xref: /netbsd-src/sys/kern/kern_synch.c (revision dd3ee07da436799d8de85f3055253118b76bf345)

/*	$NetBSD: kern_synch.c,v 1.350 2022/03/10 12:21:25 riastradh Exp $	*/

/*-
 * Copyright (c) 1999, 2000, 2004, 2006, 2007, 2008, 2009, 2019, 2020
 *    The NetBSD Foundation, Inc.
 * All rights reserved.
 *
 * This code is derived from software contributed to The NetBSD Foundation
 * by Jason R. Thorpe of the Numerical Aerospace Simulation Facility,
 * NASA Ames Research Center, by Charles M. Hannum, Andrew Doran and
 * Daniel Sieger.
 *
 * 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 NETBSD FOUNDATION, INC. 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 FOUNDATION 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.
 */

/*-
 * Copyright (c) 1982, 1986, 1990, 1991, 1993
 *	The Regents of the University of California.  All rights reserved.
 * (c) UNIX System Laboratories, Inc.
 * All or some portions of this file are derived from material licensed
 * to the University of California by American Telephone and Telegraph
 * Co. or Unix System Laboratories, Inc. and are reproduced herein with
 * the permission of UNIX System Laboratories, Inc.
 *
 * 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.
 * 3. Neither the name of the University nor the names of its contributors
 *    may be used to endorse or promote products derived from this software
 *    without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE REGENTS 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 REGENTS 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.
 *
 *	@(#)kern_synch.c	8.9 (Berkeley) 5/19/95
 */

#include <sys/cdefs.h>
__KERNEL_RCSID(0, "$NetBSD: kern_synch.c,v 1.350 2022/03/10 12:21:25 riastradh Exp $");

#include "opt_kstack.h"
#include "opt_dtrace.h"

#define	__MUTEX_PRIVATE

#include <sys/param.h>
#include <sys/systm.h>
#include <sys/proc.h>
#include <sys/kernel.h>
#include <sys/cpu.h>
#include <sys/pserialize.h>
#include <sys/resourcevar.h>
#include <sys/rwlock.h>
#include <sys/sched.h>
#include <sys/syscall_stats.h>
#include <sys/sleepq.h>
#include <sys/lockdebug.h>
#include <sys/evcnt.h>
#include <sys/intr.h>
#include <sys/lwpctl.h>
#include <sys/atomic.h>
#include <sys/syslog.h>

#include <uvm/uvm_extern.h>

#include <dev/lockstat.h>

#include <sys/dtrace_bsd.h>
int                             dtrace_vtime_active=0;
dtrace_vtime_switch_func_t      dtrace_vtime_switch_func;

static void	sched_unsleep(struct lwp *, bool);
static void	sched_changepri(struct lwp *, pri_t);
static void	sched_lendpri(struct lwp *, pri_t);

syncobj_t sleep_syncobj = {
	.sobj_flag	= SOBJ_SLEEPQ_SORTED,
	.sobj_unsleep	= sleepq_unsleep,
	.sobj_changepri	= sleepq_changepri,
	.sobj_lendpri	= sleepq_lendpri,
	.sobj_owner	= syncobj_noowner,
};

syncobj_t sched_syncobj = {
	.sobj_flag	= SOBJ_SLEEPQ_SORTED,
	.sobj_unsleep	= sched_unsleep,
	.sobj_changepri	= sched_changepri,
	.sobj_lendpri	= sched_lendpri,
	.sobj_owner	= syncobj_noowner,
};

syncobj_t kpause_syncobj = {
	.sobj_flag	= SOBJ_SLEEPQ_NULL,
	.sobj_unsleep	= sleepq_unsleep,
	.sobj_changepri	= sleepq_changepri,
	.sobj_lendpri	= sleepq_lendpri,
	.sobj_owner	= syncobj_noowner,
};

/* "Lightning bolt": once a second sleep address. */
kcondvar_t		lbolt			__cacheline_aligned;

u_int			sched_pstats_ticks	__cacheline_aligned;

/* Preemption event counters. */
static struct evcnt	kpreempt_ev_crit	__cacheline_aligned;
static struct evcnt	kpreempt_ev_klock	__cacheline_aligned;
static struct evcnt	kpreempt_ev_immed	__cacheline_aligned;

void
synch_init(void)
{

	cv_init(&lbolt, "lbolt");

	evcnt_attach_dynamic(&kpreempt_ev_crit, EVCNT_TYPE_MISC, NULL,
	   "kpreempt", "defer: critical section");
	evcnt_attach_dynamic(&kpreempt_ev_klock, EVCNT_TYPE_MISC, NULL,
	   "kpreempt", "defer: kernel_lock");
	evcnt_attach_dynamic(&kpreempt_ev_immed, EVCNT_TYPE_MISC, NULL,
	   "kpreempt", "immediate");
}

