1.\" $NetBSD: ltsleep.9,v 1.1 2002/09/29 17:36:17 wiz Exp $ 2.\" 3.\" Copyright (c) 1996, 2002 The NetBSD Foundation, Inc. 4.\" All rights reserved. 5.\" 6.\" This code is derived from software contributed to The NetBSD Foundation 7.\" by Paul Kranenburg. 8.\" 9.\" Redistribution and use in source and binary forms, with or without 10.\" modification, are permitted provided that the following conditions 11.\" are met: 12.\" 1. Redistributions of source code must retain the above copyright 13.\" notice, this list of conditions and the following disclaimer. 14.\" 2. Redistributions in binary form must reproduce the above copyright 15.\" notice, this list of conditions and the following disclaimer in the 16.\" documentation and/or other materials provided with the distribution. 17.\" 3. All advertising materials mentioning features or use of this software 18.\" must display the following acknowledgement: 19.\" This product includes software developed by the NetBSD 20.\" Foundation, Inc. and its contributors. 21.\" 4. Neither the name of The NetBSD Foundation nor the names of its 22.\" contributors may be used to endorse or promote products derived 23.\" from this software without specific prior written permission. 24.\" 25.\" THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 26.\" ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 27.\" TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 28.\" PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 29.\" BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 30.\" CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 31.\" SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 32.\" INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 33.\" CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 34.\" ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 35.\" POSSIBILITY OF SUCH DAMAGE. 36.\" 37.Dd September 29, 2002 38.Dt LTSLEEP 9 39.Os 40.Sh NAME 41.Nm ltsleep , 42.Nm tsleep , 43.Nm wakeup 44.Nd process context sleep and and wakeup 45.Sh SYNOPSIS 46.Fd #include \*[Lt]sys/proc.h\*[Gt] 47.Ft int 48.Fn "ltsleep" "void *ident" "int priority" "const char *wmesg" "int timo" "__volatile struct simplelock *slock" 49.Ft int 50.Fn "tsleep" "void *ident" "int priority" "const char *wmesg" "int timo" 51.Ft void 52.Fn "wakeup" "void *ident" 53.Sh DESCRIPTION 54These functions implement voluntary context switching. 55.Fn ltsleep 56and 57.Fn tsleep 58are used throughout the kernel whenever processing in the current context 59can not continue for any of the following reasons: 60.Bl -bullet -offset indent 61.It 62The current process needs to await the results of a pending I/O operation. 63.It 64The current process needs resources 65.Pq e.g. memory 66which are temporarily unavailable. 67.It 68The current process wants access to data-structures which are locked by 69other processes. 70.El 71.Pp 72The function 73.Fn wakeup 74is used to notify sleeping processes of possible changes to the condition 75that caused them to go to sleep. 76Typically, an awakened process will -- after it has acquired a context 77again -- retry the action that blocked its operation to see if the 78.Dq blocking 79condition has cleared. 80.Pp 81The 82.Fn ltsleep 83function takes the following arguments: 84.Bl -tag -width priority 85.It Fa ident 86An identifier of the 87.Dq wait channel 88representing the resource for which the current process needs to wait. 89This typically is the virtual address of some kernel data-structure related 90to the resource for which the process is contending. 91The same identifier must be used in a call to 92.Fn wakeup 93to get the process going again. 94.Fa ident 95should not be 96.Dv NULL . 97.It Fa priority 98The process priority to be used when the process is awakened and put on 99the queue of runnable processes. 100This mechanism is used to optimize 101.Dq throughput 102of processes executing in kernel mode. 103If the flag 104.Dv PCATCH 105is OR'ed into 106.Fa priority 107the process checks for posted signals before and after sleeping. 108If the flag 109.Dv PNORELOCK 110is OR'ed into 111.Fa priority , 112.Fa slock 113is NOT re-locked after process resume. 114.It Fa wmesg 115A pointer to a character string indicating the reason a process is sleeping. 116The kernel does not use the string, but makes it available 117.Pq through the process structure field Li p_wmesg 118for user level utilities such as 119.Xr ps 1 . 120.It Fa timo 121If non-zero, the process will sleep for at most 122.Li timo/hz 123seconds. 124If this amount of time elapses and no 125.Fn wakeup "ident" 126has occurred, and no signal 127.Pq if Dv PCATCH No was set 128was posted, 129.Fn tsleep 130will return 131.Er EWOULDBLOCK . 132.It Fa slock 133If not NULL, the 134.Fa slock 135interlock is unlocked once the scheduler lock is acquired. Unless 136.Dv PNORELOCK 137was set, 138.Fa slock 139is locked again once 140the process is resumed from sleep. This provides wakeup-before-sleep 141condition protection facility. 142.El 143.Pp 144The 145.Fn tsleep 146macro is functionally equivalent to: 147.Bd -literal -offset indent 148ltsleep(ident, priority, wmesg, timo, NULL) 149.Ed 150.Pp 151The 152.Fn wakeup 153function will mark all processes which are currently sleeping on the identifier 154.Fa ident 155as runnable. 156Eventually, each of the processes will resume execution in the kernel 157context, causing a return from 158.Fn tsleep . 159Note that processes returning from sleep should always re-evaluate the 160conditions that blocked them, since a call to 161.Fn wakeup 162merely signals a 163.Em possible 164change to the blocking conditions. 165For example, when two or more processes are waiting for an exclusive-access 166lock 167.Pq see Xr lock 9 , 168only one of them will succeed in acquiring the lock when it is released. 169All others will have to go back to sleep and wait for the next opportunity. 170.Sh RETURN VALUES 171.Fn ltsleep 172returns 0 if it returns as a result of a 173.Fn wakeup . 174If a 175.Fn ltsleep 176returns as a result of a signal, the return value is 177.Er ERESTART 178if the signal has the 179.Dv SA_RESTART 180property 181.Pq see Xr sigaction 2 , 182and 183.Er EINTR 184otherwise. 185If 186.Fn ltsleep 187returns because of a timeout it returns 188.Er EWOULDBLOCK . 189.Sh SEE ALSO 190.Xr sigaction 2 , 191.Xr hz 9 , 192.Xr lock 9 193.Sh HISTORY 194The sleep/wakeup process synchronization mechanism is very old. It 195appeared in a very early version of Unix. 196.Fn tsleep 197appeared in 198.Bx 4.4 . 199.Fn ltsleep 200appeared in 201.Nx 1.5 . 202