a mutex with the data by putting both within the same structure, or at least within the same procedure.

A call to mutex_lock() locks the mutex object referenced by mp. If the mutex is already locked, the calling thread blocks until the mutex is freed; this will return with the mutex object referenced by mp in the locked state with the calling thread as its owner. If the current owner of a mutex tries to relock the mutex, it will result in deadlock.

The mutex_trylock() function is the same as mutex_lock(), respectively, except that if the mutex object referenced by mp is locked (by any thread, including the current thread), the call returns immediately with an error.

The mutex_unlock() function are called by the owner of the mutex object referenced by mp to release it. The mutex must be locked and the calling thread must be the one that last locked the mutex (the owner). If there are threads blocked on the mutex object referenced by mp when mutex_unlock() is called, the mp is freed, and the scheduling policy will determine which thread gets the mutex. If the calling thread is not the owner of the lock, no error status is returned, and the behavior of the program is undefined.

The mutex_destroy() function destroys the mutex object referenced by mp. The mutex object becomes uninitialized. The space used by the destroyed mutex variable is not freed. It needs to be explicitly reclaimed.

If successful, these functions return 0. Otherwise, an error number is returned.

The mutex_init() function will fail if: EINVAL

The mutex_init() function will fail for LOCK_ROBUST type mutex if: EBUSY

EINVAL

The mutex_trylock() function will fail if: EBUSY

The mutex_lock() and mutex_trylock() functions will fail for a LOCK_RECURSIVE mutex if: EAGAIN

The mutex_lock() function will fail for a LOCK_ERRORCHECK and non-LOCK_RECURSIVE mutex if: EDEADLK

The mutex_lock() function may fail for a non-LOCK_ERRORCHECK and non-LOCK_RECURSIVE mutex if: EDEADLK

The mutex_unlock() function will fail for a LOCK_ERRORCHECK mutex if: EPERM

The mutex_lock() or mutex_trylock() functions will fail for LOCK_ROBUST type mutex if: EOWNERDEAD

ENOTRECOVERABLE

The mutex_consistent() function will fail if: EINVAL

The following example uses one global mutex as a gate-keeper to permit each thread exclusive sequential access to the code within the user-defined function "change_global_data." This type of synchronization will protect the state of shared data, but it also prohibits parallelism.

/* cc thisfile.c -lthread */
#define _REENTRANT
#include <stdio.h>
#include <thread.h>
#define NUM_THREADS 12
void *change_global_data(void *); /* for thr_create() */
main(int argc,char * argv[]) {
 int i=0;
 for (i=0; i< NUM_THREADS; i++) {
 thr_create(NULL, 0, change_global_data, NULL, 0, NULL);
 }
 while ((thr_join(NULL, NULL, NULL) == 0));
}

void * change_global_data(void *null){
 static mutex_t Global_mutex;
 static int Global_data = 0;
 mutex_lock(&Global_mutex);
 Global_data++;
 sleep(1);
 printf("%d is global data\en",Global_data);
 mutex_unlock(&Global_mutex);
 return NULL;
}

The previous example, the mutex, the code it owns, and the data it protects was enclosed in one function. The next example uses C++ features to accommodate many functions that use just one mutex to protect one data:

/* CC thisfile.c -lthread use C++ to compile*/

#define _REENTRANT
#include <stdlib.h>
#include <stdio.h>
#include <thread.h>
#include <errno.h>
#include <iostream.h>
#define NUM_THREADS 16
void *change_global_data(void *); /* for thr_create() */

class Mutected {
 private:
 static mutex_t Global_mutex;
 static int Global_data;
 public:
 static int add_to_global_data(void);
 static int subtract_from_global_data(void);
};

int Mutected::Global_data = 0;
mutex_t Mutected::Global_mutex;

int Mutected::add_to_global_data() {
 mutex_lock(&Global_mutex); 
 Global_data++;
 mutex_unlock(&Global_mutex); 
 return Global_data;
}

int Mutected::subtract_from_global_data() {
 mutex_lock(&Global_mutex); 
 Global_data--;
 mutex_unlock(&Global_mutex);
 return Global_data;
}

void
main(int argc,char * argv[]) {
 int i=0;
 for (i=0;i< NUM_THREADS;i++) {
 thr_create(NULL,0,change_global_data,NULL,0,NULL);
 }
 while ((thr_join(NULL,NULL,NULL) == 0));
}

void * change_global_data(void *) {
 static int switcher = 0;
 if ((switcher++ % 3) == 0) /* one-in-three threads subtracts */
 cout << Mutected::subtract_from_global_data() << endl; 
 else
 cout << Mutected::add_to_global_data() << endl; 
 return NULL;
}

A mutex can protect data that is shared among processes. The mutex would need to be initialized as USYNC_PROCESS. One process initializes the process-shared mutex and writes it to a file to be mapped into memory by all cooperating processes (see mmap(2)). Afterwards, other independent processes can run the same program (whether concurrently or not) and share mutex-protected data.

