freebsd-dev/sys/kern/kern_condvar.c

454 lines
11 KiB
C
Raw Normal View History

2001-01-16 01:00:43 +00:00
/*-
* 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.
*/
2003-06-11 00:56:59 +00:00
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
2001-01-16 01:00:43 +00:00
#include "opt_ktrace.h"
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/lock.h>
#include <sys/mutex.h>
2001-01-16 01:00:43 +00:00
#include <sys/proc.h>
#include <sys/kernel.h>
#include <sys/ktr.h>
#include <sys/condvar.h>
#include <sys/sched.h>
2001-01-16 01:00:43 +00:00
#include <sys/signalvar.h>
Switch the sleep/wakeup and condition variable implementations to use the sleep queue interface: - Sleep queues attempt to merge some of the benefits of both sleep queues and condition variables. Having sleep qeueus in a hash table avoids having to allocate a queue head for each wait channel. Thus, struct cv has shrunk down to just a single char * pointer now. However, the hash table does not hold threads directly, but queue heads. This means that once you have located a queue in the hash bucket, you no longer have to walk the rest of the hash chain looking for threads. Instead, you have a list of all the threads sleeping on that wait channel. - Outside of the sleepq code and the sleep/cv code the kernel no longer differentiates between cv's and sleep/wakeup. For example, calls to abortsleep() and cv_abort() are replaced with a call to sleepq_abort(). Thus, the TDF_CVWAITQ flag is removed. Also, calls to unsleep() and cv_waitq_remove() have been replaced with calls to sleepq_remove(). - The sched_sleep() function no longer accepts a priority argument as sleep's no longer inherently bump the priority. Instead, this is soley a propery of msleep() which explicitly calls sched_prio() before blocking. - The TDF_ONSLEEPQ flag has been dropped as it was never used. The associated TDF_SET_ONSLEEPQ and TDF_CLR_ON_SLEEPQ macros have also been dropped and replaced with a single explicit clearing of td_wchan. TD_SET_ONSLEEPQ() would really have only made sense if it had taken the wait channel and message as arguments anyway. Now that that only happens in one place, a macro would be overkill.
2004-02-27 18:52:44 +00:00
#include <sys/sleepqueue.h>
2001-01-16 01:00:43 +00:00
#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: curthread 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__)); \
2001-01-16 01:00:43 +00:00
} 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;
2001-01-16 01:00:43 +00:00
}
/*
* Destroy a condition variable. The condition variable must be re-initialized
* in order to be re-used.
*/
void
cv_destroy(struct cv *cvp)
{
Switch the sleep/wakeup and condition variable implementations to use the sleep queue interface: - Sleep queues attempt to merge some of the benefits of both sleep queues and condition variables. Having sleep qeueus in a hash table avoids having to allocate a queue head for each wait channel. Thus, struct cv has shrunk down to just a single char * pointer now. However, the hash table does not hold threads directly, but queue heads. This means that once you have located a queue in the hash bucket, you no longer have to walk the rest of the hash chain looking for threads. Instead, you have a list of all the threads sleeping on that wait channel. - Outside of the sleepq code and the sleep/cv code the kernel no longer differentiates between cv's and sleep/wakeup. For example, calls to abortsleep() and cv_abort() are replaced with a call to sleepq_abort(). Thus, the TDF_CVWAITQ flag is removed. Also, calls to unsleep() and cv_waitq_remove() have been replaced with calls to sleepq_remove(). - The sched_sleep() function no longer accepts a priority argument as sleep's no longer inherently bump the priority. Instead, this is soley a propery of msleep() which explicitly calls sched_prio() before blocking. - The TDF_ONSLEEPQ flag has been dropped as it was never used. The associated TDF_SET_ONSLEEPQ and TDF_CLR_ON_SLEEPQ macros have also been dropped and replaced with a single explicit clearing of td_wchan. TD_SET_ONSLEEPQ() would really have only made sense if it had taken the wait channel and message as arguments anyway. Now that that only happens in one place, a macro would be overkill.
