SLEEP(9) BSD Kernel Developer's Manual SLEEP(9)
msleep, msleep_sbt, msleep_spin, msleep_spin_sbt, pause, pause_sbt, tsleep, tsleep_sbt, wakeup -- wait for events
msleep(void *chan, struct mtx *mtx, int priority, const char *wmesg, int timo);
msleep_sbt(void *chan, struct mtx *mtx, int priority, const char *wmesg, sbintime_t sbt, sbintime_t pr, int flags);
msleep_spin(void *chan, struct mtx *mtx, const char *wmesg, int timo);
msleep_spin_sbt(void *chan, struct mtx *mtx, const char *wmesg, sbintime_t sbt, sbintime_t pr, int flags);
pause(const char *wmesg, int timo);
pause_sbt(const char *wmesg, sbintime_t sbt, sbintime_t pr, int flags);
tsleep(void *chan, int priority, const char *wmesg, int timo);
tsleep_sbt(void *chan, int priority, const char *wmesg, sbintime_t sbt, sbintime_t pr, int flags);
The functions tsleep(), msleep(), msleep_spin(), pause(), wakeup(), and wakeup_one() handle event-based thread blocking. If a thread must
wait for an external event, it is put to sleep by tsleep(), msleep(), msleep_spin(), or pause(). Threads may also wait using one of the
locking primitive sleep routines mtx_sleep(9), rw_sleep(9), or sx_sleep(9).
The parameter chan is an arbitrary address that uniquely identifies the event on which the thread is being put to sleep. All threads sleep-
ing on a single chan are woken up later by wakeup(), often called from inside an interrupt routine, to indicate that the resource the thread
was blocking on is available now.
The parameter priority specifies a new priority for the thread as well as some optional flags. If the new priority is not 0, then the thread
will be made runnable with the specified priority when it resumes. PZERO should never be used, as it is for compatibility only. A new pri-
ority of 0 means to use the thread's current priority when it is made runnable again.
If priority includes the PCATCH flag, pending signals are allowed to interrupt the sleep, otherwise pending signals are ignored during the
sleep. If PCATCH is set and a signal becomes pending, ERESTART is returned if the current system call should be restarted if possible, and
EINTR is returned if the system call should be interrupted by the signal (return EINTR). If the PBDRY flag is specified in addition to
PCATCH, then the sleeping thread is not stopped when SIGSTOP becomes pending or some other stop action occurs while it is sleeping. Instead,
it is woken up, with the assumption that the stop will occur on reaching a stop point when returning to usermode. The flag should be used
when the sleeping thread owns resources, for instance vnode locks, that should be released in a timely fashion.
The parameter wmesg is a string describing the sleep condition for tools like ps(1). Due to the limited space of those programs to display
arbitrary strings, this message should not be longer than 6 characters.
The parameter timo specifies a timeout for the sleep. If timo is not 0, then the thread will sleep for at most timo / hz seconds. If the
timeout expires, then the sleep function will return EWOULDBLOCK.
msleep_sbt(), msleep_spin_sbt(), pause_sbt() and tsleep_sbt() functions take sbt parameter instead of timo. It allows the caller to specify
relative or absolute wakeup time with higher resolution in form of sbintime_t. The parameter pr allows the caller to specify wanted absolute
event precision. The parameter flags allows the caller to pass additional callout_reset_sbt() flags.
Several of the sleep functions including msleep(), msleep_spin(), and the locking primitive sleep routines specify an additional lock parame-
ter. The lock will be released before sleeping and reacquired before the sleep routine returns. If priority includes the PDROP flag, then
the lock will not be reacquired before returning. The lock is used to ensure that a condition can be checked atomically, and that the cur-
rent thread can be suspended without missing a change to the condition, or an associated wakeup. In addition, all of the sleep routines will
fully drop the Giant mutex (even if recursed) while the thread is suspended and will reacquire the Giant mutex before the function returns.
Note that the Giant mutex may be specified as the lock to drop. In that case, however, the PDROP flag is not allowed.
To avoid lost wakeups, either a lock should be used to protect against races, or a timeout should be specified to place an upper bound on the
delay due to a lost wakeup. As a result, the tsleep() function should only be invoked with a timeout of 0 when the Giant mutex is held.
The msleep() function requires that mtx reference a default, i.e. non-spin, mutex. Its use is deprecated in favor of mtx_sleep(9) which pro-
vides identical behavior.
The msleep_spin() function requires that mtx reference a spin mutex. The msleep_spin() function does not accept a priority parameter and
thus does not support changing the current thread's priority, the PDROP flag, or catching signals via the PCATCH flag.
The pause() function is a wrapper around tsleep() that suspends execution of the current thread for the indicated timeout. The thread can
not be awakened early by signals or calls to wakeup() or wakeup_one().
The wakeup_one() function makes the first thread in the queue that is sleeping on the parameter chan runnable. This reduces the load when a
large number of threads are sleeping on the same address, but only one of them can actually do any useful work when made runnable.
Due to the way it works, the wakeup_one() function requires that only related threads sleep on a specific chan address. It is the program-
mer's responsibility to choose a unique chan value. The older wakeup() function did not require this, though it was never good practice for
threads to share a chan value. When converting from wakeup() to wakeup_one(), pay particular attention to ensure that no other threads wait
on the same chan.
When awakened by a call to wakeup() or wakeup_one(), if a signal is pending and PCATCH is specified, a non-zero error code is returned. If
the thread is awakened by a call to wakeup() or wakeup_one(), the msleep(), msleep_spin(), tsleep(), and locking primitive sleep functions
return 0. Otherwise, a non-zero error code is returned.
msleep(), msleep_spin(), tsleep(), and the locking primitive sleep functions will fail if:
[EINTR] The PCATCH flag was specified, a signal was caught, and the system call should be interrupted.
[ERESTART] The PCATCH flag was specified, a signal was caught, and the system call should be restarted.
[EWOULDBLOCK] A non-zero timeout was specified and the timeout expired.
ps(1), locking(9), malloc(9), mi_switch(9), mtx_sleep(9), rw_sleep(9), sx_sleep(9), timeout(9)
The functions sleep() and wakeup() were present in Version 1 AT&T UNIX. They were probably also present in the preceding PDP-7 version of
UNIX. They were the basic process synchronization model.
The tsleep() function appeared in 4.4BSD and added the parameters wmesg and timo. The sleep() function was removed in FreeBSD 2.2. The
wakeup_one() function appeared in FreeBSD 2.2. The msleep() function appeared in FreeBSD 5.0, and the msleep_spin() function appeared in
FreeBSD 6.2. The pause() function appeared in FreeBSD 7.0.
This manual page was written by Jorg Wunsch <joerg@FreeBSD.org>.
February 19, 2013 BSD