/*
 * OBSOLETE INTERFACE
 *
 * General sleep call.  Suspends the current LWP until a wakeup is
 * performed on the specified identifier.  The LWP will then be made
 * runnable with the specified priority.  Sleeps at most timo/hz seconds (0
 * means no timeout).  If pri includes PCATCH flag, signals are checked
 * before and after sleeping, else signals are not checked.  Returns 0 if
 * awakened, EWOULDBLOCK if the timeout expires.  If PCATCH is set and a
 * signal needs to be delivered, 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).
 */
int
tsleep(wchan_t ident, pri_t priority, const char *wmesg, int timo)
{
	struct lwp *l = curlwp;
	sleepq_t *sq;
	kmutex_t *mp;
	bool catch_p;

	KASSERT((l->l_pflag & LP_INTR) == 0);
	KASSERT(ident != &lbolt);

	if (sleepq_dontsleep(l)) {
		(void)sleepq_abort(NULL, 0);
		return 0;
	}

	l->l_kpriority = true;
	catch_p = priority & PCATCH;
	sq = sleeptab_lookup(&sleeptab, ident, &mp);
	sleepq_enter(sq, l, mp);
	sleepq_enqueue(sq, ident, wmesg, &sleep_syncobj, catch_p);
	return sleepq_block(timo, catch_p);
}

int
mtsleep(wchan_t ident, pri_t priority, const char *wmesg, int timo,
	kmutex_t *mtx)
{
	struct lwp *l = curlwp;
	sleepq_t *sq;
	kmutex_t *mp;
	bool catch_p;
	int error;

	KASSERT((l->l_pflag & LP_INTR) == 0);
	KASSERT(ident != &lbolt);

	if (sleepq_dontsleep(l)) {
		(void)sleepq_abort(mtx, (priority & PNORELOCK) != 0);
		return 0;
	}

	l->l_kpriority = true;
	catch_p = priority & PCATCH;
	sq = sleeptab_lookup(&sleeptab, ident, &mp);
	sleepq_enter(sq, l, mp);
	sleepq_enqueue(sq, ident, wmesg, &sleep_syncobj, catch_p);
	mutex_exit(mtx);
	error = sleepq_block(timo, catch_p);

	if ((priority & PNORELOCK) == 0)
		mutex_enter(mtx);

	return error;
}

/*
 * General sleep call for situations where a wake-up is not expected.
 */
int
kpause(const char *wmesg, bool intr, int timo, kmutex_t *mtx)
{
	struct lwp *l = curlwp;
	int error;

	KASSERT(!(timo == 0 && intr == false));

	if (sleepq_dontsleep(l))
		return sleepq_abort(NULL, 0);

	if (mtx != NULL)
		mutex_exit(mtx);
	l->l_kpriority = true;
	lwp_lock(l);
	KERNEL_UNLOCK_ALL(NULL, &l->l_biglocks);
	sleepq_enqueue(NULL, l, wmesg, &kpause_syncobj, intr);
	error = sleepq_block(timo, intr);
	if (mtx != NULL)
		mutex_enter(mtx);

	return error;
}

/*
 * OBSOLETE INTERFACE
 *
 * Make all LWPs sleeping on the specified identifier runnable.
 */
void
wakeup(wchan_t ident)
{
	sleepq_t *sq;
	kmutex_t *mp;

	if (__predict_false(cold))
		return;

	sq = sleeptab_lookup(&sleeptab, ident, &mp);
	sleepq_wake(sq, ident, (u_int)-1, mp);
}

/*
 * General yield call.  Puts the current LWP back on its run queue and
 * performs a context switch.
 */
void
yield(void)
{
	struct lwp *l = curlwp;

	KERNEL_UNLOCK_ALL(l, &l->l_biglocks);
	lwp_lock(l);

	KASSERT(lwp_locked(l, l->l_cpu->ci_schedstate.spc_lwplock));
	KASSERT(l->l_stat == LSONPROC);

	/* Voluntary - ditch kpriority boost. */
	l->l_kpriority = false;
	spc_lock(l->l_cpu);
	mi_switch(l);
	KERNEL_LOCK(l->l_biglocks, l);
}

/*
 * General preemption call.  Puts the current LWP back on its run queue
 * and performs an involuntary context switch.  Different from yield()
 * in that:
 *
 * - It's counted differently (involuntary vs. voluntary).
 * - Realtime threads go to the head of their runqueue vs. tail for yield().
 * - Priority boost is retained unless LWP has exceeded timeslice.
 */
void
preempt(void)
{
	struct lwp *l = curlwp;

	KERNEL_UNLOCK_ALL(l, &l->l_biglocks);
	lwp_lock(l);

	KASSERT(lwp_locked(l, l->l_cpu->ci_schedstate.spc_lwplock));
	KASSERT(l->l_stat == LSONPROC);

	spc_lock(l->l_cpu);
	/* Involuntary - keep kpriority boost unless a CPU hog. */
	if ((l->l_cpu->ci_schedstate.spc_flags & SPCF_SHOULDYIELD) != 0) {
		l->l_kpriority = false;
	}
	l->l_pflag |= LP_PREEMPTING;
	mi_switch(l);
	KERNEL_LOCK(l->l_biglocks, l);
}

