405 lines
13 KiB
Groff
405 lines
13 KiB
Groff
.\"
|
|
.\" Copyright (c) 1998 Berkeley Software Design, Inc. 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, this list of conditions and the following disclaimer.
|
|
.\" 2. Redistributions in binary form must reproduce the above copyright
|
|
.\" notice, this list of conditions and the following disclaimer in the
|
|
.\" documentation and/or other materials provided with the distribution.
|
|
.\" 3. Berkeley Software Design Inc's name may not be used to endorse or
|
|
.\" promote products derived from this software without specific prior
|
|
.\" written permission.
|
|
.\"
|
|
.\" THIS SOFTWARE IS PROVIDED BY BERKELEY SOFTWARE DESIGN INC ``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 BERKELEY SOFTWARE DESIGN INC 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.
|
|
.\"
|
|
.\" from BSDI $Id: mutex.4,v 1.1.2.3 1998/04/27 22:53:13 ewv Exp $
|
|
.\" $FreeBSD$
|
|
.\"
|
|
.Dd April 20, 1998
|
|
.Dt MUTEX 9
|
|
.Sh NAME
|
|
.Nm mutex ,
|
|
.Nm mtx_init ,
|
|
.Nm mtx_enter ,
|
|
.Nm mtx_try_enter ,
|
|
.Nm mtx_exit ,
|
|
.Nm mtx_destroy ,
|
|
.Nm mtx_owned ,
|
|
.Nm mtx_assert ,
|
|
.Nm MUTEX_DECLARE
|
|
.Nd kernel synchronization primitives
|
|
.Sh SYNOPSIS
|
|
.Fd #include <sys/mutex.h>
|
|
.Ft int
|
|
.Fn mtx_init "struct mtx *mutex" "char *name" "flags"
|
|
.Ft void
|
|
.Fn mtx_enter "struct mtx *mutex" "int flags"
|
|
.Ft void
|
|
.Fn mtx_try_enter "struct mtx *mutex" "int flags"
|
|
.Ft void
|
|
.Fn mtx_exit "struct mtx *mutex" "int flags"
|
|
.Ft void
|
|
.Fn mtx_destroy "struct mtx *mutex"
|
|
.Ft int
|
|
.Fn mtx_owned "struct mtx *mutex"
|
|
.Ft void
|
|
.Fn mtx_assert "struct mtx *mutex" "int what"
|
|
.Fn MUTEX_DECLARE "modifiers" "name"
|
|
.Sh DESCRIPTION
|
|
Mutexes are the most basic and primary method of process synchronization.
|
|
The major design considerations for mutexes are:
|
|
.Bl -enum
|
|
.It
|
|
Acquiring and releasing uncontested mutexes should be as cheap
|
|
as possible.
|
|
.It
|
|
They must have the information and storage space to support
|
|
priority propagation.
|
|
.It
|
|
A process must be able to recursively acquire a mutex.
|
|
.El
|
|
.Pp
|
|
There are currently two flavors of mutexes, those that context switch
|
|
when they block and those that do not.
|
|
.Pp
|
|
By default 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.
|
|
.Pp
|
|
Mutexes which do not context switch are spin mutexes.
|
|
These should only be used to protect data shared with device that
|
|
require non-preemptive 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.
|
|
.Pp
|
|
Once a spin mutex has been acquired it is not permissible to acquire a
|
|
blocking mutex.
|
|
.Pp
|
|
The storage needed to implement a mutex is provided by a
|
|
.Dv struct mtx .
|
|
In general this should be treated as an opaque object and
|
|
referenced only with the mutex primitives.
|
|
.Pp
|
|
The
|
|
.Fn mtx_init
|
|
function must be used to initialize a mutex
|
|
before it can be passed to
|
|
.Fn mtx_enter .
|
|
The
|
|
.Ar name
|
|
argument is used by the witness code
|
|
to classify a mutex when doing checks of lock ordering.
|
|
The pointer passed in as the
|
|
.Ar name
|
|
is saved rather than the data it points to.
|
|
The data pointed to must remain stable
|
|
until the mutex is destroyed.
|
|
Currently the
|
|
.Ar flag
|
|
argument is unused.
|
|
In the future it will likely be at least
|
|
used to identify spin mutexes
|
|
so that debug code can verify
|
|
consistent use of a mutex.
|
|
It is not permissible to pass the same
|
|
.Ar mutex
|
|
to
|
|
.Fn mtx_init
|
|
multiple times without intervening calls to
|
|
.Fn mtx_destroy .
