Scroll to navigation

MUTEX(9) Kernel Developer's Manual MUTEX(9)

NAME

mutex, mtx_init, mtx_destroy, mtx_lock, mtx_lock_spin, mtx_lock_flags, mtx_lock_spin_flags, mtx_trylock, mtx_trylock_flags, mtx_trylock_spin, mtx_trylock_spin_flags, mtx_unlock, mtx_unlock_spin, mtx_unlock_flags, mtx_unlock_spin_flags, mtx_sleep, mtx_initialized, mtx_owned, mtx_recursed, mtx_assert, MTX_SYSINIT
kernel synchronization primitives

SYNOPSIS

#include <sys/param.h>
#include <sys/lock.h>
#include <sys/mutex.h>

void
mtx_init(struct mtx *mutex, const char *name, const char *type, int opts);

void
mtx_destroy(struct mtx *mutex);

void
mtx_lock(struct mtx *mutex);

void
mtx_lock_spin(struct mtx *mutex);

void
mtx_lock_flags(struct mtx *mutex, int flags);

void
mtx_lock_spin_flags(struct mtx *mutex, int flags);

int
mtx_trylock(struct mtx *mutex);

int
mtx_trylock_flags(struct mtx *mutex, int flags);

void
mtx_trylock_spin(struct mtx *mutex);

int
mtx_trylock_spin_flags(struct mtx *mutex, int flags);

void
mtx_unlock(struct mtx *mutex);

void
mtx_unlock_spin(struct mtx *mutex);

void
mtx_unlock_flags(struct mtx *mutex, int flags);

void
mtx_unlock_spin_flags(struct mtx *mutex, int flags);

int
mtx_sleep(void *chan, struct mtx *mtx, int priority, const char *wmesg, int timo);

int
mtx_initialized(const struct mtx *mutex);

int
mtx_owned(const struct mtx *mutex);

int
mtx_recursed(const struct mtx *mutex);


options INVARIANTS
options INVARIANT_SUPPORT
void
mtx_assert(const struct mtx *mutex, int what);

#include <sys/kernel.h>

MTX_SYSINIT(name, struct mtx *mtx, const char *description, int opts);

DESCRIPTION

Mutexes are the most basic and primary method of thread synchronization. The major design considerations for mutexes are:
  1. Acquiring and releasing uncontested mutexes should be as cheap as possible.
  2. They must have the information and storage space to support priority propagation.
  3. A thread must be able to recursively acquire a mutex, provided that the mutex is initialized to support recursion.

There are currently two flavors of mutexes, those that context switch when they block and those that do not.

By default, MTX_DEF mutexes will context switch when they are already held. As an optimization, they may spin for some amount of time before context switching. It is important to remember that since a thread may be preempted at any time, the possible context switch introduced by acquiring a mutex is guaranteed to not break anything that is not already broken.

Mutexes which do not context switch are MTX_SPIN mutexes. These should only be used to protect data shared with primary interrupt code. This includes interrupt filters and low level scheduling code. In all architectures both acquiring and releasing of a uncontested spin mutex is more expensive than the same operation on a non-spin mutex. In order to protect an interrupt service routine from blocking against itself all interrupts are either blocked or deferred on a processor while holding a spin lock. It is permissible to hold multiple spin mutexes.

Once a spin mutex has been acquired it is not permissible to acquire a blocking mutex.

The storage needed to implement a mutex is provided by a struct mtx. In general this should be treated as an opaque object and referenced only with the mutex primitives.

The mtx_init() function must be used to initialize a mutex before it can be passed to any of the other mutex functions. The name option is used to identify the lock in debugging output etc. The type option is used by the witness code to classify a mutex when doing checks of lock ordering. If type is NULL, name is used in its place. The pointer passed in as name and type is saved rather than the data it points to. The data pointed to must remain stable until the mutex is destroyed. The opts argument is used to set the type of mutex. It may contain either MTX_DEF or MTX_SPIN but not both. If the kernel has been compiled with option INVARIANTS, mtx_init() will assert that the mutex has not been initialized multiple times without intervening calls to mtx_destroy() unless the MTX_NEW option is specified. See below for additional initialization options.

The mtx_lock() function acquires a MTX_DEF mutual exclusion lock on behalf of the currently running kernel thread. If another kernel thread is holding the mutex, the caller will be disconnected from the CPU until the mutex is available (i.e., it will block).

The mtx_lock_spin() function acquires a MTX_SPIN mutual exclusion lock on behalf of the currently running kernel thread. If another kernel thread is holding the mutex, the caller will spin until the mutex becomes available. Interrupts are disabled during the spin and remain disabled following the acquiring of the lock.

It is possible for the same thread to recursively acquire a mutex with no ill effects, provided that the MTX_RECURSE bit was passed to mtx_init() during the initialization of the mutex.

The mtx_lock_flags() and mtx_lock_spin_flags() functions acquire a MTX_DEF or MTX_SPIN lock, respectively, and also accept a flags argument. In both cases, the only flags presently available for lock acquires are MTX_QUIET and MTX_RECURSE. If the MTX_QUIET bit is turned on in the flags argument, then if KTR_LOCK tracing is being done, it will be silenced during the lock acquire. If the MTX_RECURSE bit is turned on in the flags argument, then the mutex can be acquired recursively.

The mtx_trylock() and mtx_trylock_spin() functions attempt to acquire a MTX_DEF or MTX_SPIN mutex, respectively, pointed to by mutex. If the mutex cannot be immediately acquired, the functions will return 0, otherwise the mutex will be acquired and a non-zero value will be returned.