/*
 * Return true if the current LWP should yield the processor.  Intended to
 * be used by long-running code in kernel.
 */
inline bool
preempt_needed(void)
{
	lwp_t *l = curlwp;
	int needed;

	KPREEMPT_DISABLE(l);
	needed = l->l_cpu->ci_want_resched;
	KPREEMPT_ENABLE(l);

	return (needed != 0);
}

/*
 * A breathing point for long running code in kernel.
 */
void
preempt_point(void)
{

	if (__predict_false(preempt_needed())) {
		preempt();
	}
}

/*
 * Handle a request made by another agent to preempt the current LWP
 * in-kernel.  Usually called when l_dopreempt may be non-zero.
 *
 * Character addresses for lockstat only.
 */
static char	kpreempt_is_disabled;
static char	kernel_lock_held;
static char	is_softint_lwp;
static char	spl_is_raised;

bool
kpreempt(uintptr_t where)
{
	uintptr_t failed;
	lwp_t *l;
	int s, dop, lsflag;

	l = curlwp;
	failed = 0;
	while ((dop = l->l_dopreempt) != 0) {
		if (l->l_stat != LSONPROC) {
			/*
			 * About to block (or die), let it happen.
			 * Doesn't really count as "preemption has
			 * been blocked", since we're going to
			 * context switch.
			 */
			atomic_swap_uint(&l->l_dopreempt, 0);
			return true;
		}
		KASSERT((l->l_flag & LW_IDLE) == 0);
		if (__predict_false(l->l_nopreempt != 0)) {
			/* LWP holds preemption disabled, explicitly. */
			if ((dop & DOPREEMPT_COUNTED) == 0) {
				kpreempt_ev_crit.ev_count++;
			}
			failed = (uintptr_t)&kpreempt_is_disabled;
			break;
		}
		if (__predict_false((l->l_pflag & LP_INTR) != 0)) {
			/* Can't preempt soft interrupts yet. */
			atomic_swap_uint(&l->l_dopreempt, 0);
			failed = (uintptr_t)&is_softint_lwp;
			break;
		}
		s = splsched();
		if (__predict_false(l->l_blcnt != 0 ||
		    curcpu()->ci_biglock_wanted != NULL)) {
			/* Hold or want kernel_lock, code is not MT safe. */
			splx(s);
			if ((dop & DOPREEMPT_COUNTED) == 0) {
				kpreempt_ev_klock.ev_count++;
			}
			failed = (uintptr_t)&kernel_lock_held;
			break;
		}
		if (__predict_false(!cpu_kpreempt_enter(where, s))) {
			/*
			 * It may be that the IPL is too high.
			 * kpreempt_enter() can schedule an
			 * interrupt to retry later.
			 */
			splx(s);
			failed = (uintptr_t)&spl_is_raised;
			break;
		}
		/* Do it! */
		if (__predict_true((dop & DOPREEMPT_COUNTED) == 0)) {
			kpreempt_ev_immed.ev_count++;
		}
		lwp_lock(l);
		/* Involuntary - keep kpriority boost. */
		l->l_pflag |= LP_PREEMPTING;
		spc_lock(l->l_cpu);
		mi_switch(l);
		l->l_nopreempt++;
		splx(s);

		/* Take care of any MD cleanup. */
		cpu_kpreempt_exit(where);
		l->l_nopreempt--;
	}

	if (__predict_true(!failed)) {
		return false;
	}

	/* Record preemption failure for reporting via lockstat. */
	atomic_or_uint(&l->l_dopreempt, DOPREEMPT_COUNTED);
	lsflag = 0;
	LOCKSTAT_ENTER(lsflag);
	if (__predict_false(lsflag)) {
		if (where == 0) {
			where = (uintptr_t)__builtin_return_address(0);
		}
		/* Preemption is on, might recurse, so make it atomic. */
		if (atomic_cas_ptr_ni((void *)&l->l_pfailaddr, NULL,
		    (void *)where) == NULL) {
			LOCKSTAT_START_TIMER(lsflag, l->l_pfailtime);
			l->l_pfaillock = failed;
		}
	}
	LOCKSTAT_EXIT(lsflag);
	return true;
}

/*
 * Return true if preemption is explicitly disabled.
 */
bool
kpreempt_disabled(void)
{
	const lwp_t *l = curlwp;

	return l->l_nopreempt != 0 || l->l_stat == LSZOMB ||
	    (l->l_flag & LW_IDLE) != 0 || (l->l_pflag & LP_INTR) != 0 ||
	    cpu_kpreempt_disabled();
}

/*
 * Disable kernel preemption.
 */
void
kpreempt_disable(void)
{

	KPREEMPT_DISABLE(curlwp);
}

/*
 * Reenable kernel preemption.
 */
void
kpreempt_enable(void)
{

	KPREEMPT_ENABLE(curlwp);
}

/*
 * Compute the amount of time during which the current lwp was running.
 *
 * - update l_rtime unless it's an idle lwp.
 */

void
updatertime(lwp_t *l, const struct bintime *now)
{

	if (__predict_false(l->l_flag & LW_IDLE))
		return;