2004-02-27 18:52:44 +00:00
#ifdef INVARIANTS
struct sleepqueue *sq;
2001-01-16 01:00:43 +00:00
sleepq_lock(cvp);
Switch the sleep/wakeup and condition variable implementations to use the sleep queue interface: - Sleep queues attempt to merge some of the benefits of both sleep queues and condition variables. Having sleep qeueus in a hash table avoids having to allocate a queue head for each wait channel. Thus, struct cv has shrunk down to just a single char * pointer now. However, the hash table does not hold threads directly, but queue heads. This means that once you have located a queue in the hash bucket, you no longer have to walk the rest of the hash chain looking for threads. Instead, you have a list of all the threads sleeping on that wait channel. - Outside of the sleepq code and the sleep/cv code the kernel no longer differentiates between cv's and sleep/wakeup. For example, calls to abortsleep() and cv_abort() are replaced with a call to sleepq_abort(). Thus, the TDF_CVWAITQ flag is removed. Also, calls to unsleep() and cv_waitq_remove() have been replaced with calls to sleepq_remove(). - The sched_sleep() function no longer accepts a priority argument as sleep's no longer inherently bump the priority. Instead, this is soley a propery of msleep() which explicitly calls sched_prio() before blocking. - The TDF_ONSLEEPQ flag has been dropped as it was never used. The associated TDF_SET_ONSLEEPQ and TDF_CLR_ON_SLEEPQ macros have also been dropped and replaced with a single explicit clearing of td_wchan. TD_SET_ONSLEEPQ() would really have only made sense if it had taken the wait channel and message as arguments anyway. Now that that only happens in one place, a macro would be overkill.
2004-02-27 18:52:44 +00:00
sq = sleepq_lookup(cvp);
sleepq_release(cvp);
KASSERT(sq == NULL, ("%s: associated sleep queue non-empty", __func__));
#endif
2001-01-16 01:00:43 +00:00
}
/*
* Wait on a condition variable. The current thread is placed on the condition
2001-01-16 01:00:43 +00:00
* 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
2001-01-16 01:00:43 +00:00
* 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)
2001-01-16 01:00:43 +00:00
{
WITNESS_SAVE_DECL(lock_witness);
struct lock_class *class;
struct thread *td;
int lock_state;
2001-01-16 01:00:43 +00:00
td = curthread;
lock_state = 0;
#ifdef KTRACE
if (KTRPOINT(td, KTR_CSW))
ktrcsw(1, 0);
#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);
#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;
2001-01-16 01:00:43 +00:00
#ifdef KTRACE
if (KTRPOINT(td, KTR_CSW))
ktrcsw(1, 0);
2001-01-16 01:00:43 +00:00
#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);
2001-01-16 01:00:43 +00:00
Switch the sleep/wakeup and condition variable implementations to use the sleep queue interface: - Sleep queues attempt to merge some of the benefits of both sleep queues and condition variables. Having sleep qeueus in a hash table avoids having to allocate a queue head for each wait channel. Thus, struct cv has shrunk down to just a single char * pointer now. However, the hash table does not hold threads directly, but queue heads. This means that once you have located a queue in the hash bucket, you no longer have to walk the rest of the hash chain looking for threads. Instead, you have a list of all the threads sleeping on that wait channel. - Outside of the sleepq code and the sleep/cv code the kernel no longer differentiates between cv's and sleep/wakeup. For example, calls to abortsleep() and cv_abort() are replaced with a call to sleepq_abort(). Thus, the TDF_CVWAITQ flag is removed. Also, calls to unsleep() and cv_waitq_remove() have been replaced with calls to sleepq_remove(). - The sched_sleep() function no longer accepts a priority argument as sleep's no longer inherently bump the priority. Instead, this is soley a propery of msleep() which explicitly calls sched_prio() before blocking. - The TDF_ONSLEEPQ flag has been dropped as it was never used. The associated TDF_SET_ONSLEEPQ and TDF_CLR_ON_SLEEPQ macros have also been dropped and replaced with a single explicit clearing of td_wchan. TD_SET_ONSLEEPQ() would really have only made sense if it had taken the wait channel and message as arguments anyway. Now that that only happens in one place, a macro would be overkill.
2004-02-27 18:52:44 +00:00
if (cold || panicstr) {
2001-01-16 01:00:43 +00:00
/*
* 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.
