freebsd-dev/sys/kern/kern_condvar.c
Hans Petter Selasky 1a26c3c047 Major callout subsystem cleanup and rewrite:
- Close a migration race where callout_reset() failed to set the
  CALLOUT_ACTIVE flag.
- Callout callback functions are now allowed to be protected by
  spinlocks.
- Switching the callout CPU number cannot always be done on a
  per-callout basis. See the updated timeout(9) manual page for more
  information.
- The timeout(9) manual page has been updated to reflect how all the
  functions inside the callout API are working. The manual page has
  been made function oriented to make it easier to deduce how each of
  the functions making up the callout API are working without having
  to first read the whole manual page. Group all functions into a
  handful of sections which should give a quick top-level overview
  when the different functions should be used.
- The CALLOUT_SHAREDLOCK flag and its functionality has been removed
  to reduce the complexity in the callout code and to avoid problems
  about atomically stopping callouts via callout_stop(). If someone
  needs it, it can be re-added. From my quick grep there are no
  CALLOUT_SHAREDLOCK clients in the kernel.
- A new callout API function named "callout_drain_async()" has been
  added. See the updated timeout(9) manual page for a complete
  description.
- Update the callout clients in the "kern/" folder to use the callout
  API properly, like cv_timedwait(). Previously there was some custom
  sleepqueue code in the callout subsystem, which has been removed,
  because we now allow callouts to be protected by spinlocks. This
  allows us to tear down the callout like done with regular mutexes,
  and a "td_slpmutex" has been added to "struct thread" to atomically
  teardown the "td_slpcallout". Further the "TDF_TIMOFAIL" and
  "SWT_SLEEPQTIMO" states can now be completely removed. Currently
  they are marked as available and will be cleaned up in a follow up
  commit.
- Bump the __FreeBSD_version to indicate kernel modules need
  recompilation.
- There has been several reports that this patch "seems to squash a
  serious bug leading to a callout timeout and panic".

Kernel build testing:	all architectures were built
MFC after:		2 weeks
Differential Revision:	https://reviews.freebsd.org/D1438
Sponsored by:		Mellanox Technologies
Reviewed by:		jhb, adrian, sbruno and emaste
2015-01-15 15:32:30 +00:00

454 lines
12 KiB
C

/*-
* Copyright (c) 2000 Jake Burkholder <jake@freebsd.org>.
* 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 AUTHOR AND CONTRIBUTORS ``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 AUTHOR OR CONTRIBUTORS 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.
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include "opt_ktrace.h"
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/lock.h>
#include <sys/mutex.h>
#include <sys/proc.h>
#include <sys/kernel.h>
#include <sys/ktr.h>
#include <sys/condvar.h>
#include <sys/sched.h>
#include <sys/signalvar.h>
#include <sys/sleepqueue.h>
#include <sys/resourcevar.h>
#ifdef KTRACE
#include <sys/uio.h>
#include <sys/ktrace.h>
#endif
/*
* Common sanity checks for cv_wait* functions.
*/
#define CV_ASSERT(cvp, lock, td) do { \
KASSERT((td) != NULL, ("%s: td NULL", __func__)); \
KASSERT(TD_IS_RUNNING(td), ("%s: not TDS_RUNNING", __func__)); \
KASSERT((cvp) != NULL, ("%s: cvp NULL", __func__)); \
KASSERT((lock) != NULL, ("%s: lock NULL", __func__)); \
} while (0)
/*
* Initialize a condition variable. Must be called before use.
*/
void
cv_init(struct cv *cvp, const char *desc)
{
cvp->cv_description = desc;
cvp->cv_waiters = 0;
}
/*
* Destroy a condition variable. The condition variable must be re-initialized
* in order to be re-used.