	/* rtime += now - stime */
	bintime_add(&l->l_rtime, now);
	bintime_sub(&l->l_rtime, &l->l_stime);
}

/*
 * Select next LWP from the current CPU to run..
 */
static inline lwp_t *
nextlwp(struct cpu_info *ci, struct schedstate_percpu *spc)
{
	lwp_t *newl;

	/*
	 * Let sched_nextlwp() select the LWP to run the CPU next.
	 * If no LWP is runnable, select the idle LWP.
	 *
	 * On arrival here LWPs on a run queue are locked by spc_mutex which
	 * is currently held.  Idle LWPs are always locked by spc_lwplock,
	 * which may or may not be held here.  On exit from this code block,
	 * in all cases newl is locked by spc_lwplock.
	 */
	newl = sched_nextlwp();
	if (newl != NULL) {
		sched_dequeue(newl);
		KASSERT(lwp_locked(newl, spc->spc_mutex));
		KASSERT(newl->l_cpu == ci);
		newl->l_stat = LSONPROC;
		newl->l_pflag |= LP_RUNNING;
		spc->spc_curpriority = lwp_eprio(newl);
		spc->spc_flags &= ~(SPCF_SWITCHCLEAR | SPCF_IDLE);
		lwp_setlock(newl, spc->spc_lwplock);
	} else {
		/*
		 * The idle LWP does not get set to LSONPROC, because
		 * otherwise it screws up the output from top(1) etc.
		 */
		newl = ci->ci_data.cpu_idlelwp;
		newl->l_pflag |= LP_RUNNING;
		spc->spc_curpriority = PRI_IDLE;
		spc->spc_flags = (spc->spc_flags & ~SPCF_SWITCHCLEAR) |
		    SPCF_IDLE;
	}

	/*
	 * Only clear want_resched if there are no pending (slow) software
	 * interrupts.  We can do this without an atomic, because no new
	 * LWPs can appear in the queue due to our hold on spc_mutex, and
	 * the update to ci_want_resched will become globally visible before
	 * the release of spc_mutex becomes globally visible.
	 */
	ci->ci_want_resched = ci->ci_data.cpu_softints;

	return newl;
}

/*
 * The machine independent parts of context switch.
 *
 * NOTE: l->l_cpu is not changed in this routine, because an LWP never
 * changes its own l_cpu (that would screw up curcpu on many ports and could
 * cause all kinds of other evil stuff).  l_cpu is always changed by some
 * other actor, when it's known the LWP is not running (the LP_RUNNING flag
 * is checked under lock).
 */
void
mi_switch(lwp_t *l)
{
	struct cpu_info *ci;
	struct schedstate_percpu *spc;
	struct lwp *newl;
	kmutex_t *lock;
	int oldspl;
	struct bintime bt;
	bool returning;

	KASSERT(lwp_locked(l, NULL));
	KASSERT(kpreempt_disabled());
	KASSERT(mutex_owned(curcpu()->ci_schedstate.spc_mutex));
	KASSERTMSG(l->l_blcnt == 0, "kernel_lock leaked");

	kstack_check_magic(l);

	binuptime(&bt);

	KASSERTMSG(l == curlwp, "l %p curlwp %p", l, curlwp);
	KASSERT((l->l_pflag & LP_RUNNING) != 0);
	KASSERT(l->l_cpu == curcpu() || l->l_stat == LSRUN);
	ci = curcpu();
	spc = &ci->ci_schedstate;
	returning = false;
	newl = NULL;

	/*
	 * If we have been asked to switch to a specific LWP, then there
	 * is no need to inspect the run queues.  If a soft interrupt is
	 * blocking, then return to the interrupted thread without adjusting
	 * VM context or its start time: neither have been changed in order
	 * to take the interrupt.
	 */
	if (l->l_switchto != NULL) {
		if ((l->l_pflag & LP_INTR) != 0) {
			returning = true;
			softint_block(l);
			if ((l->l_pflag & LP_TIMEINTR) != 0)
				updatertime(l, &bt);
		}
		newl = l->l_switchto;
		l->l_switchto = NULL;
	}
#ifndef __HAVE_FAST_SOFTINTS
	else if (ci->ci_data.cpu_softints != 0) {
		/* There are pending soft interrupts, so pick one. */
		newl = softint_picklwp();
		newl->l_stat = LSONPROC;
		newl->l_pflag |= LP_RUNNING;
	}
#endif	/* !__HAVE_FAST_SOFTINTS */

	/*
	 * If on the CPU and we have gotten this far, then we must yield.
	 */
	if (l->l_stat == LSONPROC && l != newl) {
		KASSERT(lwp_locked(l, spc->spc_lwplock));
		KASSERT((l->l_flag & LW_IDLE) == 0);
		l->l_stat = LSRUN;
		lwp_setlock(l, spc->spc_mutex);
		sched_enqueue(l);
		sched_preempted(l);