2001-01-16 01:00:43 +00:00
*/
class->lc_unlock(lock);
2001-01-16 01:00:43 +00:00
return;
}
sleepq_lock(cvp);
2001-01-16 01:00:43 +00:00
cvp->cv_waiters++;
Change the preemption code for software interrupt thread schedules and mutex releases to not require flags for the cases when preemption is not allowed: The purpose of the MTX_NOSWITCH and SWI_NOSWITCH flags is to prevent switching to a higher priority thread on mutex releease and swi schedule, respectively when that switch is not safe. Now that the critical section API maintains a per-thread nesting count, the kernel can easily check whether or not it should switch without relying on flags from the programmer. This fixes a few bugs in that all current callers of swi_sched() used SWI_NOSWITCH, when in fact, only the ones called from fast interrupt handlers and the swi_sched of softclock needed this flag. Note that to ensure that swi_sched()'s in clock and fast interrupt handlers do not switch, these handlers have to be explicitly wrapped in critical_enter/exit pairs. Presently, just wrapping the handlers is sufficient, but in the future with the fully preemptive kernel, the interrupt must be EOI'd before critical_exit() is called. (critical_exit() can switch due to a deferred preemption in a fully preemptive kernel.) I've tested the changes to the interrupt code on i386 and alpha. I have not tested ia64, but the interrupt code is almost identical to the alpha code, so I expect it will work fine. PowerPC and ARM do not yet have interrupt code in the tree so they shouldn't be broken. Sparc64 is broken, but that's been ok'd by jake and tmm who will be fixing the interrupt code for sparc64 shortly. Reviewed by: peter Tested on: i386, alpha
2002-01-05 08:47:13 +00:00
DROP_GIANT();
2001-01-16 01:00:43 +00:00
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);
2001-01-16 01:00:43 +00:00
#ifdef KTRACE
if (KTRPOINT(td, KTR_CSW))
ktrcsw(0, 0);
#endif
2001-01-16 01:00:43 +00:00
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
2001-01-16 01:00:43 +00:00
* 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)
2001-01-16 01:00:43 +00:00
{
WITNESS_SAVE_DECL(lock_witness);
struct lock_class *class;
struct thread *td;
int lock_state, rval;
2001-01-16 01:00:43 +00:00
td = curthread;
lock_state = 0;
2001-01-16 01:00:43 +00:00
#ifdef KTRACE
if (KTRPOINT(td, KTR_CSW))
ktrcsw(1, 0);
2001-01-16 01:00:43 +00:00
#endif
CV_ASSERT(cvp, lock, td);
WITNESS_WARN(WARN_GIANTOK | WARN_SLEEPOK, lock,
"Waiting on \"%s\"", cvp->cv_description);
class = LOCK_CLASS(lock);
2001-01-16 01:00:43 +00:00
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);
2001-01-16 01:00:43 +00:00
}
sleepq_lock(cvp);
cvp->cv_waiters++;
if (lock == &Giant.lock_object)
mtx_assert(&Giant, MA_OWNED);
Change the preemption code for software interrupt thread schedules and mutex releases to not require flags for the cases when preemption is not allowed: The purpose of the MTX_NOSWITCH and SWI_NOSWITCH flags is to prevent switching to a higher priority thread on mutex releease and swi schedule, respectively when that switch is not safe. Now that the critical section API maintains a per-thread nesting count, the kernel can easily check whether or not it should switch without relying on flags from the programmer. This fixes a few bugs in that all current callers of swi_sched() used SWI_NOSWITCH, when in fact, only the ones called from fast interrupt handlers and the swi_sched of softclock needed this flag. Note that to ensure that swi_sched()'s in clock and fast interrupt handlers do not switch, these handlers have to be explicitly wrapped in critical_enter/exit pairs. Presently, just wrapping the handlers is sufficient, but in the future with the fully preemptive kernel, the interrupt must be EOI'd before critical_exit() is called. (critical_exit() can switch due to a deferred preemption in a fully preemptive kernel.) I've tested the changes to the interrupt code on i386 and alpha. I have not tested ia64, but the interrupt code is almost identical to the alpha code, so I expect it will work fine. PowerPC and ARM do not yet have interrupt code in the tree so they shouldn't be broken. Sparc64 is broken, but that's been ok'd by jake and tmm who will be fixing the interrupt code for sparc64 shortly. Reviewed by: peter Tested on: i386, alpha
2002-01-05 08:47:13 +00:00
DROP_GIANT();
2001-01-16 01:00:43 +00:00
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);
2001-01-16 01:00:43 +00:00
#ifdef KTRACE
if (KTRPOINT(td, KTR_CSW))
ktrcsw(0, 0);
2001-01-16 01:00:43 +00:00
#endif
PICKUP_GIANT();
if (lock != &Giant.lock_object) {
class->lc_lock(lock, lock_state);
WITNESS_RESTORE(lock, lock_witness);
}
2001-01-16 01:00:43 +00:00
return (rval);
}
/*
* Wait on a condition variable for at most timo/hz seconds. Returns 0 if the
* process was resumed by cv_signal or cv_broadcast, EWOULDBLOCK if the timeout
* expires.