*/
void
cv_destroy(struct cv *cvp)
{
#ifdef INVARIANTS
struct sleepqueue *sq;
sleepq_lock(cvp);
sq = sleepq_lookup(cvp);
sleepq_release(cvp);
KASSERT(sq == NULL, ("%s: associated sleep queue non-empty", __func__));
#endif
}
/*
* Wait on a condition variable. The current thread is placed on the condition
* variable's wait queue and suspended. A cv_signal or cv_broadcast on the same
* condition variable will resume the thread. The mutex is released before
* sleeping and will be held on return. It is recommended that the mutex be
* held when cv_signal or cv_broadcast are called.
*/
void
_cv_wait(struct cv *cvp, struct lock_object *lock)
{
WITNESS_SAVE_DECL(lock_witness);
struct lock_class *class;
struct thread *td;
uintptr_t lock_state;
td = curthread;
lock_state = 0;
#ifdef KTRACE
if (KTRPOINT(td, KTR_CSW))
ktrcsw(1, 0, cv_wmesg(cvp));
#endif
CV_ASSERT(cvp, lock, td);
WITNESS_WARN(WARN_GIANTOK | WARN_SLEEPOK, lock,
"Waiting on \"%s\"", cvp->cv_description);
class = LOCK_CLASS(lock);
if (cold || panicstr) {
/*
* During autoconfiguration, just give interrupts
* a chance, then just return. Don't run any other
* thread or panic below, in case this is the idle
* process and already asleep.
*/
return;
}
sleepq_lock(cvp);
cvp->cv_waiters++;
if (lock == &Giant.lock_object)
mtx_assert(&Giant, MA_OWNED);
DROP_GIANT();
sleepq_add(cvp, lock, cvp->cv_description, SLEEPQ_CONDVAR, 0);
if (lock != &Giant.lock_object) {
if (class->lc_flags & LC_SLEEPABLE)
sleepq_release(cvp);
WITNESS_SAVE(lock, lock_witness);
lock_state = class->lc_unlock(lock);
if (class->lc_flags & LC_SLEEPABLE)
sleepq_lock(cvp);
}
sleepq_wait(cvp, 0);
#ifdef KTRACE
if (KTRPOINT(td, KTR_CSW))
ktrcsw(0, 0, cv_wmesg(cvp));
#endif
PICKUP_GIANT();
if (lock != &Giant.lock_object) {
class->lc_lock(lock, lock_state);
WITNESS_RESTORE(lock, lock_witness);
}
}
/*
* Wait on a condition variable. This function differs from cv_wait by
* not aquiring the mutex after condition variable was signaled.
*/
void
_cv_wait_unlock(struct cv *cvp, struct lock_object *lock)
{
struct lock_class *class;
struct thread *td;
td = curthread;
#ifdef KTRACE
if (KTRPOINT(td, KTR_CSW))
ktrcsw(1, 0, cv_wmesg(cvp));
#endif
CV_ASSERT(cvp, lock, td);
WITNESS_WARN(WARN_GIANTOK | WARN_SLEEPOK, lock,
"Waiting on \"%s\"", cvp->cv_description);
KASSERT(lock != &Giant.lock_object,
("cv_wait_unlock cannot be used with Giant"));
class = LOCK_CLASS(lock);
if (cold || panicstr) {
/*
* During autoconfiguration, just give interrupts
* a chance, then just return. Don't run any other
* thread or panic below, in case this is the idle
* process and already asleep.
*/
class->lc_unlock(lock);
return;
}
sleepq_lock(cvp);
cvp->cv_waiters++;
DROP_GIANT();
sleepq_add(cvp, lock, cvp->cv_description, SLEEPQ_CONDVAR, 0);
if (class->lc_flags & LC_SLEEPABLE)
sleepq_release(cvp);
class->lc_unlock(lock);
if (class->lc_flags & LC_SLEEPABLE)
sleepq_lock(cvp);
sleepq_wait(cvp, 0);
#ifdef KTRACE
if (KTRPOINT(td, KTR_CSW))
ktrcsw(0, 0, cv_wmesg(cvp));
#endif
PICKUP_GIANT();
}
/*
* Wait on a condition variable, allowing interruption by signals. Return 0 if
* the thread was resumed with cv_signal or cv_broadcast, EINTR or ERESTART if
* a signal was caught. If ERESTART is returned the system call should be
* restarted if possible.