		/*
		 * Handle migration.  Note that "migrating LWP" may
		 * be reset here, if interrupt/preemption happens
		 * early in idle LWP.
		 */
		if (l->l_target_cpu != NULL && (l->l_pflag & LP_BOUND) == 0) {
			KASSERT((l->l_pflag & LP_INTR) == 0);
			spc->spc_migrating = l;
		}
	}

	/* Pick new LWP to run. */
	if (newl == NULL) {
		newl = nextlwp(ci, spc);
	}

	/* Items that must be updated with the CPU locked. */
	if (!returning) {
		/* Count time spent in current system call */
		SYSCALL_TIME_SLEEP(l);

		updatertime(l, &bt);

		/* Update the new LWP's start time. */
		newl->l_stime = bt;

		/*
		 * ci_curlwp changes when a fast soft interrupt occurs.
		 * We use ci_onproc to keep track of which kernel or
		 * user thread is running 'underneath' the software
		 * interrupt.  This is important for time accounting,
		 * itimers and forcing user threads to preempt (aston).
		 */
		ci->ci_onproc = newl;
	}

	/*
	 * Preemption related tasks.  Must be done holding spc_mutex.  Clear
	 * l_dopreempt without an atomic - it's only ever set non-zero by
	 * sched_resched_cpu() which also holds spc_mutex, and only ever
	 * cleared by the LWP itself (us) with atomics when not under lock.
	 */
	l->l_dopreempt = 0;
	if (__predict_false(l->l_pfailaddr != 0)) {
		LOCKSTAT_FLAG(lsflag);
		LOCKSTAT_ENTER(lsflag);
		LOCKSTAT_STOP_TIMER(lsflag, l->l_pfailtime);
		LOCKSTAT_EVENT_RA(lsflag, l->l_pfaillock, LB_NOPREEMPT|LB_SPIN,
		    1, l->l_pfailtime, l->l_pfailaddr);
		LOCKSTAT_EXIT(lsflag);
		l->l_pfailtime = 0;
		l->l_pfaillock = 0;
		l->l_pfailaddr = 0;
	}

	if (l != newl) {
		struct lwp *prevlwp;

		/* Release all locks, but leave the current LWP locked */
		if (l->l_mutex == spc->spc_mutex) {
			/*
			 * Drop spc_lwplock, if the current LWP has been moved
			 * to the run queue (it is now locked by spc_mutex).
			 */
			mutex_spin_exit(spc->spc_lwplock);
		} else {
			/*
			 * Otherwise, drop the spc_mutex, we are done with the
			 * run queues.
			 */
			mutex_spin_exit(spc->spc_mutex);
		}

		/* We're down to only one lock, so do debug checks. */
		LOCKDEBUG_BARRIER(l->l_mutex, 1);

		/* Count the context switch. */
		CPU_COUNT(CPU_COUNT_NSWTCH, 1);
		l->l_ncsw++;
		if ((l->l_pflag & LP_PREEMPTING) != 0) {
			l->l_nivcsw++;
			l->l_pflag &= ~LP_PREEMPTING;
		}

		/*
		 * Increase the count of spin-mutexes before the release
		 * of the last lock - we must remain at IPL_SCHED after
		 * releasing the lock.
		 */
		KASSERTMSG(ci->ci_mtx_count == -1,
		    "%s: cpu%u: ci_mtx_count (%d) != -1 "
		    "(block with spin-mutex held)",
		     __func__, cpu_index(ci), ci->ci_mtx_count);
		oldspl = MUTEX_SPIN_OLDSPL(ci);
		ci->ci_mtx_count = -2;

		/* Update status for lwpctl, if present. */
		if (l->l_lwpctl != NULL) {
			l->l_lwpctl->lc_curcpu = (l->l_stat == LSZOMB ?
			    LWPCTL_CPU_EXITED : LWPCTL_CPU_NONE);
		}

		/*
		 * If curlwp is a soft interrupt LWP, there's nobody on the
		 * other side to unlock - we're returning into an assembly
		 * trampoline.  Unlock now.  This is safe because this is a
		 * kernel LWP and is bound to current CPU: the worst anyone
		 * else will do to it, is to put it back onto this CPU's run
		 * queue (and the CPU is busy here right now!).
		 */
		if (returning) {
			/* Keep IPL_SCHED after this; MD code will fix up. */
			l->l_pflag &= ~LP_RUNNING;
			lwp_unlock(l);
		} else {
			/* A normal LWP: save old VM context. */
			pmap_deactivate(l);
		}

		/*
		 * If DTrace has set the active vtime enum to anything
		 * other than INACTIVE (0), then it should have set the
		 * function to call.
		 */
		if (__predict_false(dtrace_vtime_active)) {
			(*dtrace_vtime_switch_func)(newl);
		}

		/*
		 * We must ensure not to come here from inside a read section.
		 */
		KASSERT(pserialize_not_in_read_section());