/* cc thisfile.c -lthread */
/* To execute, run the command line "a.out 0 &; a.out 1" */

#define _REENTRANT
#include <sys/types.h>
#include <sys/mman.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <stdio.h>
#include <thread.h>
#define INTERPROCESS_FILE "ipc-sharedfile"
#define NUM_ADDTHREADS 12
#define NUM_SUBTRACTTHREADS 10
#define INCREMENT '0'
#define DECREMENT '1'
typedef struct {
 mutex_t Interprocess_mutex;
 int Interprocess_data;
} buffer_t;
buffer_t *buffer;

void *add_interprocess_data(), *subtract_interprocess_data();
void create_shared_memory(), test_argv();
int zeroed[sizeof(buffer_t)];
int ipc_fd, i=0;

void
main(int argc,char * argv[]){
 test_argv(argv[1]);

 switch (*argv[1]) {
 case INCREMENT:
 /* Initializes the process-shared mutex */
 /* Should be run prior to running a DECREMENT process */
 create_shared_memory();
 ipc_fd = open(INTERPROCESS_FILE, O_RDWR);
 buffer = (buffer_t *)mmap(NULL, sizeof(buffer_t),
 PROT_READ | PROT_WRITE, MAP_SHARED, ipc_fd, 0);
 buffer->Interprocess_data = 0;
 mutex_init(&buffer->Interprocess_mutex, USYNC_PROCESS,0);
 for (i=0; i< NUM_ADDTHREADS; i++)
 thr_create(NULL, 0, add_interprocess_data, argv[1],
 0, NULL);
 break;

 case DECREMENT:
 /* Should be run after the INCREMENT process has run. */
 while(ipc_fd = open(INTERPROCESS_FILE, O_RDWR)) == -1)
 sleep(1);
 buffer = (buffer_t *)mmap(NULL, sizeof(buffer_t),
 PROT_READ | PROT_WRITE, MAP_SHARED, ipc_fd, 0);
 for (i=0; i< NUM_SUBTRACTTHREADS; i++)
 thr_create(NULL, 0, subtract_interprocess_data, argv[1],
 0, NULL);
 break;
 } /* end switch */

 while ((thr_join(NULL,NULL,NULL) == 0));
} /* end main */

void *add_interprocess_data(char argv_1[]){
 mutex_lock(&buffer->Interprocess_mutex);
 buffer->Interprocess_data++;
 sleep(2);
 printf("%d is add-interprocess data, and %c is argv1\en",
 buffer->Interprocess_data, argv_1[0]);
 mutex_unlock(&buffer->Interprocess_mutex);
 return NULL;
}

void *subtract_interprocess_data(char argv_1[]) {
 mutex_lock(&buffer->Interprocess_mutex);
 buffer->Interprocess_data--;
 sleep(2);
 printf("%d is subtract-interprocess data, and %c is argv1\en",
 buffer->Interprocess_data, argv_1[0]);
 mutex_unlock(&buffer->Interprocess_mutex);
 return NULL;
}

void create_shared_memory(){
 int i;
 ipc_fd = creat(INTERPROCESS_FILE, O_CREAT | O_RDWR );
 for (i=0; i<sizeof(buffer_t); i++){
 zeroed[i] = 0;
 write(ipc_fd, &zeroed[i],2);
 }
 close(ipc_fd);
 chmod(INTERPROCESS_FILE, S_IRWXU | S_IRWXG | S_IRWXO);
}

void test_argv(char argv1[]) {
 if (argv1 == NULL) {
 printf("use 0 as arg1 for initial process\en \e
 or use 1 as arg1 for the second process\en");
 exit(NULL);
 }
}

A mutex can protect data that is shared among processes robustly. The mutex would need to be initialized as USYNC_PROCESS | LOCK_ROBUST. One process initializes the robust process-shared mutex and writes it to a file to be mapped into memory by all cooperating processes (see mmap(2)). Afterwards, other independent processes can run the same program (whether concurrently or not) and share mutex-protected data.

The following example shows how to use a USYNC_PROCESS | LOCK_ROBUST type mutex.