*/
int
_cv_timedwait(struct cv *cvp, struct lock_object *lock, int timo)
2001-01-16 01:00:43 +00:00
{
WITNESS_SAVE_DECL(lock_witness);
struct lock_class *class;
struct thread *td;
int lock_state, rval;
2001-01-16 01:00:43 +00:00
td = curthread;
lock_state = 0;
2001-01-16 01:00:43 +00:00
#ifdef KTRACE
if (KTRPOINT(td, KTR_CSW))
ktrcsw(1, 0);
2001-01-16 01:00:43 +00:00
#endif
CV_ASSERT(cvp, lock, td);
WITNESS_WARN(WARN_GIANTOK | WARN_SLEEPOK, lock,
"Waiting on \"%s\"", cvp->cv_description);
class = LOCK_CLASS(lock);
2001-01-16 01:00:43 +00:00
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
2001-01-16 01:00:43 +00:00
* already asleep.
*/
return 0;
}
sleepq_lock(cvp);
2001-01-16 01:00:43 +00:00
cvp->cv_waiters++;
if (lock == &Giant.lock_object)
mtx_assert(&Giant, MA_OWNED);
Change the preemption code for software interrupt thread schedules and mutex releases to not require flags for the cases when preemption is not allowed: The purpose of the MTX_NOSWITCH and SWI_NOSWITCH flags is to prevent switching to a higher priority thread on mutex releease and swi schedule, respectively when that switch is not safe. Now that the critical section API maintains a per-thread nesting count, the kernel can easily check whether or not it should switch without relying on flags from the programmer. This fixes a few bugs in that all current callers of swi_sched() used SWI_NOSWITCH, when in fact, only the ones called from fast interrupt handlers and the swi_sched of softclock needed this flag. Note that to ensure that swi_sched()'s in clock and fast interrupt handlers do not switch, these handlers have to be explicitly wrapped in critical_enter/exit pairs. Presently, just wrapping the handlers is sufficient, but in the future with the fully preemptive kernel, the interrupt must be EOI'd before critical_exit() is called. (critical_exit() can switch due to a deferred preemption in a fully preemptive kernel.) I've tested the changes to the interrupt code on i386 and alpha. I have not tested ia64, but the interrupt code is almost identical to the alpha code, so I expect it will work fine. PowerPC and ARM do not yet have interrupt code in the tree so they shouldn't be broken. Sparc64 is broken, but that's been ok'd by jake and tmm who will be fixing the interrupt code for sparc64 shortly. Reviewed by: peter Tested on: i386, alpha
2002-01-05 08:47:13 +00:00
DROP_GIANT();
2001-01-16 01:00:43 +00:00
sleepq_add(cvp, lock, cvp->cv_description, SLEEPQ_CONDVAR, 0);
sleepq_set_timeout(cvp, timo);
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_timedwait(cvp, 0);
2001-01-16 01:00:43 +00:00
#ifdef KTRACE
if (KTRPOINT(td, KTR_CSW))
ktrcsw(0, 0);
2001-01-16 01:00:43 +00:00
#endif
PICKUP_GIANT();
if (lock != &Giant.lock_object) {
class->lc_lock(lock, lock_state);
WITNESS_RESTORE(lock, lock_witness);
}
2001-01-16 01:00:43 +00:00
return (rval);
}
/*
* Wait on a condition variable for at most timo/hz seconds, allowing
* interruption by signals. Returns 0 if the thread was resumed by cv_signal
2001-01-16 01:00:43 +00:00
* or cv_broadcast, EWOULDBLOCK if the timeout expires, and EINTR or ERESTART if
* a signal was caught.