		/* Switch to the new LWP.. */
#ifdef MULTIPROCESSOR
		KASSERT(curlwp == ci->ci_curlwp);
#endif
		KASSERTMSG(l == curlwp, "l %p curlwp %p", l, curlwp);
		prevlwp = cpu_switchto(l, newl, returning);
		ci = curcpu();
#ifdef MULTIPROCESSOR
		KASSERT(curlwp == ci->ci_curlwp);
#endif
		KASSERTMSG(l == curlwp, "l %p curlwp %p prevlwp %p",
		    l, curlwp, prevlwp);
		KASSERT(prevlwp != NULL);
		KASSERT(l->l_cpu == ci);
		KASSERT(ci->ci_mtx_count == -2);

		/*
		 * Immediately mark the previous LWP as no longer running
		 * and unlock (to keep lock wait times short as possible).
		 * We'll still be at IPL_SCHED afterwards.  If a zombie,
		 * don't touch after clearing LP_RUNNING as it could be
		 * reaped by another CPU.  Issue a memory barrier to ensure
		 * this.
		 *
		 * atomic_store_release matches atomic_load_acquire in
		 * lwp_free.
		 */
		KASSERT((prevlwp->l_pflag & LP_RUNNING) != 0);
		lock = prevlwp->l_mutex;
		if (__predict_false(prevlwp->l_stat == LSZOMB)) {
			atomic_store_release(&prevlwp->l_pflag,
			    prevlwp->l_pflag & ~LP_RUNNING);
		} else {
			prevlwp->l_pflag &= ~LP_RUNNING;
		}
		mutex_spin_exit(lock);

		/*
		 * Switched away - we have new curlwp.
		 * Restore VM context and IPL.
		 */
		pmap_activate(l);
		pcu_switchpoint(l);

		/* Update status for lwpctl, if present. */
		if (l->l_lwpctl != NULL) {
			l->l_lwpctl->lc_curcpu = (int)cpu_index(ci);
			l->l_lwpctl->lc_pctr++;
		}

		/*
		 * Normalize the spin mutex count and restore the previous
		 * SPL.  Note that, unless the caller disabled preemption,
		 * we can be preempted at any time after this splx().
		 */
		KASSERT(l->l_cpu == ci);
		KASSERT(ci->ci_mtx_count == -1);
		ci->ci_mtx_count = 0;
		splx(oldspl);
	} else {
		/* Nothing to do - just unlock and return. */
		mutex_spin_exit(spc->spc_mutex);
		l->l_pflag &= ~LP_PREEMPTING;
		lwp_unlock(l);
	}

	KASSERT(l == curlwp);
	KASSERT(l->l_stat == LSONPROC || (l->l_flag & LW_IDLE) != 0);

	SYSCALL_TIME_WAKEUP(l);
	LOCKDEBUG_BARRIER(NULL, 1);
}

/*
 * setrunnable: change LWP state to be runnable, placing it on the run queue.
 *
 * Call with the process and LWP locked.  Will return with the LWP unlocked.
 */
void
setrunnable(struct lwp *l)
{
	struct proc *p = l->l_proc;
	struct cpu_info *ci;
	kmutex_t *oldlock;

	KASSERT((l->l_flag & LW_IDLE) == 0);
	KASSERT((l->l_flag & LW_DBGSUSPEND) == 0);
	KASSERT(mutex_owned(p->p_lock));
	KASSERT(lwp_locked(l, NULL));
	KASSERT(l->l_mutex != l->l_cpu->ci_schedstate.spc_mutex);

	switch (l->l_stat) {
	case LSSTOP:
		/*
		 * If we're being traced (possibly because someone attached us
		 * while we were stopped), check for a signal from the debugger.
		 */
		if ((p->p_slflag & PSL_TRACED) != 0 && p->p_xsig != 0)
			signotify(l);
		p->p_nrlwps++;
		break;
	case LSSUSPENDED:
		KASSERT(lwp_locked(l, l->l_cpu->ci_schedstate.spc_lwplock));
		l->l_flag &= ~LW_WSUSPEND;
		p->p_nrlwps++;
		cv_broadcast(&p->p_lwpcv);
		break;
	case LSSLEEP:
		KASSERT(l->l_wchan != NULL);
		break;
	case LSIDL:
		KASSERT(lwp_locked(l, l->l_cpu->ci_schedstate.spc_lwplock));
		break;
	default:
		panic("setrunnable: lwp %p state was %d", l, l->l_stat);
	}

	/*
	 * If the LWP was sleeping, start it again.
	 */
	if (l->l_wchan != NULL) {
		l->l_stat = LSSLEEP;
		/* lwp_unsleep() will release the lock. */
		lwp_unsleep(l, true);
		return;
	}

	/*
	 * If the LWP is still on the CPU, mark it as LSONPROC.  It may be
	 * about to call mi_switch(), in which case it will yield.
	 */
	if ((l->l_pflag & LP_RUNNING) != 0) {
		l->l_stat = LSONPROC;
		l->l_slptime = 0;
		lwp_unlock(l);
		return;
	}

	/*
	 * Look for a CPU to run.
	 * Set the LWP runnable.
	 */
	ci = sched_takecpu(l);
	l->l_cpu = ci;
	spc_lock(ci);
	oldlock = lwp_setlock(l, l->l_cpu->ci_schedstate.spc_mutex);
	sched_setrunnable(l);
	l->l_stat = LSRUN;
	l->l_slptime = 0;
	sched_enqueue(l);
	sched_resched_lwp(l, true);
	/* SPC & LWP now unlocked. */
	mutex_spin_exit(oldlock);
}

