.\" .\" Copyright (c) 1996 Joerg Wunsch .\" .\" 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 DEVELOPERS ``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 DEVELOPERS 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: src/share/man/man9/sleep.9,v 1.18.2.5 2001/12/17 11:30:19 ru Exp $ .\" $DragonFly: src/share/man/man9/sleep.9,v 1.2 2003/06/17 04:37:01 dillon Exp $ .\" " .Dd December 17, 1998 .Os .Dt SLEEP 9 .Sh NAME .Nm sleep , .Nm tsleep , .Nm asleep , .Nm await , .Nm wakeup .Nd wait for events .Sh SYNOPSIS .In sys/param.h .In sys/systm.h .In sys/proc.h .Ft int .Fn tsleep "void *ident" "int priority" "const char *wmesg" "int timo" .Ft int .Fn asleep "void *ident" "int priority" "const char *wmesg" "int timo" .Ft int .Fn await "int priority" "int timo" .Ft void .Fn wakeup "void *ident" .Ft void .Fn wakeup_one "void *ident" .Sh DESCRIPTION The functions .Fn tsleep and .Fn wakeup handle event-based process blocking. If a process must wait for an external event, it is put on sleep by .Nm tsleep . The parameter .Ar ident is an arbitrary address that uniquely identifies the event on which the process is being asleep. All processes sleeping on a single .Ar ident are woken up later by .Nm wakeup , often called from inside an interrupt routine, to indicate that the resource the process was blocking on is available now. .Pp The parameter .Ar wmesg is a string describing the sleep condition for tools like .Xr ps 1 . Due to the limited space of those programs to display arbitrary strings, this message should not be longer than 6 characters. .Pp The .Fn wakeup_one function is used to make the first process in the queue that is sleeping on the parameter .Fa ident runnable. This can prevent the system from becoming saturated when a large number of processes are sleeping on the same address, but only one of them can actually do any useful work when made runnable. .Pp .Nm Tsleep is the general sleep call. Suspends the current process until a wakeup is performed on the specified identifier. The process will then be made runnable with the specified .Ar priority . Sleeps at most .Ar timo \&/ hz seconds (0 means no timeout). If .Ar pri includes the .Dv PCATCH flag, signals are checked before and after sleeping, else signals are not checked. Returns 0 if awakened, .Er EWOULDBLOCK if the timeout expires. If .Dv PCATCH is set and a signal needs to be delivered, .Er ERESTART is returned if the current system call should be restarted if possible, and .Er EINTR is returned if the system call should be interrupted by the signal (return .Er EINTR ) . .Pp .Nm Asleep implements the new asynchronous sleep function. It takes the same arguments as .Fn tsleep and places the process on the appropriate wait queue, but .Fn asleep leaves the process runnable and returns immediately. The caller is then expected to, at some point in the future, call .Fn await to actually wait for the previously queued wait condition. If .Fn asleep is called several times, only the most recent call is effective. .Fn asleep may be called with an .Ar ident value of NULL to remove any previously queued condition. .Pp .Nm Await implements the new asynchronous wait function. When .Fn asleep is called on an identifier it associates the process with that identifier but does not block. .Fn await will actually block the process until .Fn wakeup is called on that identifier any time after the .Fn asleep . If .Fn wakeup is called after you .Fn asleep but before you .Fn await then the .Fn await call is effectively a NOP. If .Fn await is called multiple times without an intervening .Fn asleep , the .Fn await is effectively a NOP but will also call .Fn mswitch for safety. The .Fn await function allows you to override the priority and timeout values to be used. If the value -1 is specified for an argument, the value is taken from the previous .Fn asleep call. If -1 is passed for the priority you must be prepared to catch signal conditions if the prior call to .Fn asleep specified it in its priority. If -1 is passed for the timeout you must be prepared to catch a timeout condition if the prior call to .Fn asleep specified a timeout. When you use -1, it is usually a good idea to not make assumptions as to the arguments used by the prior .Fn asleep call. .Pp The .Fn asleep and .Fn await functions are mainly used by the kernel to shift the burden of blocking away from extremely low level routines and to push it onto their callers. This in turn allows more complex interlocking code to .Em backout of a temporary resource failure (such as lack of memory) in order to release major locks prior to actually blocking, and to then retry the operation on wakeup. This key feature is expected to be heavily used in SMP situations in order to allow code to make better use of spinlocks. A spinlock, by its very nature, cannot be used around code that might block. It is hoped that these capabilities will make it easier to migrate the SMP master locks deeper into the kernel. .Pp These routines may also be used to avoid nasty spl*() calls to get around race conditions with simple conditional test/wait interlocks. You simply call .Fn asleep prior to your test, then conditionally .Fn await only if the test fails. It is usually a good idea to cancel an .Fn asleep if you wind up never calling the related .Fn await , but it is not required. If you do not want to waste cpu calling .Fn asleep unnecessarily, you can surround the whole thing with a second test. The race condition is still handled by the inside .Fn asleep call. .Sh RETURN VALUES See above. .Sh SEE ALSO .Xr ps 1 , .Xr malloc 9 .Sh HISTORY The sleep/wakeup process synchronization mechanism is very old. It appeared in a very early version of Unix. .Pp .Nm Tsleep appeared in .Bx 4.4 . .Pp .Nm Asleep Ns / Ns Nm await first appeared in .Fx 3.0 and is designed to shift the burden of blocking away from extremely low level routines and push it up to their callers. .Pp .Nm Sleep used to be the traditional form. It doesn't let you specify a timeout or a .Ar wmesg , hence it has been discontinued. .Sh AUTHORS .An -nosplit This man page was written by .An J\(:org Wunsch . .Nm Asleep and .Nm await were designed and written by .An Matthew Dillon .