|
|
.Pp
|
|
The
|
|
.Fn mtx_enter
|
|
function acquires a 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),
|
|
spin wait for the mutex,
|
|
or possibly a combination of both.
|
|
.Pp
|
|
It is possible for the same thread to recursively acquire a mutex
|
|
with no ill effects;
|
|
if recursion on a given mutex can be avoided,
|
|
faster and smaller code will usually be generated.
|
|
.Pp
|
|
The
|
|
.Fn mtx_try_enter
|
|
function is used to acquire exclusive access
|
|
to those objects protected by the mutex
|
|
pointed to by
|
|
.Ar mutex .
|
|
The
|
|
.Ar flag
|
|
argument is used to specify various options,
|
|
typically
|
|
.Dv MTX_DEF
|
|
is supplied.
|
|
If the mutex can not be immediately acquired
|
|
.Fn mtx_try_enter
|
|
will return 0,
|
|
otherwise the mutex will be acquired
|
|
and a non-zero value will be returned.
|
|
.Pp
|
|
The
|
|
.Fn mtx_exit
|
|
function releases a mutual exclusion lock;
|
|
if a higher priority thread is waiting for the mutex,
|
|
the releasing thread may be disconnected
|
|
to allow the higher priority thread to acquire the mutex and run.
|
|
.Pp
|
|
The
|
|
.Fn mtx_destroy
|
|
function is used to destroy
|
|
.Ar mutex
|
|
so the data associated with it may be freed
|
|
or otherwise overwritten.
|
|
Any mutex which is destroyed
|
|
must previously have been initialized with
|
|
.Fn 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.
|
|
.Pp
|
|
The
|
|
.Fn mtx_owned
|
|
function returns non-zero
|
|
if the current process holds
|
|
.Ar mutex .
|
|
If the current process does not hold
|
|
.Ar mutex
|
|
zero is returned.
|
|
.Pp
|
|
The
|
|
.Fn mtx_assert
|
|
function allows assertions to be made about
|
|
.Ar mutex .
|
|
If the assertions are not true and the kernel is compiled with
|
|
.Dv INVARIANTS
|
|
then the kernel will panic.
|
|
Currently the following assertions are supported:
|
|
.Bl -enum
|
|
.It
|
|
.Dv MA_OWNED
|
|
Assert that the current thread
|
|
holds the mutex
|
|
pointed to by the first argument.
|
|
.It
|
|
.Dv MA_NOTOWNED
|
|
Assert that the current thread
|
|
does not hold the mutex
|
|
pointed to by the first argument.
|
|
.El
|
|
.Pp
|
|
The
|
|
.Fn MUTEX_DECLARE
|
|
macro is used to declare a mutex that is initialized before
|
|
.Xr malloc 9
|
|
is operating.
|
|
Unfortunately, mutex initialization may require
|
|
.Xr malloc 9 .
|
|
However, some mutexes are intialized and used before
|
|
.Xr malloc 9
|
|
can be used.
|
|
Declaring these mutexes with the
|
|
.Fn MUTEX_DECLARE
|
|
macro and then using the
|
|
.Dv MTX_COLD
|
|
flag when calling
|
|
.Fn mtx_init
|
|
allows these early mutexes to be initialized and used before
|
|
.Xr malloc 9
|
|
is available.
|
|
The
|
|
.Ar modifiers
|
|
argument is a list of attributes to be applied to the mutex structure being
|
|
declared such as
|
|
.Dq static .
|
|
The
|
|
.Ar name
|
|
argument is the name of the mutex structure to declare.
|
|
.Pp
|
|
The type of a mutex is not an attribute of the mutex,
|
|
but instead a function of the
|
|
.Fa flags
|
|
argument passed to
|
|
.Fn mtx_enter
|
|
and
|
|
.Fn mtx_exit ;
|
|
this allows code to be generated for the specific mutex type
|
|
at compile time
|
|
and avoids wasting run time on the determination of lock features.
|
|
This does place on the programmer,
|
|
the burden of using matching forms of the
|
|
.Fn mtx_enter
|
|
and
|
|
.Fn mtx_exit
|
|
functions for a given mutex.
|
|
It is an error to acquire a mutex in one mode (e.g. spin)
|
|
and release it in another (e.g. default).
|
|
It is also an error to get the lock in one mode
|
|
and allow another thread to attempt to get the lock in another mode.
|
|
A good general rule is to always use a given mutex in one mode only.