*/
int
_cv_wait_sig(struct cv *cvp, struct lock_object *lock)
{
WITNESS_SAVE_DECL(lock_witness);
struct lock_class *class;
struct thread *td;
uintptr_t lock_state;
int rval;
td = curthread;
lock_state = 0;
#ifdef KTRACE
if (KTRPOINT(td, KTR_CSW))
ktrcsw(1, 0, cv_wmesg(cvp));
#endif
CV_ASSERT(cvp, lock, td);
WITNESS_WARN(WARN_GIANTOK | WARN_SLEEPOK, lock,
"Waiting on \"%s\"", cvp->cv_description);
class = LOCK_CLASS(lock);
if (cold || panicstr) {
/*
* After a panic, or during autoconfiguration, just give
* interrupts a chance, then just return; don't run any other
* procs or panic below, in case this is the idle process and
* already asleep.
*/
return (0);
}
sleepq_lock(cvp);
cvp->cv_waiters++;
if (lock == &Giant.lock_object)
mtx_assert(&Giant, MA_OWNED);
DROP_GIANT();
sleepq_add(cvp, lock, cvp->cv_description, SLEEPQ_CONDVAR |
SLEEPQ_INTERRUPTIBLE, 0);
if (lock != &Giant.lock_object) {
if (class->lc_flags & LC_SLEEPABLE)
sleepq_release(cvp);
WITNESS_SAVE(lock, lock_witness);
lock_state = class->lc_unlock(lock);
if (class->lc_flags & LC_SLEEPABLE)
sleepq_lock(cvp);
}
rval = sleepq_wait_sig(cvp, 0);
#ifdef KTRACE
if (KTRPOINT(td, KTR_CSW))
ktrcsw(0, 0, cv_wmesg(cvp));
#endif
PICKUP_GIANT();
if (lock != &Giant.lock_object) {
class->lc_lock(lock, lock_state);
WITNESS_RESTORE(lock, lock_witness);
}
return (rval);
}
/*
* Wait on a condition variable for (at most) the value specified in sbt
* argument. Returns 0 if the process was resumed by cv_signal or cv_broadcast,
* EWOULDBLOCK if the timeout expires.
*/
int
_cv_timedwait_sbt(struct cv *cvp, struct lock_object *lock, sbintime_t sbt,
sbintime_t pr, int flags)
{
WITNESS_SAVE_DECL(lock_witness);
struct lock_class *class;
struct thread *td;
int lock_state, rval;
td = curthread;
lock_state = 0;
#ifdef KTRACE
if (KTRPOINT(td, KTR_CSW))
ktrcsw(1, 0, cv_wmesg(cvp));
#endif
CV_ASSERT(cvp, lock, td);
WITNESS_WARN(WARN_GIANTOK | WARN_SLEEPOK, lock,
"Waiting on \"%s\"", cvp->cv_description);
class = LOCK_CLASS(lock);
if (cold || panicstr) {
/*
* After a panic, or during autoconfiguration, just give
* interrupts a chance, then just return; don't run any other
* thread or panic below, in case this is the idle process and
* already asleep.
*/
return 0;
}
sleepq_lock(cvp);
cvp->cv_waiters++;
if (lock == &Giant.lock_object)
mtx_assert(&Giant, MA_OWNED);
DROP_GIANT();
sleepq_add(cvp, lock, cvp->cv_description, SLEEPQ_CONDVAR, 0);
sleepq_release(cvp);
sleepq_set_timeout_sbt(cvp, sbt, pr, flags);
if (lock != &Giant.lock_object) {
WITNESS_SAVE(lock, lock_witness);
lock_state = class->lc_unlock(lock);
}
sleepq_lock(cvp);
rval = sleepq_timedwait(cvp, 0);
#ifdef KTRACE
if (KTRPOINT(td, KTR_CSW))
ktrcsw(0, 0, cv_wmesg(cvp));
#endif
PICKUP_GIANT();
if (lock != &Giant.lock_object) {
class->lc_lock(lock, lock_state);
WITNESS_RESTORE(lock, lock_witness);
}
return (rval);
}
/*
* Wait on a condition variable for (at most) the value specified in sbt
* argument, allowing interruption by signals.