The mtx_trylock_flags() and mtx_trylock_spin_flags() functions have the same behavior as mtx_trylock() and mtx_trylock_spin() respectively, but should be used when the caller desires to pass in a flags value. Presently, the only valid value in the mtx_trylock() and mtx_trylock_spin() cases is MTX_QUIET, and its effects are identical to those described for mtx_lock() above.

The mtx_unlock() function releases a MTX_DEF mutual exclusion lock. The current thread may be preempted if a higher priority thread is waiting for the mutex.

The mtx_unlock_spin() function releases a MTX_SPIN mutual exclusion lock.

The mtx_unlock_flags() and mtx_unlock_spin_flags() functions behave in exactly the same way as do the standard mutex unlock routines above, while also allowing a flags argument which may specify MTX_QUIET. The behavior of MTX_QUIET is identical to its behavior in the mutex lock routines.

The mtx_destroy() function is used to destroy mutex so the data associated with it may be freed or otherwise overwritten. Any mutex which is destroyed must previously have been initialized with mtx_init(). It is permissible to have a single hold count on a mutex when it is destroyed. It is not permissible to hold the mutex recursively, or have another thread blocked on the mutex when it is destroyed.

The mtx_sleep() function is used to atomically release mtx while waiting for an event. For more details on the parameters to this function, see sleep(9).

The mtx_initialized() function returns non-zero if mutex has been initialized and zero otherwise.

The mtx_owned() function returns non-zero if the current thread holds mutex. If the current thread does not hold mutex zero is returned.

The mtx_recursed() function returns non-zero if the mutex is recursed. This check should only be made if the running thread already owns mutex.

The mtx_assert() function allows assertions specified in what to be made about mutex. If the assertions are not true and the kernel is compiled with options INVARIANTS and options INVARIANT_SUPPORT, the kernel will panic. Currently the following assertions are supported:

Assert that the current thread holds the mutex pointed to by the first argument.
Assert that the current thread does not hold the mutex pointed to by the first argument.
Assert that the current thread has recursed on the mutex pointed to by the first argument. This assertion is only valid in conjunction with MA_OWNED.
Assert that the current thread has not recursed on the mutex pointed to by the first argument. This assertion is only valid in conjunction with MA_OWNED.

The MTX_SYSINIT() macro is used to generate a call to the mtx_sysinit() routine at system startup in order to initialize a given mutex lock. The parameters are the same as mtx_init() but with an additional argument, name, that is used in generating unique variable names for the related structures associated with the lock and the sysinit routine.

The Default Mutex Type

Most kernel code should use the default lock type, MTX_DEF. The default lock type will allow the thread to be disconnected from the CPU if the lock is already held by another thread. The implementation may treat the lock as a short term spin lock under some circumstances. However, it is always safe to use these forms of locks in an interrupt thread without fear of deadlock against an interrupted thread on the same CPU.

The Spin Mutex Type

A MTX_SPIN mutex will not relinquish the CPU when it cannot immediately get the requested lock, but will loop, waiting for the mutex to be released by another CPU. This could result in deadlock if another thread interrupted the thread which held a mutex and then tried to acquire the mutex. For this reason spin locks disable all interrupts on the local CPU.

Spin locks are fairly specialized locks that are intended to be held for very short periods of time. Their primary purpose is to protect portions of the code that implement other synchronization primitives such as default mutexes, thread scheduling, and interrupt threads.

Initialization Options

The options passed in the opts argument of mtx_init() specify the mutex type. One of the MTX_DEF or MTX_SPIN options is required and only one of those two options may be specified. The possibilities are:
Default mutexes will always allow the current thread to be suspended to avoid deadlock conditions against interrupt threads. The implementation of this lock type may spin for a while before suspending the current thread.
Spin mutexes will never relinquish the CPU. All interrupts are disabled on the local CPU while any spin lock is held.
Specifies that the initialized mutex is allowed to recurse. This bit must be present if the mutex is permitted to recurse.

Note that neither mtx_trylock() nor mtx_trylock_spin() support recursion; that is, attempting to acquire an already-owned mutex fails.

Do not log any mutex operations for this lock.
Instruct witness(4) to ignore this lock.
Witness should not log messages about duplicate locks being acquired.
Do not profile this lock.
Do not check for double-init.

Lock and Unlock Flags

The flags passed to the mtx_lock_flags(), mtx_lock_spin_flags(), mtx_unlock_flags(), and mtx_unlock_spin_flags() functions provide some basic options to the caller, and are often used only under special circumstances to modify lock or unlock behavior. Standard locking and unlocking should be performed with the mtx_lock(), mtx_lock_spin(), mtx_unlock(), and mtx_unlock_spin() functions. Only if a flag is required should the corresponding flags-accepting routines be used.

Options that modify mutex behavior:

This option is used to quiet logging messages during individual mutex operations. This can be used to trim superfluous logging messages for debugging purposes.

Giant

If Giant must be acquired, it must be acquired prior to acquiring other mutexes. Put another way: it is impossible to acquire Giant non-recursively while holding another mutex. It is possible to acquire other mutexes while holding Giant, and it is possible to acquire Giant recursively while holding other mutexes.

Sleeping

Sleeping while holding a mutex (except for Giant) is never safe and should be avoided. There are numerous assertions which will fail if this is attempted.

Functions Which Access Memory in Userspace

No mutexes should be held (except for Giant) across functions which access memory in userspace, such as copyin(9), copyout(9), uiomove(9), fuword(9), etc. No locks are needed when calling these functions.

SEE ALSO

condvar(9), LOCK_PROFILING(9), locking(9), mtx_pool(9), panic(9), rwlock(9), sema(9), sleep(9), sx(9)

HISTORY

These functions appeared in BSD/OS 4.1 and FreeBSD 5.0. The mtx_trylock_spin() function was added in FreeBSD 11.1.
May 24, 2017 Linux 4.19.0-10-amd64