*/
int
_cv_timedwait_sig(struct cv *cvp, struct lock_object *lock, int timo)
2001-01-16 01:00:43 +00:00
{
WITNESS_SAVE_DECL(lock_witness);
struct lock_class *class;
struct thread *td;
int lock_state, rval;
2001-01-16 01:00:43 +00:00
td = curthread;
lock_state = 0;
2001-01-16 01:00:43 +00:00
#ifdef KTRACE
if (KTRPOINT(td, KTR_CSW))
ktrcsw(1, 0);
2001-01-16 01:00:43 +00:00
#endif
CV_ASSERT(cvp, lock, td);
WITNESS_WARN(WARN_GIANTOK | WARN_SLEEPOK, lock,
"Waiting on \"%s\"", cvp->cv_description);
class = LOCK_CLASS(lock);
2001-01-16 01:00:43 +00:00
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
2001-01-16 01:00:43 +00:00
* already asleep.
*/
return 0;
}
sleepq_lock(cvp);
2001-01-16 01:00:43 +00:00
cvp->cv_waiters++;
if (lock == &Giant.lock_object)
mtx_assert(&Giant, MA_OWNED);
Change the preemption code for software interrupt thread schedules and mutex releases to not require flags for the cases when preemption is not allowed: The purpose of the MTX_NOSWITCH and SWI_NOSWITCH flags is to prevent switching to a higher priority thread on mutex releease and swi schedule, respectively when that switch is not safe. Now that the critical section API maintains a per-thread nesting count, the kernel can easily check whether or not it should switch without relying on flags from the programmer. This fixes a few bugs in that all current callers of swi_sched() used SWI_NOSWITCH, when in fact, only the ones called from fast interrupt handlers and the swi_sched of softclock needed this flag. Note that to ensure that swi_sched()'s in clock and fast interrupt handlers do not switch, these handlers have to be explicitly wrapped in critical_enter/exit pairs. Presently, just wrapping the handlers is sufficient, but in the future with the fully preemptive kernel, the interrupt must be EOI'd before critical_exit() is called. (critical_exit() can switch due to a deferred preemption in a fully preemptive kernel.) I've tested the changes to the interrupt code on i386 and alpha. I have not tested ia64, but the interrupt code is almost identical to the alpha code, so I expect it will work fine. PowerPC and ARM do not yet have interrupt code in the tree so they shouldn't be broken. Sparc64 is broken, but that's been ok'd by jake and tmm who will be fixing the interrupt code for sparc64 shortly. Reviewed by: peter Tested on: i386, alpha
2002-01-05 08:47:13 +00:00
DROP_GIANT();
2001-01-16 01:00:43 +00:00
sleepq_add(cvp, lock, cvp->cv_description, SLEEPQ_CONDVAR |
SLEEPQ_INTERRUPTIBLE, 0);
sleepq_set_timeout(cvp, timo);
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_timedwait_sig(cvp, 0);
2001-01-16 01:00:43 +00:00
#ifdef KTRACE
if (KTRPOINT(td, KTR_CSW))
ktrcsw(0, 0);
2001-01-16 01:00:43 +00:00
#endif
PICKUP_GIANT();
if (lock != &Giant.lock_object) {
class->lc_lock(lock, lock_state);
WITNESS_RESTORE(lock, lock_witness);
}
2001-01-16 01:00:43 +00:00
return (rval);
}
/*
* Signal a condition variable, wakes up one waiting thread. Will also wakeup
2001-01-16 01:00:43 +00:00
* 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
2001-01-16 01:00:43 +00:00
* being made runnable. Should be called with the same mutex as was passed to
* cv_wait held.