* Returns 0 if the thread was resumed by cv_signal or cv_broadcast,
* EWOULDBLOCK if the timeout expires, and EINTR or ERESTART if a signal
* was caught.
*/
int
_cv_timedwait_sig_sbt(struct cv *cvp, struct lock_object *lock,
sbintime_t sbt, sbintime_t pr, int flags)
{
WITNESS_SAVE_DECL(lock_witness);
struct lock_class *class;
struct thread *td;
int lock_state, rval;
td = curthread;
lock_state = 0;
#ifdef KTRACE
if (KTRPOINT(td, KTR_CSW))
ktrcsw(1, 0, cv_wmesg(cvp));
#endif
CV_ASSERT(cvp, lock, td);
WITNESS_WARN(WARN_GIANTOK | WARN_SLEEPOK, lock,
"Waiting on \"%s\"", cvp->cv_description);
class = LOCK_CLASS(lock);
if (cold || panicstr) {
/*
* After a panic, or during autoconfiguration, just give
* interrupts a chance, then just return; don't run any other
* thread or panic below, in case this is the idle process and
* already asleep.
*/
return 0;
}
sleepq_lock(cvp);
cvp->cv_waiters++;
if (lock == &Giant.lock_object)
mtx_assert(&Giant, MA_OWNED);
DROP_GIANT();
sleepq_add(cvp, lock, cvp->cv_description, SLEEPQ_CONDVAR |
SLEEPQ_INTERRUPTIBLE, 0);
sleepq_release(cvp);
sleepq_set_timeout_sbt(cvp, sbt, pr, flags);
if (lock != &Giant.lock_object) {
WITNESS_SAVE(lock, lock_witness);
lock_state = class->lc_unlock(lock);
}
sleepq_lock(cvp);
rval = sleepq_timedwait_sig(cvp, 0);
#ifdef KTRACE
if (KTRPOINT(td, KTR_CSW))
ktrcsw(0, 0, cv_wmesg(cvp));
#endif
PICKUP_GIANT();
if (lock != &Giant.lock_object) {
class->lc_lock(lock, lock_state);
WITNESS_RESTORE(lock, lock_witness);
}
return (rval);
}
/*
* Signal a condition variable, wakes up one waiting thread. Will also wakeup
* the swapper if the process is not in memory, so that it can bring the
* sleeping process in. Note that this may also result in additional threads
* being made runnable. Should be called with the same mutex as was passed to
* cv_wait held.
*/
void
cv_signal(struct cv *cvp)
{
int wakeup_swapper;
wakeup_swapper = 0;
sleepq_lock(cvp);
if (cvp->cv_waiters > 0) {
cvp->cv_waiters--;
wakeup_swapper = sleepq_signal(cvp, SLEEPQ_CONDVAR, 0, 0);
}
sleepq_release(cvp);
if (wakeup_swapper)
kick_proc0();
}
/*
* Broadcast a signal to a condition variable. Wakes up all waiting threads.
* Should be called with the same mutex as was passed to cv_wait held.
*/
void
cv_broadcastpri(struct cv *cvp, int pri)
{
int wakeup_swapper;
/*
* XXX sleepq_broadcast pri argument changed from -1 meaning
* no pri to 0 meaning no pri.
*/
wakeup_swapper = 0;
if (pri == -1)
pri = 0;
sleepq_lock(cvp);
if (cvp->cv_waiters > 0) {
cvp->cv_waiters = 0;
wakeup_swapper = sleepq_broadcast(cvp, SLEEPQ_CONDVAR, pri, 0);
}
sleepq_release(cvp);
if (wakeup_swapper)
kick_proc0();
}