SLEEP(10.2) SLEEP(10.2)
NAME
sleep, wakeup, tsleep, return0 - process synchronisation
SYNOPSIS
void sleep(Rendez *r, int (*f)(void*), void *arg)
void wakeup(Rendez *r)
void tsleep(Rendez *r, int (*f)(void*), void *arg, int ms)
int return0(void *arg)
DESCRIPTION
A process running in the kernel can use these functions to
synchronise with an interrupt handler or another kernel pro-
cess. In particular, they are used by device drivers to
wait for an event to be signalled on receipt of an inter-
rupt. (In practice, they are most often used indirectly,
through qio(10.2) for instance.)
The caller of sleep and a caller of wakeup share a Rendez
structure, to provide a rendezvous point between them to
synchronise on an event. Sleep uses a condition function f
that returns true if the event has occurred.
Sleep evaluates f(arg). If true, the event has happened and
sleep returns immediately. Otherwise, sleep blocks on the
event variable r, awaiting wakeup.
Wakeup is called by either a process or an interrupt handler
to wake any process sleeping at r, signifying that the cor-
responding condition is true (the event has occurred). It
has no effect if there is no sleeping process.
Tsleep is similar to sleep, except that if the condition
f(arg) is false and the caller does sleep, and nothing else
wakes it within ms millliseconds, the system will wake it.
Tsleep's caller must check its environment to decide whether
timeout or the event occurred. The timing provided by
tsleep is imprecise, but adequate in practice for the normal
use of protecting against lost interrupts and otherwise
unresponsive devices or software.
Return0 ignores its arguments and returns zero. It is com-
monly used as the predicate f in a call to tsleep to obtain
a time delay, using a Rendez variable sleep in the Proc
structure, for example:
tsleep(&up->sleep, return0, nil, 10);
Page 1 Plan 9 (printed 12/21/24)
SLEEP(10.2) SLEEP(10.2)
Both sleep and tsleep can be interrupted by swiproc (see
kproc(10.2)), causing a non-local goto through a call to
error(10.2).
DIAGNOSTICS
There can be at most one process waiting on a Rendez, and if
two processes collide, the system will panic(10.2) (``double
sleep''). Access to a Rendez must therefore be serialised
by some other mechanism, usually qlock(10.2).
SOURCE
/os/port/proc.c
/emu/port/proc.c
SEE ALSO
lock(10.2), qlock(10.2), delay(10.2)
``Process Sleep and Wakeup on a Shared-memory Multiproces-
sor'', in Plan 9 Programmer's Manual: Volume 2 .
Page 2 Plan 9 (printed 12/21/24)