|
|
.Ss The default Mutex Type
|
|
Most kernel code should use the default lock type;
|
|
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.
|
|
.Ss The spin Mutex Type
|
|
A 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)
|
|
by default.
|
|
.Pp
|
|
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.
|
|
.Ss Flags
|
|
The flags passed to the
|
|
.Fn mtx_enter
|
|
and
|
|
.Fn mtx_exit
|
|
functions determine what type of mutex is being used
|
|
and also provide various options
|
|
used to generate more efficient code under certain circumstances.
|
|
.Pp
|
|
Both lock types (default and spin)
|
|
can be acquired recursively by the same thread.
|
|
This behavior can be changed with flags.
|
|
.Pp
|
|
The type of the mutex must always be specified:
|
|
.Bl -tag -width MTX_NORECURSE
|
|
.It Dv 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.
|
|
Most locks should be of this type.
|
|
.It Dv MTX_SPIN
|
|
Spin lock;
|
|
will never relinquish the CPU.
|
|
By default all interrupts are disabled on the local CPU
|
|
while any spin lock is held.
|
|
.El
|
|
.Pp
|
|
Options that modify mutex behavior:
|
|
.Bl -tag -width MTX_NORECURSE
|
|
.It Dv MTX_NORECURSE
|
|
If it is known, absolutely,
|
|
that the mutex will not be recursively acquired at this invocation
|
|
then this flag should be specified.
|
|
.Pp
|
|
If the lock is already held by the current thread,
|
|
then a kernel with
|
|
.Dv MUTEX_DEBUG
|
|
defined will panic;
|
|
without debugging enabled,
|
|
the thread may deadlock against itself
|
|
or leave the mutex in a corrupted state.
|
|
.Pp
|
|
This flag prevents generation of additional inline code
|
|
to deal with recursive lock acquisitions
|
|
and should be specified whenever possible
|
|
in the interests of efficiency.
|
|
Not specifying this flag will only cause the generated code
|
|
to be a little larger than necessary;
|
|
it will still operate correctly.
|
|
.It Dv MTX_RLIKELY
|
|
This provides a hint that it is likely that this mutex
|
|
will be held recursively at this invocation.
|
|
The actual optimization used is machine dependent;
|
|
generally, this will inline code to handle recursion
|
|
where a function call would otherwise be needed.
|
|
.Pp
|
|
This is a hint only;
|
|
leaving it out or specifying it inappropriately
|
|
will not cause any great harm other than
|
|
possibly generating less efficient code.
|
|
.It Dv MTX_TOPHALF
|
|
This option applies to spin locks only.
|
|
It indicates that the mutex is never acquired
|
|
from an interrupt thread,
|
|
so it is safe to leave interrupts enabled while holding the lock.
|
|
Since an interrupt may occur while holding the lock,
|
|
this may be detrimental to other processors
|
|
spin waiting for the lock.
|
|
Do not forget to include this option when the lock is released.
|
|
.Pp
|
|
This option should not be used in new code;
|
|
it is documented here for completeness only.
|
|
.It Dv MTX_FIRST
|
|
This option applies to spin locks only.
|
|
It indicates this is the first spin lock acquired by the thread.
|
|
No other spin locks may be held,
|
|
and the requested lock also may not be currently held.
|
|
Do not forget to include this option when the lock is released.
|
|
.It Dv MTX_NOSWITCH
|
|
When releasing a mutex,
|
|
this flag prevents a thread switch that might occur
|
|
if another higher priority thread was waiting for the mutex.
|
|
This may cause priority inversion and should be used carefully.
|
|
.Pp
|
|
This flag is used internally by the lock code.
|
|
It should not be used in general kernel code
|
|
and is documented here for completeness only.
|
|
.It Dv MTX_NOSPIN
|
|
For default locks,
|
|
this hint will prevent spinning before relinquishing the CPU.
|
|
This should be specified when it is known
|
|
that the lock will usually remain unavailable for some time
|
|
when it is not immediately available
|
|
(i.e.: coarse grained locks protecting large subsystems).
|
|
.It Dv MTX_COLD
|
|
This option is only used in
|
|
.Fn mtx_init
|
|
and is used in conjunction with mutexes declared with
|
|
.Fn MUTEX_DECLARE
|
|
to initialize mutexes that are needed before
|
|
.Xr malloc 9
|
|
is available for use.
|
|
.El
|
|
.Sh HISTORY
|
|
These
|
|
functions appeared in BSD/OS 4.1 and
|
|
.Fx 5.0 .
|