*/
void
cv_signal(struct cv *cvp)
{
If a thread that is swapped out is made runnable, then the setrunnable() routine wakes up proc0 so that proc0 can swap the thread back in. Historically, this has been done by waking up proc0 directly from setrunnable() itself via a wakeup(). When waking up a sleeping thread that was swapped out (the usual case when waking proc0 since only sleeping threads are eligible to be swapped out), this resulted in a bit of recursion (e.g. wakeup() -> setrunnable() -> wakeup()). With sleep queues having separate locks in 6.x and later, this caused a spin lock LOR (sleepq lock -> sched_lock/thread lock -> sleepq lock). An attempt was made to fix this in 7.0 by making the proc0 wakeup use the ithread mechanism for doing the wakeup. However, this required grabbing proc0's thread lock to perform the wakeup. If proc0 was asleep elsewhere in the kernel (e.g. waiting for disk I/O), then this degenerated into the same LOR since the thread lock would be some other sleepq lock. Fix this by deferring the wakeup of the swapper until after the sleepq lock held by the upper layer has been locked. The setrunnable() routine now returns a boolean value to indicate whether or not proc0 needs to be woken up. The end result is that consumers of the sleepq API such as *sleep/wakeup, condition variables, sx locks, and lockmgr, have to wakeup proc0 if they get a non-zero return value from sleepq_abort(), sleepq_broadcast(), or sleepq_signal(). Discussed with: jeff Glanced at by: sam Tested by: Jurgen Weber jurgen - ish com au MFC after: 2 weeks
2008-08-05 20:02:31 +00:00
int wakeup_swapper;
2001-01-16 01:00:43 +00:00
If a thread that is swapped out is made runnable, then the setrunnable() routine wakes up proc0 so that proc0 can swap the thread back in. Historically, this has been done by waking up proc0 directly from setrunnable() itself via a wakeup(). When waking up a sleeping thread that was swapped out (the usual case when waking proc0 since only sleeping threads are eligible to be swapped out), this resulted in a bit of recursion (e.g. wakeup() -> setrunnable() -> wakeup()). With sleep queues having separate locks in 6.x and later, this caused a spin lock LOR (sleepq lock -> sched_lock/thread lock -> sleepq lock). An attempt was made to fix this in 7.0 by making the proc0 wakeup use the ithread mechanism for doing the wakeup. However, this required grabbing proc0's thread lock to perform the wakeup. If proc0 was asleep elsewhere in the kernel (e.g. waiting for disk I/O), then this degenerated into the same LOR since the thread lock would be some other sleepq lock. Fix this by deferring the wakeup of the swapper until after the sleepq lock held by the upper layer has been locked. The setrunnable() routine now returns a boolean value to indicate whether or not proc0 needs to be woken up. The end result is that consumers of the sleepq API such as *sleep/wakeup, condition variables, sx locks, and lockmgr, have to wakeup proc0 if they get a non-zero return value from sleepq_abort(), sleepq_broadcast(), or sleepq_signal(). Discussed with: jeff Glanced at by: sam Tested by: Jurgen Weber jurgen - ish com au MFC after: 2 weeks
2008-08-05 20:02:31 +00:00
wakeup_swapper = 0;
sleepq_lock(cvp);
if (cvp->cv_waiters > 0) {
cvp->cv_waiters--;
If a thread that is swapped out is made runnable, then the setrunnable() routine wakes up proc0 so that proc0 can swap the thread back in. Historically, this has been done by waking up proc0 directly from setrunnable() itself via a wakeup(). When waking up a sleeping thread that was swapped out (the usual case when waking proc0 since only sleeping threads are eligible to be swapped out), this resulted in a bit of recursion (e.g. wakeup() -> setrunnable() -> wakeup()). With sleep queues having separate locks in 6.x and later, this caused a spin lock LOR (sleepq lock -> sched_lock/thread lock -> sleepq lock). An attempt was made to fix this in 7.0 by making the proc0 wakeup use the ithread mechanism for doing the wakeup. However, this required grabbing proc0's thread lock to perform the wakeup. If proc0 was asleep elsewhere in the kernel (e.g. waiting for disk I/O), then this degenerated into the same LOR since the thread lock would be some other sleepq lock. Fix this by deferring the wakeup of the swapper until after the sleepq lock held by the upper layer has been locked. The setrunnable() routine now returns a boolean value to indicate whether or not proc0 needs to be woken up. The end result is that consumers of the sleepq API such as *sleep/wakeup, condition variables, sx locks, and lockmgr, have to wakeup proc0 if they get a non-zero return value from sleepq_abort(), sleepq_broadcast(), or sleepq_signal(). Discussed with: jeff Glanced at by: sam Tested by: Jurgen Weber jurgen - ish com au MFC after: 2 weeks
