mutex, mtx_init, mtx_lock, mtx_lock_spin, mtx_lock_flags,
mtx_lock_spin_flags, mtx_trylock, mtx_trylock_flags, mtx_unlock,
mtx_unlock_spin, mtx_unlock_flags, mtx_unlock_spin_flags, mtx_destroy,
mtx_initialized, mtx_owned, mtx_recursed, mtx_assert, MTX_SYSINIT -- kernel
synchronization primitives
#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_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_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);
void
mtx_destroy(struct mtx *mutex);
int
mtx_initialized(struct mtx *mutex);
int
mtx_owned(struct mtx *mutex);
int
mtx_recursed(struct mtx *mutex);
void
mtx_assert(struct mtx *mutex, int what);
MTX_SYSINIT(name, struct mutex *mtx, const char *description, int opts);
Mutexes are the most basic and primary method of process 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 process 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 a machine dependent optimization they may spin for some amount
of time before context switching. It is important to remember that since
a process may be preempted at any time, the possible context switch
introduced by acquiring a mutex is guaranteed to not break anything that
isn't already broken.
Mutexes which do not context switch are MTX_SPIN mutexes. These should
only be used to protect data shared with any devices that require nonpreemptive
interrupts, and low level scheduling code. In most/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 blocked on a processor while holding a spin lock. It is permissible
to hold multiple spin mutexes. In this case it is a required
that they be released in the opposite order to that which they were
acquired.
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 mtx_lock(). 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. See below for additional initialization options. It is not permissible
to pass the same mutex to mtx_init() multiple times without
intervening calls to mtx_destroy().
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 sleep).
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 flag presently available for lock acquires is
MTX_QUIET. 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.
The mtx_trylock() function is used to acquire exclusive access to those
objects protected by the mutex pointed to by mutex. If the mutex cannot
be immediately acquired mtx_trylock() will return 0, otherwise the mutex
will be acquired and a non-zero value will be returned.
The mtx_trylock_flags() function has the same behavior as mtx_trylock()
but should be used when the caller desires to pass in a flags value.
Presently, the only valid value in the mtx_trylock() case is MTX_QUIET,
and its effects are identical to those described for mtx_lock() and
mtx_lock_spin() above.
The mtx_unlock() function releases a MTX_DEF mutual exclusion lock; if a
higher priority thread is waiting for the mutex, the releasing thread
will be disconnected to allow the higher priority thread to acquire the
mutex and run unless the current thread is executing in a critical section.
The mtx_unlock_spin() function releases a MTX_SPIN mutual exclusion lock;
interrupt state prior to the acquiring of the lock is restored.
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
process blocked on the mutex when it is destroyed.
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 process holds
mutex. If the current process 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 to be made about mutex. If
the assertions are not true and the kernel is compiled with INVARIANTS
then the kernel will panic. Currently the following assertions are supported:
MA_OWNED Assert that the current thread holds the mutex pointed to
by the first argument.
MA_NOTOWNED Assert that the current thread does not hold the mutex
pointed to by the first argument.
MA_RECURSED 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.
MA_NOTRECURSED 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 [Toc] [Back]
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 it
cannot get the lock. The machine dependent 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 [Toc] [Back]
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 a thread
interrupted the thread which held a mutex and then tried to acquire the
mutex; for this reason spin locks will disable all interrupts (on the
local CPU only).
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 default (i.e. sleep) locks.
Initialization Options [Toc] [Back]
The options passed in the opts argument of mtx_init() specify the mutex
type. The possibilities are:
MTX_DEF Default lock type; will always allow the current thread to
be suspended to avoid deadlock conditions against interrupt
threads. The machine dependent implementation of
this lock type may spin for a while before suspending the
current thread. If this flag is specified, clearly
MTX_SPIN must NOT be specified.
MTX_SPIN Spin lock type; will never relinquish the CPU. All interrupts
are disabled on the local CPU while any spin lock is
held.
MTX_RECURSE Recursion option bit; specifies that the initialized mutex
is allowed to recurse. This bit must be present if the
mutex is going to be permitted to recurse.
MTX_QUIET Do not log any mutex operations for this lock.
MTX_NOWITNESS Instruct witness(4) to ignore this lock.
MTX_DUPOK Witness should not log messages about duplicate locks
being acquired.
Lock and Unlock Flags [Toc] [Back]
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:
MTX_QUIET 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 [Toc] [Back]
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 [Toc] [Back]
Sleeping while holding a mutex (except for Giant) is almost never safe
and should be avoided. There are numerous assertions which will fail if
this is attempted.
Functions Which Access Memory in Userspace [Toc] [Back]
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.
condvar(9), msleep(9), mtx_pool(9), sema(9), sx(9)
These functions appeared in BSD/OS 4.1 and FreeBSD 5.0.
FreeBSD 5.2.1 February 12, 2001 FreeBSD 5.2.1 [ Back ] |