/*
 * suspendsched:
 *
 *	Convert all non-LW_SYSTEM LSSLEEP or LSRUN LWPs to LSSUSPENDED.
 */
void
suspendsched(void)
{
	CPU_INFO_ITERATOR cii;
	struct cpu_info *ci;
	struct lwp *l;
	struct proc *p;

	/*
	 * We do this by process in order not to violate the locking rules.
	 */
	mutex_enter(&proc_lock);
	PROCLIST_FOREACH(p, &allproc) {
		mutex_enter(p->p_lock);
		if ((p->p_flag & PK_SYSTEM) != 0) {
			mutex_exit(p->p_lock);
			continue;
		}

		if (p->p_stat != SSTOP) {
			if (p->p_stat != SZOMB && p->p_stat != SDEAD) {
				p->p_pptr->p_nstopchild++;
				p->p_waited = 0;
			}
			p->p_stat = SSTOP;
		}

		LIST_FOREACH(l, &p->p_lwps, l_sibling) {
			if (l == curlwp)
				continue;

			lwp_lock(l);

			/*
			 * Set L_WREBOOT so that the LWP will suspend itself
			 * when it tries to return to user mode.  We want to
			 * try and get to get as many LWPs as possible to
			 * the user / kernel boundary, so that they will
			 * release any locks that they hold.
			 */
			l->l_flag |= (LW_WREBOOT | LW_WSUSPEND);

			if (l->l_stat == LSSLEEP &&
			    (l->l_flag & LW_SINTR) != 0) {
				/* setrunnable() will release the lock. */
				setrunnable(l);
				continue;
			}

			lwp_unlock(l);
		}

		mutex_exit(p->p_lock);
	}
	mutex_exit(&proc_lock);

	/*
	 * Kick all CPUs to make them preempt any LWPs running in user mode.
	 * They'll trap into the kernel and suspend themselves in userret().
	 *
	 * Unusually, we don't hold any other scheduler object locked, which
	 * would keep preemption off for sched_resched_cpu(), so disable it
	 * explicitly.
	 */
	kpreempt_disable();
	for (CPU_INFO_FOREACH(cii, ci)) {
		spc_lock(ci);
		sched_resched_cpu(ci, PRI_KERNEL, true);
		/* spc now unlocked */
	}
	kpreempt_enable();
}

/*
 * sched_unsleep:
 *
 *	The is called when the LWP has not been awoken normally but instead
 *	interrupted: for example, if the sleep timed out.  Because of this,
 *	it's not a valid action for running or idle LWPs.
 */
static void
sched_unsleep(struct lwp *l, bool cleanup)
{

	lwp_unlock(l);
	panic("sched_unsleep");
}

static void
sched_changepri(struct lwp *l, pri_t pri)
{
	struct schedstate_percpu *spc;
	struct cpu_info *ci;

	KASSERT(lwp_locked(l, NULL));

	ci = l->l_cpu;
	spc = &ci->ci_schedstate;

	if (l->l_stat == LSRUN) {
		KASSERT(lwp_locked(l, spc->spc_mutex));
		sched_dequeue(l);
		l->l_priority = pri;
		sched_enqueue(l);
		sched_resched_lwp(l, false);
	} else if (l->l_stat == LSONPROC && l->l_class != SCHED_OTHER) {
		/* On priority drop, only evict realtime LWPs. */
		KASSERT(lwp_locked(l, spc->spc_lwplock));
		l->l_priority = pri;
		spc_lock(ci);
		sched_resched_cpu(ci, spc->spc_maxpriority, true);
		/* spc now unlocked */
	} else {
		l->l_priority = pri;
	}
}

static void
sched_lendpri(struct lwp *l, pri_t pri)
{
	struct schedstate_percpu *spc;
	struct cpu_info *ci;

	KASSERT(lwp_locked(l, NULL));

	ci = l->l_cpu;
	spc = &ci->ci_schedstate;

	if (l->l_stat == LSRUN) {
		KASSERT(lwp_locked(l, spc->spc_mutex));
		sched_dequeue(l);
		l->l_inheritedprio = pri;
		l->l_auxprio = MAX(l->l_inheritedprio, l->l_protectprio);
		sched_enqueue(l);
		sched_resched_lwp(l, false);
	} else if (l->l_stat == LSONPROC && l->l_class != SCHED_OTHER) {
		/* On priority drop, only evict realtime LWPs. */
		KASSERT(lwp_locked(l, spc->spc_lwplock));
		l->l_inheritedprio = pri;
		l->l_auxprio = MAX(l->l_inheritedprio, l->l_protectprio);
		spc_lock(ci);
		sched_resched_cpu(ci, spc->spc_maxpriority, true);
		/* spc now unlocked */
	} else {
		l->l_inheritedprio = pri;
		l->l_auxprio = MAX(l->l_inheritedprio, l->l_protectprio);
	}
}

struct lwp *
syncobj_noowner(wchan_t wchan)
{

	return NULL;
}

/* Decay 95% of proc::p_pctcpu in 60 seconds, ccpu = exp(-1/20) */
const fixpt_t ccpu = 0.95122942450071400909 * FSCALE;