2008-08-05 20:02:31 +00:00
wakeup_swapper = sleepq_signal(cvp, SLEEPQ_CONDVAR, 0, 0);
}
sleepq_release(cvp);
If a thread that is swapped out is made runnable, then the setrunnable() routine wakes up proc0 so that proc0 can swap the thread back in. Historically, this has been done by waking up proc0 directly from setrunnable() itself via a wakeup(). When waking up a sleeping thread that was swapped out (the usual case when waking proc0 since only sleeping threads are eligible to be swapped out), this resulted in a bit of recursion (e.g. wakeup() -> setrunnable() -> wakeup()). With sleep queues having separate locks in 6.x and later, this caused a spin lock LOR (sleepq lock -> sched_lock/thread lock -> sleepq lock). An attempt was made to fix this in 7.0 by making the proc0 wakeup use the ithread mechanism for doing the wakeup. However, this required grabbing proc0's thread lock to perform the wakeup. If proc0 was asleep elsewhere in the kernel (e.g. waiting for disk I/O), then this degenerated into the same LOR since the thread lock would be some other sleepq lock. Fix this by deferring the wakeup of the swapper until after the sleepq lock held by the upper layer has been locked. The setrunnable() routine now returns a boolean value to indicate whether or not proc0 needs to be woken up. The end result is that consumers of the sleepq API such as *sleep/wakeup, condition variables, sx locks, and lockmgr, have to wakeup proc0 if they get a non-zero return value from sleepq_abort(), sleepq_broadcast(), or sleepq_signal(). Discussed with: jeff Glanced at by: sam Tested by: Jurgen Weber jurgen - ish com au MFC after: 2 weeks
2008-08-05 20:02:31 +00:00
if (wakeup_swapper)
kick_proc0();
2001-01-16 01:00:43 +00:00
}
/*
* Broadcast a signal to a condition variable. Wakes up all waiting threads.
2001-01-16 01:00:43 +00:00
* Should be called with the same mutex as was passed to cv_wait held.
*/
void
cv_broadcastpri(struct cv *cvp, int pri)
2001-01-16 01:00:43 +00:00
{
If a thread that is swapped out is made runnable, then the setrunnable() routine wakes up proc0 so that proc0 can swap the thread back in. Historically, this has been done by waking up proc0 directly from setrunnable() itself via a wakeup(). When waking up a sleeping thread that was swapped out (the usual case when waking proc0 since only sleeping threads are eligible to be swapped out), this resulted in a bit of recursion (e.g. wakeup() -> setrunnable() -> wakeup()). With sleep queues having separate locks in 6.x and later, this caused a spin lock LOR (sleepq lock -> sched_lock/thread lock -> sleepq lock). An attempt was made to fix this in 7.0 by making the proc0 wakeup use the ithread mechanism for doing the wakeup. However, this required grabbing proc0's thread lock to perform the wakeup. If proc0 was asleep elsewhere in the kernel (e.g. waiting for disk I/O), then this degenerated into the same LOR since the thread lock would be some other sleepq lock. Fix this by deferring the wakeup of the swapper until after the sleepq lock held by the upper layer has been locked. The setrunnable() routine now returns a boolean value to indicate whether or not proc0 needs to be woken up. The end result is that consumers of the sleepq API such as *sleep/wakeup, condition variables, sx locks, and lockmgr, have to wakeup proc0 if they get a non-zero return value from sleepq_abort(), sleepq_broadcast(), or sleepq_signal(). Discussed with: jeff Glanced at by: sam Tested by: Jurgen Weber jurgen - ish com au MFC after: 2 weeks
2008-08-05 20:02:31 +00:00
int wakeup_swapper;
/*
* XXX sleepq_broadcast pri argument changed from -1 meaning
* no pri to 0 meaning no pri.
