rwlock.9

.\" Copyright (c) 2006 Gleb Smirnoff <glebius@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, 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.
.\"
.\" THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``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 AUTHOR OR CONTRIBUTORS 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.
.\"
.Dd November 11, 2017
.Dt RWLOCK 9
.Os
.Sh NAME
.Nm rwlock ,
.Nm rw_init ,
.Nm rw_init_flags ,
.Nm rw_destroy ,
.Nm rw_rlock ,
.Nm rw_wlock ,
.Nm rw_runlock ,
.Nm rw_wunlock ,
.Nm rw_unlock ,
.Nm rw_try_rlock ,
.Nm rw_try_upgrade ,
.Nm rw_try_wlock ,
.Nm rw_downgrade ,
.Nm rw_sleep ,
.Nm rw_initialized ,
.Nm rw_wowned ,
.Nm rw_assert ,
.Nm RW_SYSINIT ,
.Nm RW_SYSINIT_FLAGS
.Nd kernel reader/writer lock
.Sh SYNOPSIS
.In sys/param.h
.In sys/lock.h
.In sys/rwlock.h
.Ft void
.Fn rw_init "struct rwlock *rw" "const char *name"
.Ft void
.Fn rw_init_flags "struct rwlock *rw" "const char *name" "int opts"
.Ft void
.Fn rw_destroy "struct rwlock *rw"
.Ft void
.Fn rw_rlock "struct rwlock *rw"
.Ft void
.Fn rw_wlock "struct rwlock *rw"
.Ft int
.Fn rw_try_rlock "struct rwlock *rw"
.Ft int
.Fn rw_try_wlock "struct rwlock *rw"
.Ft void
.Fn rw_runlock "struct rwlock *rw"
.Ft void
.Fn rw_wunlock "struct rwlock *rw"
.Ft void
.Fn rw_unlock "struct rwlock *rw"
.Ft int
.Fn rw_try_upgrade "struct rwlock *rw"
.Ft void
.Fn rw_downgrade "struct rwlock *rw"
.Ft int
.Fn rw_sleep "void *chan" "struct rwlock *rw" "int priority" "const char *wmesg" "int timo"
.Ft int
.Fn rw_initialized "const struct rwlock *rw"
.Ft int
.Fn rw_wowned "const struct rwlock *rw"
.Pp
.Cd "options INVARIANTS"
.Cd "options INVARIANT_SUPPORT"
.Ft void
.Fn rw_assert "const struct rwlock *rw" "int what"
.In sys/kernel.h
.Fn RW_SYSINIT "name" "struct rwlock *rw" "const char *desc"
.Fn RW_SYSINIT_FLAGS "name" "struct rwlock *rw" "const char *desc" "int flags"
.Sh DESCRIPTION
Reader/writer locks allow shared access to protected data by multiple threads,
or exclusive access by a single thread.
The threads with shared access are known as
.Em readers
since they only read the protected data.
A thread with exclusive access is known as a
.Em writer
since it can modify protected data.
.Pp
Although reader/writer locks look very similar to
.Xr sx 9
locks, their usage pattern is different.
Reader/writer locks can be treated as mutexes (see
.Xr mutex 9 )
with shared/exclusive semantics.
Unlike
.Xr sx 9 ,
an
.Nm
can be locked while holding a non-spin mutex, and an
.Nm
cannot be held while sleeping.
The
.Nm
locks have priority propagation like mutexes, but priority
can be propagated only to writers.
This limitation comes from the fact that readers
are anonymous.
Another important property is that readers can always recurse,
and exclusive locks can be made recursive selectively.
.Ss Macros and Functions
.Bl -tag -width indent
.It Fn rw_init "struct rwlock *rw" "const char *name"
Initialize structure located at
.Fa rw
as reader/writer lock, described by name
.Fa name .
The description is used solely for debugging purposes.
This function must be called before any other operations
on the lock.
.It Fn rw_init_flags "struct rwlock *rw" "const char *name" "int opts"
Initialize the rw lock just like the
.Fn rw_init
function, but specifying a set of optional flags to alter the
behaviour of
.Fa rw ,
through the
.Fa opts
argument.
It contains one or more of the following flags:
.Bl -tag -width ".Dv RW_NOPROFILE"
.It Dv RW_DUPOK
Witness should not log messages about duplicate locks being acquired.
.It Dv RW_NOPROFILE
Do not profile this lock.
.It Dv RW_NOWITNESS
Instruct
.Xr witness 4
to ignore this lock.
.It Dv RW_QUIET
Do not log any operations for this lock via
.Xr ktr 4 .
.It Dv RW_RECURSE
Allow threads to recursively acquire exclusive locks for
.Fa rw .
.It Dv RW_NEW
If the kernel has been compiled with
.Cd "option INVARIANTS" ,
.Fn rw_init_flags
will assert that the
.Fa rw
has not been initialized multiple times without intervening calls to
.Fn rw_destroy
unless this option is specified.
.El
.It Fn rw_rlock "struct rwlock *rw"
Lock
.Fa rw
as a reader.
