CALLOUT(9) BSD Kernel Developer's Manual CALLOUT(9)NAME
callout_init, callout_destroy, callout_halt, callout_reset,
callout_schedule, callout_setfunc, callout_stop, callout_pending,
callout_expired, callout_invoking, callout_ack — execute a function after
a specified length of time
SYNOPSIS
#include <sys/callout.h>
void
callout_init(callout_t *c, u_int flags);
void
callout_destroy(callout_t *c);
void
callout_reset(callout_t *c, int ticks, void (*func)(void *), void *arg);
void
callout_schedule(callout_t *c, int ticks);
void
callout_setfunc(callout_t *c, void (*func)(void *), void *arg);
bool
callout_stop(callout_t *c);
bool
callout_halt(callout_t *c, kmutex_t *interlock);
bool
callout_pending(callout_t *c);
bool
callout_expired(callout_t *c);
bool
callout_active(callout_t *c);
bool
callout_invoking(callout_t *c);
bool
callout_ack(callout_t *c);
DESCRIPTION
The callout facility provides a mechanism to execute a function at a
given time. The timer is based on the hardclock timer which ticks hz
times per second. The function is called at softclock interrupt level.
Clients of the callout facility are responsible for providing pre-allo‐
cated callout structures, or “handles”. The callout facility replaces
the historic UNIX functions timeout() and untimeout().
FUNCTIONS
The callout_init() function initializes the callout handle c for use. No
operations can be performed on the callout before it is initialized. If
the flags argument is CALLOUT_MPSAFE, the handler will be called without
getting the global kernel lock. In this case it should only use func‐
tions that are multiprocessor safe.
callout_destroy() destroys the callout, preventing further use. It is
provided as a diagnostic facility intended to catch bugs. To ensure
future compatibility, callout_destroy() should always be called when the
callout is no longer required (for instance, when a device is being
detached).
The callout_reset() function resets and starts the timer associated with
the callout handle c. When the timer expires after ticks/hz seconds, the
function specified by func will be called with the argument arg. If the
timer associated with the callout handle is already running, the callout
will simply be rescheduled to execute at the newly specified time. Once
the timer is started, the callout handle is marked as PENDING. Once the
timer expires, the handle is marked as EXPIRED and INVOKING, and the
PENDING status is cleared.
The callout_setfunc() function sets the function and argument of the
callout handle c to func and arg respectively. The callout handle must
already be initialized. If a callout will always be used with the same
function and argument, then callout_setfunc() used in conjunction with
callout_schedule() is slightly more efficient than using callout_reset().
The callout_stop() function requests that the timer associated with the
callout handle c be stopped. The PENDING and EXPIRED status for the
callout handle is cleared. It is safe to call callout_stop() on a call‐
out handle that is not pending, so long as it is initialized.
callout_stop() will return a non-zero value if the callout was EXPIRED.
Note that callout_stop() can return while the callout is running on a
different CPU or at a different interrupt priority level on the current
CPU. It can only be said to prevent the callout from firing in the
future, unless explicitly re-scheduled. To stop a callout and wait for
completion, use callout_halt().
callout_halt() acts much like callout_stop(), but waits for the callout
to complete if it is currently in-flight. callout_halt() may not be
called from a hard interrupt handler as it will sleep if the callout is
currently executing. If the callout can take locks (such as mutexes or
RW locks), the caller of callout_halt() must not hold any of those locks,
otherwise the two could deadlock. To facilitate this, callout_halt() can
optionally release a single mutex specified by the interlock parameter.
If interlock is not NULL and the calling thread must wait for the callout
to complete, interlock will be released before waiting and re-acquired
before returning. If no wait is required, interlock will not be
released. However, to avoid race conditions the caller should always
assume that interlock has been released and reacquired, and act accord‐
ingly.
The callout_pending() function tests the PENDING status of the callout
handle c. A PENDING callout is one that has been started and whose func‐
tion has not yet been called. Note that it is possible for a callout's
timer to have expired without its function being called if interrupt
level has not dropped low enough to let softclock interrupts through.
Note that it is only safe to test PENDING status when at softclock inter‐
rupt level or higher.
The callout_expired() function tests to see if the callout's timer has
expired and its function called.
The callout_active() function returns true if a timer has been started
but not explicitly stopped, even if it has already fired.
callout_active(foo) is logically the same as callout_pending(foo) ||
callout_expired(foo); it is implemented as a separate function for com‐
patibility with FreeBSD and for the special case of TCP_TIMER_ISARMED().
Its use is not recommended.
The callout_invoking() function tests the INVOKING status of the callout
handle c. This flag is set just before a callout's function is being
called. Since the priority level is lowered prior to invocation of the
callout function, other pending higher-priority code may run before the
callout function is allowed to run. This may create a race condition if
this higher-priority code deallocates storage containing one or more
callout structures whose callout functions are about to be run. In such
cases, one technique to prevent references to deallocated storage would
be to test whether any callout functions are in the INVOKING state using
callout_invoking(), and if so, to mark the data structure and defer stor‐
age deallocation until the callout function is allowed to run. For this
handshake protocol to work, the callout function will have to use the
callout_ack() function to clear this flag.
The callout_ack() function clears the INVOKING state in the callout han‐
dle c. This is used in situations where it is necessary to protect
against the race condition described under callout_invoking().
CONCURRENCY
The callout facility performs locking internally in order to guarantee
the atomicity of individual operations performed on callouts. It does
not provide life cycle management of user-provided callout data struc‐
tures, nor does it ensure that groups of operations (multiple function
calls) are performed atomically. These aspects of callout management are
the responsibility of the user of the callout facility.
Scheduled callouts may be active concurrently in a context different to
the user of the callout facility: on another CPU, or at a different
interrupt priority level or thread on the current CPU. The callout
facility provides only one guarantee in this regard: any given callout
will never have multiple concurrent invocations.
SEE ALSOcondvar(9), hz(9), softint(9), workqueue(9)HISTORY
The callout facility was implemented by Artur Grabowski and Thomas
Nordin, based on the work of G. Varghese and A. Lauck, described in the
paper Hashed and Hierarchical Timing Wheels: Data Structures for the
Efficient Implementation of a Timer Facility in the Proceedings of the
11th ACM Annual Symposium on Operating System Principles, Austin, Texas,
November 1987. It was adapted to the NetBSD kernel by Jason R. Thorpe.
BSD August 3, 2009 BSD