*/
If a thread that is swapped out is made runnable, then the setrunnable() routine wakes up proc0 so that proc0 can swap the thread back in. Historically, this has been done by waking up proc0 directly from setrunnable() itself via a wakeup(). When waking up a sleeping thread that was swapped out (the usual case when waking proc0 since only sleeping threads are eligible to be swapped out), this resulted in a bit of recursion (e.g. wakeup() -> setrunnable() -> wakeup()). With sleep queues having separate locks in 6.x and later, this caused a spin lock LOR (sleepq lock -> sched_lock/thread lock -> sleepq lock). An attempt was made to fix this in 7.0 by making the proc0 wakeup use the ithread mechanism for doing the wakeup. However, this required grabbing proc0's thread lock to perform the wakeup. If proc0 was asleep elsewhere in the kernel (e.g. waiting for disk I/O), then this degenerated into the same LOR since the thread lock would be some other sleepq lock. Fix this by deferring the wakeup of the swapper until after the sleepq lock held by the upper layer has been locked. The setrunnable() routine now returns a boolean value to indicate whether or not proc0 needs to be woken up. The end result is that consumers of the sleepq API such as *sleep/wakeup, condition variables, sx locks, and lockmgr, have to wakeup proc0 if they get a non-zero return value from sleepq_abort(), sleepq_broadcast(), or sleepq_signal(). Discussed with: jeff Glanced at by: sam Tested by: Jurgen Weber jurgen - ish com au MFC after: 2 weeks
2008-08-05 20:02:31 +00:00
wakeup_swapper = 0;
if (pri == -1)
pri = 0;
sleepq_lock(cvp);
if (cvp->cv_waiters > 0) {
cvp->cv_waiters = 0;
If a thread that is swapped out is made runnable, then the setrunnable() routine wakes up proc0 so that proc0 can swap the thread back in. Historically, this has been done by waking up proc0 directly from setrunnable() itself via a wakeup(). When waking up a sleeping thread that was swapped out (the usual case when waking proc0 since only sleeping threads are eligible to be swapped out), this resulted in a bit of recursion (e.g. wakeup() -> setrunnable() -> wakeup()). With sleep queues having separate locks in 6.x and later, this caused a spin lock LOR (sleepq lock -> sched_lock/thread lock -> sleepq lock). An attempt was made to fix this in 7.0 by making the proc0 wakeup use the ithread mechanism for doing the wakeup. However, this required grabbing proc0's thread lock to perform the wakeup. If proc0 was asleep elsewhere in the kernel (e.g. waiting for disk I/O), then this degenerated into the same LOR since the thread lock would be some other sleepq lock. Fix this by deferring the wakeup of the swapper until after the sleepq lock held by the upper layer has been locked. The setrunnable() routine now returns a boolean value to indicate whether or not proc0 needs to be woken up. The end result is that consumers of the sleepq API such as *sleep/wakeup, condition variables, sx locks, and lockmgr, have to wakeup proc0 if they get a non-zero return value from sleepq_abort(), sleepq_broadcast(), or sleepq_signal(). Discussed with: jeff Glanced at by: sam Tested by: Jurgen Weber jurgen - ish com au MFC after: 2 weeks
2008-08-05 20:02:31 +00:00
wakeup_swapper = sleepq_broadcast(cvp, SLEEPQ_CONDVAR, pri, 0);
}
sleepq_release(cvp);
If a thread that is swapped out is made runnable, then the setrunnable() routine wakes up proc0 so that proc0 can swap the thread back in. Historically, this has been done by waking up proc0 directly from setrunnable() itself via a wakeup(). When waking up a sleeping thread that was swapped out (the usual case when waking proc0 since only sleeping threads are eligible to be swapped out), this resulted in a bit of recursion (e.g. wakeup() -> setrunnable() -> wakeup()). With sleep queues having separate locks in 6.x and later, this caused a spin lock LOR (sleepq lock -> sched_lock/thread lock -> sleepq lock). An attempt was made to fix this in 7.0 by making the proc0 wakeup use the ithread mechanism for doing the wakeup. However, this required grabbing proc0's thread lock to perform the wakeup. If proc0 was asleep elsewhere in the kernel (e.g. waiting for disk I/O), then this degenerated into the same LOR since the thread lock would be some other sleepq lock. Fix this by deferring the wakeup of the swapper until after the sleepq lock held by the upper layer has been locked. The setrunnable() routine now returns a boolean value to indicate whether or not proc0 needs to be woken up. The end result is that consumers of the sleepq API such as *sleep/wakeup, condition variables, sx locks, and lockmgr, have to wakeup proc0 if they get a non-zero return value from sleepq_abort(), sleepq_broadcast(), or sleepq_signal(). Discussed with: jeff Glanced at by: sam Tested by: Jurgen Weber jurgen - ish com au MFC after: 2 weeks
2008-08-05 20:02:31 +00:00
if (wakeup_swapper)
kick_proc0();
2001-01-16 01:00:43 +00:00
}