1
0
mirror of https://git.FreeBSD.org/src.git synced 2024-12-23 11:18:54 +00:00
freebsd/share/man/man9/rmlock.9
John Baldwin cd32bd7ad1 Several improvements to rmlock(9). Many of these are based on patches
provided by Isilon.
- Add an rm_assert() supporting various lock assertions similar to other
  locking primitives.  Because rmlocks track readers the assertions are
  always fully accurate unlike rw_assert() and sx_assert().
- Flesh out the lock class methods for rmlocks to support sleeping via
  condvars and rm_sleep() (but only while holding write locks), rmlock
  details in 'show lock' in DDB, and the lc_owner method used by
  dtrace.
- Add an internal destroyed cookie so that API functions can assert
  that an rmlock is not destroyed.
- Make use of rm_assert() to add various assertions to the API (e.g.
  to assert locks are held when an unlock routine is called).
- Give RM_SLEEPABLE locks their own lock class and always use the
  rmlock's own lock_object with WITNESS.
- Use THREAD_NO_SLEEPING() / THREAD_SLEEPING_OK() to disallow sleeping
  while holding a read lock on an rmlock.

Submitted by:	andre
Obtained from:	EMC/Isilon
2013-06-25 18:44:15 +00:00

315 lines
9.0 KiB
Groff

.\" Copyright (c) 2007 Stephan Uphoff <ups@FreeBSD.org>
.\" 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.
.\"
.\" $FreeBSD$
.\"
.\" Based on rwlock.9 man page
.Dd June 25, 2013
.Dt RMLOCK 9
.Os
.Sh NAME
.Nm rmlock ,
.Nm rm_init ,
.Nm rm_init_flags ,
.Nm rm_destroy ,
.Nm rm_rlock ,
.Nm rm_try_rlock ,
.Nm rm_wlock ,
.Nm rm_runlock ,
.Nm rm_wunlock ,
.Nm rm_wowned ,
.Nm rm_sleep ,
.Nm rm_assert ,
.Nm RM_SYSINIT
.Nd kernel reader/writer lock optimized for read-mostly access patterns
.Sh SYNOPSIS
.In sys/param.h
.In sys/lock.h
.In sys/rmlock.h
.Ft void
.Fn rm_init "struct rmlock *rm" "const char *name"
.Ft void
.Fn rm_init_flags "struct rmlock *rm" "const char *name" "int opts"
.Ft void
.Fn rm_destroy "struct rmlock *rm"
.Ft void
.Fn rm_rlock "struct rmlock *rm" "struct rm_priotracker* tracker"
.Ft int
.Fn rm_try_rlock "struct rmlock *rm" "struct rm_priotracker* tracker"
.Ft void
.Fn rm_wlock "struct rmlock *rm"
.Ft void
.Fn rm_runlock "struct rmlock *rm" "struct rm_priotracker* tracker"
.Ft void
.Fn rm_wunlock "struct rmlock *rm"
.Ft int
.Fn rm_wowned "const struct rmlock *rm"
.Ft int
.Fn rm_sleep "void *wchan" "struct rmlock *rm" "int priority" "const char *wmesg" "int timo"
.Pp
.Cd "options INVARIANTS"
.Cd "options INVARIANT_SUPPORT"
.Ft void
.Fn rm_assert "struct rmlock *rm" "int what"
.In sys/kernel.h
.Fn RM_SYSINIT "name" "struct rmlock *rm" "const char *desc" "int opts"
.Sh DESCRIPTION
Read-mostly 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
Read-mostly locks are designed to be efficient for locks almost exclusively
used as reader locks and as such should be used for protecting data that
rarely changes.
Acquiring an exclusive lock after the lock has been locked for shared access
is an expensive operation.
.Pp
Normal read-mostly locks are similar to
.Xr rwlock 9
locks and follow the same lock ordering rules as
.Xr rwlock 9
locks.
Read-mostly locks have full priority propagation like mutexes.
Unlike
.Xr rwlock 9 ,
read-mostly locks propagate priority to both readers and writers.
This is implemented via the
.Va rm_priotracker
structure argument supplied to
.Fn rm_rlock
and
.Fn rm_runlock .
Readers can recurse if the lock is initialized with the
.Dv RM_RECURSE
option;
however, writers are never allowed to recurse.
.Pp
Sleepable read-mostly locks are created by passing
.Dv RM_SLEEPABLE
to
.Fn rm_init_flags .
Unlike normal read-mostly locks,
sleepable read-mostly locks follow the same lock ordering rules as
.Xr sx 9
locks.
Sleepable read-mostly locks do not propagate priority to writers,
but they do propagate priority to readers.
Writers are permitted to sleep while holding a read-mostly lock,
but readers are not.
Unlike other sleepable locks such as
.Xr sx 9
locks,
readers must use try operations on other sleepable locks to avoid sleeping.
.Ss Macros and Functions
.Bl -tag -width indent
.It Fn rm_init "struct rmlock *rm" "const char *name"
Initialize the read-mostly lock
.Fa rm .
The
.Fa name