/* cc thisfile.c -lthread */
 /* To execute, run the command line "a.out & a.out 1" */
 #include <sys/types.h>
 #include <sys/mman.h>
 #include <fcntl.h>
 #include <stdio.h>
 #include <thread.h>
 #define INTERPROCESS_FILE "ipc-sharedfile"
 typedef struct {
 mutex_t Interprocess_mutex;
 int Interprocess_data;
 } buffer_t;
 buffer_t *buffer;
 int make_date_consistent();
 void create_shared_memory();
 int zeroed[sizeof(buffer_t)];
 int ipc_fd, i=0;
 main(int argc,char * argv[]) {
 int rc;
 if (argc > 1) {
 while((ipc_fd = open(INTERPROCESS_FILE, O_RDWR)) == -1)
 sleep(1);
 buffer = (buffer_t *)mmap(NULL, sizeof(buffer_t),
 PROT_READ | PROT_WRITE, MAP_SHARED, ipc_fd, 0);
 mutex_init(&buffer->Interprocess_mutex,
 USYNC_PROCESS | LOCK_ROBUST, 0);
 } else {
 create_shared_memory();
 ipc_fd = open(INTERPROCESS_FILE, O_RDWR);
 buffer = (buffer_t *)mmap(NULL, sizeof(buffer_t),
 PROT_READ | PROT_WRITE, MAP_SHARED, ipc_fd, 0);
 buffer->Interprocess_data = 0;
 mutex_init(&buffer->Interprocess_mutex,
 USYNC_PROCESS | LOCK_ROBUST, 0);
 }
 for(;;) {
 rc = mutex_lock(&buffer->Interprocess_mutex);
 switch (rc) {
 case EOWNERDEAD:
 /*
 * The lock is acquired.
 * The last owner died holding the lock.
 * Try to make the state associated with
 * the mutex consistent.
 * If successful, make the robust lock consistent.
 */
 if (make_data_consistent())
 mutex_consistent(&buffer->Interprocess_mutex);
 mutex_unlock(&buffer->Interprocess_mutex);
 break;
 case ENOTRECOVERABLE:
 /*
 * The lock is not acquired.
 * The last owner got the mutex with EOWNERDEAD
 * and failed to make the data consistent.
 * There is no way to recover, so just exit.
 */
 exit(1);
 case 0:
 /*
 * There is no error - data is consistent.
 * Do something with data.
 */
 mutex_unlock(&buffer->Interprocess_mutex);
 break;
 }
 }
} /* end main */
void create_shared_memory() {
 int i;
 ipc_fd = creat(INTERPROCESS_FILE, O_CREAT | O_RDWR );
 for (i=0; i<sizeof(buffer_t); i++) {
 zeroed[i] = 0;
 write(ipc_fd, &zeroed[i],2);
 }
 close(ipc_fd);
 chmod(INTERPROCESS_FILE, S_IRWXU | S_IRWXG | S_IRWXO);
 }

 /* return 1 if able to make data consistent, otherwise 0. */
 int make_data_consistent () {
 buffer->Interprocess_data = 0;
 return (1);
 }

The following example allocates and frees memory in which a mutex is embedded.

struct record {
 int field1;
 int field2;
 mutex_t m;
} *r;
r = malloc(sizeof(struct record));
mutex_init(&r->m, USYNC_THREAD, NULL);
/*
 * The fields in this record are accessed concurrently
 * by acquiring the embedded lock.
 */

The thread execution in this example is as follows:

Thread 1 executes: Thread 2 executes:

... ...
mutex_lock(&r->m); mutex_lock(&r->m);
r->field1++; localvar = r->field1;
mutex_unlock(&r->m); mutex_unlock(&r->m);
... ...

Later, when a thread decides to free the memory pointed to by r, the thread should call mutex_destroy(\|) on the mutexes in this memory.

In the following example, the main thread can do a thr_join(\|) on both of the above threads. If there are no other threads using the memory in r, the main thread can now safely free r:

for (i = 0; i < 2; i++)
 thr_join(0, 0, 0);
mutex_destroy(&r->m); /* first destroy mutex */
free(r); /* then free memory */

If the mutex is not destroyed, the program could have memory leaks.

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

ATTRIBUTE TYPE ATTRIBUTE VALUE

Interface Stability Stable

MT-Level MT-Safe

mmap(2), shmop(2), pthread_mutexattr_getprioceiling(3C), pthread_mutexattr_getprotocol(3C), pthread_mutexattr_getrobust(3C), pthread_mutexattr_gettype(3C), pthread_mutex_getprioceiling(3C), pthread_mutex_init(3C), attributes(5), mutex(5), standards(5)

Previous releases of Solaris provided the USYNC_PROCESS_ROBUST mutex type. This type is now deprecated but is still supported for source and binary compatibility. When passed to mutex_init(), it is transformed into (USYNC_PROCESS | LOCK_ROBUST). The former method for restoring a USYNC_PROCESS_ROBUST mutex to a consistent state was to reinitialize it by calling mutex_init(). This method is still supported for source and binary compatibility, but the proper method is to call mutex_consistent().

The USYNC_PROCESS_ROBUST type permitted an alternate error value, ELOCKUNMAPPED, to be returned by mutex_lock() if the process containing a locked robust mutex unmapped the memory containing the mutex or performed one of the exec(2) functions. The ELOCKUNMAPPED error value implies all of the consequences of the EOWNERDEAD error value and as such is just a synonym for EOWNERDEAD. For full source and binary compatibility, the ELOCKUNMAPPED error value is still returned from mutex_lock() in these circumstances, but only if the mutex was initialized with the USYNC_PROCESS_ROBUST type. Otherwise, EOWNERDEAD is returned in these circumstances.

The mutex_lock(), mutex_unlock(), and mutex_trylock() functions do not validate the mutex type. An uninitialized mutex or a mutex with an invalid type does not return EINVAL. Interfaces for mutexes with an invalid type have unspecified behavior.

Uninitialized mutexes that are allocated locally could contain junk data. Such mutexes need to be initialized using mutex_init().

By default, if multiple threads are waiting for a mutex, the order of acquisition is undefined.