.\" $OpenBSD: rwlock.9,v 1.9 2010/08/01 14:21:10 blambert Exp $
.\"
.\" Copyright (c) 2006 Pedro Martelletto <pedro@ambientworks.net>
.\" All rights reserved.
.\"
.\" Permission to use, copy, modify, and distribute this software for any
.\" purpose with or without fee is hereby granted, provided that the above
.\" copyright notice and this permission notice appear in all copies.
.\"
.\" THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
.\" WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
.\" MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
.\" ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
.\" WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
.\" ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
.\" OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
.\"
.Dd $Mdocdate: August 1 2010 $
.Dt RWLOCK 9
.Os
.Sh NAME
.Nm rwlock ,
.Nm rw_init ,
.Nm rw_enter ,
.Nm rw_exit ,
.Nm rw_enter_read ,
.Nm rw_enter_write ,
.Nm rw_exit_read ,
.Nm rw_exit_write
.Nd interface to read/write locks
.Sh SYNOPSIS
.Fd #include <sys/rwlock.h>
.Ft void
.Fn rw_init "struct rwlock *rwl" "const char *name"
.Ft int
.Fn rw_enter "struct rwlock *rwl" "int flags"
.Ft void
.Fn rw_exit "struct rwlock *rwl"
.Ft void
.Fn rw_enter_read "struct rwlock *rwl"
.Ft void
.Fn rw_enter_write "struct rwlock *rwl"
.Ft void
.Fn rw_exit_read "struct rwlock *rwl"
.Ft void
.Fn rw_exit_write "struct rwlock *rwl"
.Ft void
.Fn rw_assert_wrlock "struct rwlock *rwl"
.Ft void
.Fn rw_assert_rdlock "struct rwlock *rwl"
.Ft void
.Fn rw_assert_unlocked "struct rwlock *rwl"
.Sh DESCRIPTION
The
.Nm
set of functions provides a multiple-reader, single-writer locking mechanism to
ensure mutual exclusion between different processes.
.Pp
The
.Fn rw_init
function is used to initiate the lock pointed to by
.Fa rwl .
The
.Fa name
argument specifies the name of the lock, which is used as the wait message
if the process needs to sleep.
.Pp
The
.Fn rw_enter
function acquires a lock.
The
.Fa flags
argument specifies what kind of lock should be obtained and also
modifies the operation.
The possible flags are:
.Pp
.Bl -tag -offset indent -width RW_DOWNGRADEXXX -compact
.It Dv RW_READ
Acquire a shared lock.
.It Dv RW_WRITE
Acquire an exclusive lock.
.It Dv RW_DOWNGRADE
Safely release an exclusive lock and acquire a shared lock without
letting other exclusive locks in between.
.It Dv RW_INTR
When waiting for a lock, allow signals to interrupt the sleep.
.It Dv RW_NOSLEEP
Do not wait for busy locks, fail with
.Dv EBUSY
instead.
.It Dv RW_SLEEPFAIL
Wait for busy locks, but do not obtain them, fail with
.Dv EAGAIN
instead.
.El
.Pp
The
.Fn rw_exit
function is used to release a held lock.
.Pp
The
.Fn rw_enter_read
function acquires a read lock, sleeping if necessary.
.Pp
The
.Fn rw_enter_write
function acquires a write lock, sleeping if necessary.
.Pp
The
.Fn rw_exit_read
function releases a read lock.
.Pp
The
.Fn rw_exit_write
function releases a write lock.
.Pp
Read locks can be acquired while the write lock is not held, and may coexist in
distinct processes at any time.
A write lock, however, can only be acquired when there are no read locks held,
granting exclusive access to a single process.
.Pp
The
.Fn rw_assert_wrlock ,
.Fn rw_assert_rdlock ,
and
.Fn rw_assert_unlocked
functions check the status
.Fa rwl ,
panicking if it is not write-, read-, or unlocked, respectively.
.Sh SEE ALSO
.Xr lockmgr 9 ,
.Xr mutex 9 ,
.Xr spl 9
.Sh HISTORY
The
.Nm
functions first appeared in
.Ox 3.5 .
.Sh AUTHORS
The
.Nm
functions were written by
.An Artur Grabowski
.Aq art@openbsd.org .
.Sh CAVEATS
While it is safe to sleep with an rwlock held, they cannot
be used in an interrupt handler as an rwlock is bound to a process.