If any thread holds this lock exclusively, the current thread blocks,
and its priority is propagated to the exclusive holder.
The
.Fn rw_rlock
function can be called when the thread has already acquired reader
access on
.Fa rw .
This is called
.Dq "recursing on a lock" .
.It Fn rw_wlock "struct rwlock *rw"
Lock
.Fa rw
as a writer.
If there are any shared owners of the lock, the current thread blocks.
The
.Fn rw_wlock
function can be called recursively only if
.Fa rw
has been initialized with the
.Dv RW_RECURSE
option enabled.
.It Fn rw_try_rlock "struct rwlock *rw"
Try to lock
.Fa rw
as a reader.
This function will return true if the operation succeeds, otherwise 0
will be returned.
.It Fn rw_try_wlock "struct rwlock *rw"
Try to lock
.Fa rw
as a writer.
This function will return true if the operation succeeds, otherwise 0
will be returned.
.It Fn rw_runlock "struct rwlock *rw"
This function releases a shared lock previously acquired by
.Fn rw_rlock .
.It Fn rw_wunlock "struct rwlock *rw"
This function releases an exclusive lock previously acquired by
.Fn rw_wlock .
.It Fn rw_unlock "struct rwlock *rw"
This function releases a shared lock previously acquired by
.Fn rw_rlock
or an exclusive lock previously acquired by
.Fn rw_wlock .
.It Fn rw_try_upgrade "struct rwlock *rw"
Attempt to upgrade a single shared lock to an exclusive lock.
The current thread must hold a shared lock of
.Fa rw .
This will only succeed if the current thread holds the only shared lock on
.Fa rw ,
and it only holds a single shared lock.
If the attempt succeeds
.Fn rw_try_upgrade
will return a non-zero value,
and the current thread will hold an exclusive lock.
If the attempt fails
.Fn rw_try_upgrade
will return zero,
and the current thread will still hold a shared lock.
.It Fn rw_downgrade "struct rwlock *rw"
Convert an exclusive lock into a single shared lock.
The current thread must hold an exclusive lock of
.Fa rw .
.It Fn rw_sleep "void *chan" "struct rwlock *rw" "int priority" "const char *wmesg" "int timo"
Atomically release
.Fa rw
while waiting for an event.
For more details on the parameters to this function,
see
.Xr sleep 9 .
.It Fn rw_initialized "const struct rwlock *rw"
This function returns non-zero if
.Fa rw
has been initialized, and zero otherwise.
.It Fn rw_destroy "struct rwlock *rw"
This functions destroys a lock previously initialized with
.Fn rw_init .
The
.Fa rw
lock must be unlocked.
.It Fn rw_wowned "const struct rwlock *rw"
This function returns a non-zero value if the current thread owns an
exclusive lock on
.Fa rw .
.It Fn rw_assert "const struct rwlock *rw" "int what"
This function allows assertions specified in
.Fa what
to be made about
.Fa rw .
If the assertions are not true and the kernel is compiled
with
.Cd "options INVARIANTS"
and
.Cd "options INVARIANT_SUPPORT" ,
the kernel will panic.
Currently the following base assertions are supported:
.Bl -tag -width ".Dv RA_UNLOCKED"
.It Dv RA_LOCKED
Assert that current thread holds either a shared or exclusive lock
of
.Fa rw .
.It Dv RA_RLOCKED
Assert that current thread holds a shared lock of
.Fa rw .
.It Dv RA_WLOCKED
Assert that current thread holds an exclusive lock of
.Fa rw .
.It Dv RA_UNLOCKED
Assert that current thread holds neither a shared nor exclusive lock of
.Fa rw .
.El
.Pp
In addition, one of the following optional flags may be specified with
.Dv RA_LOCKED ,
.Dv RA_RLOCKED ,
or
.Dv RA_WLOCKED :
.Bl -tag -width ".Dv RA_NOTRECURSED"
.It Dv RA_RECURSED
Assert that the current thread holds a recursive lock of
.Fa rw .
.It Dv RA_NOTRECURSED
Assert that the current thread does not hold a recursive lock of
.Fa rw .
.El
.El
.Sh SEE ALSO
.Xr locking 9 ,
.Xr mutex 9 ,
.Xr panic 9 ,
.Xr sema 9 ,
.Xr sx 9
.Sh HISTORY
These
functions appeared in
.Fx 7.0 .
.Sh AUTHORS
.An -nosplit
The
.Nm
facility was written by
.An "John Baldwin" .
This manual page was written by
.An "Gleb Smirnoff" .
.Sh BUGS
A kernel without
.Dv WITNESS
cannot assert whether the current thread does or does not hold a read lock.
.Dv RA_LOCKED
and
.Dv RA_RLOCKED
can only assert that
.Em any
thread holds a read lock.
They cannot ensure that the current thread holds a read lock.
Further,
.Dv RA_UNLOCKED
can only assert that the current thread does not hold a write lock.
.Pp
Reader/writer is a bit of an awkward name.
An
.Nm
can also be called a
.Dq Robert Watson
lock if desired.