freebsd-dev/share/man/man9/rwlock.9
John Baldwin e7573e7ad7 Allow threads to atomically release rw and sx locks while waiting for an
event.  Locking primitives that support this (mtx, rw, and sx) now each
include their own foo_sleep() routine.
- Rename msleep() to _sleep() and change it's 'struct mtx' object to a
  'struct lock_object' pointer.  _sleep() uses the recently added
  lc_unlock() and lc_lock() function pointers for the lock class of the
  specified lock to release the lock while the thread is suspended.
- Add wrappers around _sleep() for mutexes (mtx_sleep()), rw locks
  (rw_sleep()), and sx locks (sx_sleep()).  msleep() still exists and
  is now identical to mtx_sleep(), but it is deprecated.
- Rename SLEEPQ_MSLEEP to SLEEPQ_SLEEP.
- Rewrite much of sleep.9 to not be msleep(9) centric.
- Flesh out the 'RETURN VALUES' section in sleep.9 and add an 'ERRORS'
  section.
- Add __nonnull(1) to _sleep() and msleep_spin() so that the compiler will
  warn if you try to pass a NULL wait channel.  The functions already have
  a KASSERT to that effect.
2007-03-09 22:41:01 +00:00

254 lines
7.0 KiB
Groff

.\" 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.
.\"
.\" $FreeBSD$
.\"
.Dd April 19, 2006
.Dt RWLOCK 9
.Os
.Sh NAME
.Nm rwlock ,
.Nm rw_init ,
.Nm rw_destroy ,
.Nm rw_rlock ,
.Nm rw_wlock ,
.Nm rw_runlock ,
.Nm rw_wunlock ,
.Nm rw_try_upgrade ,
.Nm rw_downgrade ,
.Nm rw_sleep ,
.Nm rw_initialized ,
.Nm rw_wowned ,
.Nm rw_assert ,
.Nm RW_SYSINIT
.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_destroy "struct rwlock *rw"
.Ft void
.Fn rw_rlock "struct rwlock *rw"
.Ft void
.Fn rw_wlock "struct rwlock *rw"
.Ft void
.Fn rw_runlock "struct rwlock *rw"
.Ft void
.Fn rw_wunlock "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 "struct rwlock *rw"
.Ft int
.Fn rw_wowned "struct rwlock *rw"
.Pp
.Cd "options INVARIANTS"
.Cd "options INVARIANT_SUPPORT"
.Ft void
.Fn rw_assert "struct rwlock *rw" "int what"
.In sys/kernel.h
.Fn RW_SYSINIT "name" "struct rwlock *rw" "const char *desc"
.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 an exclusive holder.
This limitation comes from the fact that shared owners
are anonymous.
Another important property is that shared holders of
.Nm
can recurse,
but exclusive locks are not allowed to recurse.
.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_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 cannot be called recursively.
.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_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 "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 "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 "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 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
.El
.Sh SEE ALSO
.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
If
.Dv WITNESS
is not included in the kernel,
then it is impossible to assert that the current thread does or does not
hold a read lock.
In the
.Pf non- Dv WITNESS
case, the
.Dv RA_LOCKED
and
.Dv RA_RLOCKED
assertions merely check that some thread holds a read lock.
.Pp
Reader/writer is a bit of an awkward name.
An
.Nm
can also be called a
.Dq Robert Watson
lock if desired.