/*
 * Constants for averages over 1, 5 and 15 minutes when sampling at
 * 5 second intervals.
 */
static const fixpt_t cexp[ ] = {
	0.9200444146293232 * FSCALE,	/* exp(-1/12) */
	0.9834714538216174 * FSCALE,	/* exp(-1/60) */
	0.9944598480048967 * FSCALE,	/* exp(-1/180) */
};

/*
 * sched_pstats:
 *
 * => Update process statistics and check CPU resource allocation.
 * => Call scheduler-specific hook to eventually adjust LWP priorities.
 * => Compute load average of a quantity on 1, 5 and 15 minute intervals.
 */
void
sched_pstats(void)
{
	extern struct loadavg averunnable;
	struct loadavg *avg = &averunnable;
	const int clkhz = (stathz != 0 ? stathz : hz);
	static bool backwards = false;
	static u_int lavg_count = 0;
	struct proc *p;
	int nrun;

	sched_pstats_ticks++;
	if (++lavg_count >= 5) {
		lavg_count = 0;
		nrun = 0;
	}
	mutex_enter(&proc_lock);
	PROCLIST_FOREACH(p, &allproc) {
		struct lwp *l;
		struct rlimit *rlim;
		time_t runtm;
		int sig;

		/* Increment sleep time (if sleeping), ignore overflow. */
		mutex_enter(p->p_lock);
		runtm = p->p_rtime.sec;
		LIST_FOREACH(l, &p->p_lwps, l_sibling) {
			fixpt_t lpctcpu;
			u_int lcpticks;

			if (__predict_false((l->l_flag & LW_IDLE) != 0))
				continue;
			lwp_lock(l);
			runtm += l->l_rtime.sec;
			l->l_swtime++;
			sched_lwp_stats(l);

			/* For load average calculation. */
			if (__predict_false(lavg_count == 0) &&
			    (l->l_flag & (LW_SINTR | LW_SYSTEM)) == 0) {
				switch (l->l_stat) {
				case LSSLEEP:
					if (l->l_slptime > 1) {
						break;
					}
					/* FALLTHROUGH */
				case LSRUN:
				case LSONPROC:
				case LSIDL:
					nrun++;
				}
			}
			lwp_unlock(l);

			l->l_pctcpu = (l->l_pctcpu * ccpu) >> FSHIFT;
			if (l->l_slptime != 0)
				continue;

			lpctcpu = l->l_pctcpu;
			lcpticks = atomic_swap_uint(&l->l_cpticks, 0);
			lpctcpu += ((FSCALE - ccpu) *
			    (lcpticks * FSCALE / clkhz)) >> FSHIFT;
			l->l_pctcpu = lpctcpu;
		}
		/* Calculating p_pctcpu only for ps(1) */
		p->p_pctcpu = (p->p_pctcpu * ccpu) >> FSHIFT;

		if (__predict_false(runtm < 0)) {
			if (!backwards) {
				backwards = true;
				printf("WARNING: negative runtime; "
				    "monotonic clock has gone backwards\n");
			}
			mutex_exit(p->p_lock);
			continue;
		}

		/*
		 * Check if the process exceeds its CPU resource allocation.
		 * If over the hard limit, kill it with SIGKILL.
		 * If over the soft limit, send SIGXCPU and raise
		 * the soft limit a little.
		 */
		rlim = &p->p_rlimit[RLIMIT_CPU];
		sig = 0;
		if (__predict_false(runtm >= rlim->rlim_cur)) {
			if (runtm >= rlim->rlim_max) {
				sig = SIGKILL;
				log(LOG_NOTICE,
				    "pid %d, command %s, is killed: %s\n",
				    p->p_pid, p->p_comm, "exceeded RLIMIT_CPU");
				uprintf("pid %d, command %s, is killed: %s\n",
				    p->p_pid, p->p_comm, "exceeded RLIMIT_CPU");
			} else {
				sig = SIGXCPU;
				if (rlim->rlim_cur < rlim->rlim_max)
					rlim->rlim_cur += 5;
			}
		}
		mutex_exit(p->p_lock);
		if (__predict_false(sig)) {
			KASSERT((p->p_flag & PK_SYSTEM) == 0);
			psignal(p, sig);
		}
	}

	/* Load average calculation. */
	if (__predict_false(lavg_count == 0)) {
		int i;
		CTASSERT(__arraycount(cexp) == __arraycount(avg->ldavg));
		for (i = 0; i < __arraycount(cexp); i++) {
			avg->ldavg[i] = (cexp[i] * avg->ldavg[i] +
			    nrun * FSCALE * (FSCALE - cexp[i])) >> FSHIFT;
		}
	}

	/* Lightning bolt. */
	cv_broadcast(&lbolt);

	mutex_exit(&proc_lock);
}