description is used solely for debugging purposes.
This function must be called before any other operations
on the lock.
.It Fn rm_init_flags "struct rmlock *rm" "const char *name" "int opts"
Similar to
.Fn rm_init ,
initialize the read-mostly lock
.Fa rm
with a set of optional flags.
The
.Fa opts
arguments contains one or more of the following flags:
.Bl -tag -width ".Dv RM_NOWITNESS"
.It Dv RM_NOWITNESS
Instruct
.Xr witness 4
to ignore this lock.
.It Dv RM_RECURSE
Allow threads to recursively acquire shared locks for
.Fa rm .
.It Dv RM_SLEEPABLE
Create a sleepable read-mostly lock.
.El
.It Fn rm_rlock "struct rmlock *rm" "struct rm_priotracker* tracker"
Lock
.Fa rm
as a reader using
.Fa tracker
to track read owners of a lock for priority propagation.
This data structure is only used internally by
.Nm
and must persist until
.Fn rm_runlock
has been called.
This data structure can be allocated on the stack since
readers cannot sleep.
If any thread holds this lock exclusively, the current thread blocks,
and its priority is propagated to the exclusive holder.
If the lock was initialized with the
.Dv RM_RECURSE
option the
.Fn rm_rlock
function can be called when the current thread has already acquired reader
access on
.Fa rm .
.It Fn rm_try_rlock "struct rmlock *rm" "struct rm_priotracker* tracker"
Try to lock
.Fa rm
as a reader.
.Fn rm_try_rlock
will return 0 if the lock cannot be acquired immediately;
otherwise,
the lock will be acquired and a non-zero value will be returned.
Note that
.Fn rm_try_rlock
may fail even while the lock is not currently held by a writer.
If the lock was initialized with the
.Dv RM_RECURSE
option,
.Fn rm_try_rlock
will succeed if the current thread has already acquired reader access.
.It Fn rm_wlock "struct rmlock *rm"
Lock
.Fa rm
as a writer.
If there are any shared owners of the lock, the current thread blocks.
The
.Fn rm_wlock
function cannot be called recursively.
.It Fn rm_runlock "struct rmlock *rm" "struct rm_priotracker* tracker"
This function releases a shared lock previously acquired by
.Fn rm_rlock .
The
.Fa tracker
argument must match the
.Fa tracker
argument used for acquiring the shared lock
.It Fn rm_wunlock "struct rmlock *rm"
This function releases an exclusive lock previously acquired by
.Fn rm_wlock .
.It Fn rm_destroy "struct rmlock *rm"
This functions destroys a lock previously initialized with
.Fn rm_init .
The
.Fa rm
lock must be unlocked.
.It Fn rm_wowned "const struct rmlock *rm"
This function returns a non-zero value if the current thread owns an
exclusive lock on
.Fa rm .
.It Fn rm_sleep "void *wchan" "struct rmlock *rm" "int priority" "const char *wmesg" "int timo"
This function atomically releases
.Fa rm
while waiting for an event.
The
.Fa rm
lock must be exclusively locked.
For more details on the parameters to this function,
see
.Xr sleep 9 .
.It Fn rm_assert "struct rmlock *rm" "int what"
This function asserts that the
.Fa rm
lock is in the state specified by
.Fa what .
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 rm .
.It Dv RA_RLOCKED
Assert that current thread holds a shared lock of
.Fa rm .
.It Dv RA_WLOCKED
Assert that current thread holds an exclusive lock of
.Fa rm .
.It Dv RA_UNLOCKED
Assert that current thread holds neither a shared nor exclusive lock of
.Fa rm .
.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 rm .
.It Dv RA_NOTRECURSED
Assert that the current thread does not hold a recursive lock of
.Fa rm .
.El
.El
.Sh SEE ALSO
.Xr locking 9 ,
.Xr mutex 9 ,
.Xr panic 9 ,
.Xr rwlock 9 ,
.Xr sleep 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 "Stephan Uphoff" .
This manual page was written by
.An "Gleb Smirnoff"
for rwlock and modified to reflect rmlock by
.An "Stephan Uphoff" .
.Sh BUGS
The
.Nm
implementation is currently not optimized for single processor systems.
.Pp
.Fn rm_try_rlock
can fail transiently even when there is no writer, while another reader
updates the state on the local CPU.
.Pp
The
.Nm
implementation uses a single per CPU list shared by all
rmlocks in the system.
If rmlocks become popular, hashing to multiple per CPU queues may
be needed to speed up the writer lock process.