Home | History | Annotate | Line # | Download | only in kern
kern_synch.c revision 1.149
      1 /*	$NetBSD: kern_synch.c,v 1.149 2005/05/29 22:24:15 christos Exp $	*/
      2 
      3 /*-
      4  * Copyright (c) 1999, 2000, 2004 The NetBSD Foundation, Inc.
      5  * All rights reserved.
      6  *
      7  * This code is derived from software contributed to The NetBSD Foundation
      8  * by Jason R. Thorpe of the Numerical Aerospace Simulation Facility,
      9  * NASA Ames Research Center.
     10  * This code is derived from software contributed to The NetBSD Foundation
     11  * by Charles M. Hannum.
     12  *
     13  * Redistribution and use in source and binary forms, with or without
     14  * modification, are permitted provided that the following conditions
     15  * are met:
     16  * 1. Redistributions of source code must retain the above copyright
     17  *    notice, this list of conditions and the following disclaimer.
     18  * 2. Redistributions in binary form must reproduce the above copyright
     19  *    notice, this list of conditions and the following disclaimer in the
     20  *    documentation and/or other materials provided with the distribution.
     21  * 3. All advertising materials mentioning features or use of this software
     22  *    must display the following acknowledgement:
     23  *	This product includes software developed by the NetBSD
     24  *	Foundation, Inc. and its contributors.
     25  * 4. Neither the name of The NetBSD Foundation nor the names of its
     26  *    contributors may be used to endorse or promote products derived
     27  *    from this software without specific prior written permission.
     28  *
     29  * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
     30  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
     31  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
     32  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
     33  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
     34  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
     35  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
     36  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
     37  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
     38  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
     39  * POSSIBILITY OF SUCH DAMAGE.
     40  */
     41 
     42 /*-
     43  * Copyright (c) 1982, 1986, 1990, 1991, 1993
     44  *	The Regents of the University of California.  All rights reserved.
     45  * (c) UNIX System Laboratories, Inc.
     46  * All or some portions of this file are derived from material licensed
     47  * to the University of California by American Telephone and Telegraph
     48  * Co. or Unix System Laboratories, Inc. and are reproduced herein with
     49  * the permission of UNIX System Laboratories, Inc.
     50  *
     51  * Redistribution and use in source and binary forms, with or without
     52  * modification, are permitted provided that the following conditions
     53  * are met:
     54  * 1. Redistributions of source code must retain the above copyright
     55  *    notice, this list of conditions and the following disclaimer.
     56  * 2. Redistributions in binary form must reproduce the above copyright
     57  *    notice, this list of conditions and the following disclaimer in the
     58  *    documentation and/or other materials provided with the distribution.
     59  * 3. Neither the name of the University nor the names of its contributors
     60  *    may be used to endorse or promote products derived from this software
     61  *    without specific prior written permission.
     62  *
     63  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
     64  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     65  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     66  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
     67  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     68  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     69  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     70  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     71  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     72  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     73  * SUCH DAMAGE.
     74  *
     75  *	@(#)kern_synch.c	8.9 (Berkeley) 5/19/95
     76  */
     77 
     78 #include <sys/cdefs.h>
     79 __KERNEL_RCSID(0, "$NetBSD: kern_synch.c,v 1.149 2005/05/29 22:24:15 christos Exp $");
     80 
     81 #include "opt_ddb.h"
     82 #include "opt_ktrace.h"
     83 #include "opt_kstack.h"
     84 #include "opt_lockdebug.h"
     85 #include "opt_multiprocessor.h"
     86 #include "opt_perfctrs.h"
     87 
     88 #include <sys/param.h>
     89 #include <sys/systm.h>
     90 #include <sys/callout.h>
     91 #include <sys/proc.h>
     92 #include <sys/kernel.h>
     93 #include <sys/buf.h>
     94 #if defined(PERFCTRS)
     95 #include <sys/pmc.h>
     96 #endif
     97 #include <sys/signalvar.h>
     98 #include <sys/resourcevar.h>
     99 #include <sys/sched.h>
    100 #include <sys/sa.h>
    101 #include <sys/savar.h>
    102 
    103 #include <uvm/uvm_extern.h>
    104 
    105 #ifdef KTRACE
    106 #include <sys/ktrace.h>
    107 #endif
    108 
    109 #include <machine/cpu.h>
    110 
    111 int	lbolt;			/* once a second sleep address */
    112 int	rrticks;		/* number of hardclock ticks per roundrobin() */
    113 
    114 /*
    115  * The global scheduler state.
    116  */
    117 struct prochd sched_qs[RUNQUE_NQS];	/* run queues */
    118 __volatile u_int32_t sched_whichqs;	/* bitmap of non-empty queues */
    119 struct slpque sched_slpque[SLPQUE_TABLESIZE]; /* sleep queues */
    120 
    121 struct simplelock sched_lock = SIMPLELOCK_INITIALIZER;
    122 
    123 void schedcpu(void *);
    124 void updatepri(struct lwp *);
    125 void endtsleep(void *);
    126 
    127 __inline void sa_awaken(struct lwp *);
    128 __inline void awaken(struct lwp *);
    129 
    130 struct callout schedcpu_ch = CALLOUT_INITIALIZER_SETFUNC(schedcpu, NULL);
    131 
    132 
    133 
    134 /*
    135  * Force switch among equal priority processes every 100ms.
    136  * Called from hardclock every hz/10 == rrticks hardclock ticks.
    137  */
    138 /* ARGSUSED */
    139 void
    140 roundrobin(struct cpu_info *ci)
    141 {
    142 	struct schedstate_percpu *spc = &ci->ci_schedstate;
    143 
    144 	spc->spc_rrticks = rrticks;
    145 
    146 	if (curlwp != NULL) {
    147 		if (spc->spc_flags & SPCF_SEENRR) {
    148 			/*
    149 			 * The process has already been through a roundrobin
    150 			 * without switching and may be hogging the CPU.
    151 			 * Indicate that the process should yield.
    152 			 */
    153 			spc->spc_flags |= SPCF_SHOULDYIELD;
    154 		} else
    155 			spc->spc_flags |= SPCF_SEENRR;
    156 	}
    157 	need_resched(curcpu());
    158 }
    159 
    160 /*
    161  * Constants for digital decay and forget:
    162  *	90% of (p_estcpu) usage in 5 * loadav time
    163  *	95% of (p_pctcpu) usage in 60 seconds (load insensitive)
    164  *          Note that, as ps(1) mentions, this can let percentages
    165  *          total over 100% (I've seen 137.9% for 3 processes).
    166  *
    167  * Note that hardclock updates p_estcpu and p_cpticks independently.
    168  *
    169  * We wish to decay away 90% of p_estcpu in (5 * loadavg) seconds.
    170  * That is, the system wants to compute a value of decay such
    171  * that the following for loop:
    172  * 	for (i = 0; i < (5 * loadavg); i++)
    173  * 		p_estcpu *= decay;
    174  * will compute
    175  * 	p_estcpu *= 0.1;
    176  * for all values of loadavg:
    177  *
    178  * Mathematically this loop can be expressed by saying:
    179  * 	decay ** (5 * loadavg) ~= .1
    180  *
    181  * The system computes decay as:
    182  * 	decay = (2 * loadavg) / (2 * loadavg + 1)
    183  *
    184  * We wish to prove that the system's computation of decay
    185  * will always fulfill the equation:
    186  * 	decay ** (5 * loadavg) ~= .1
    187  *
    188  * If we compute b as:
    189  * 	b = 2 * loadavg
    190  * then
    191  * 	decay = b / (b + 1)
    192  *
    193  * We now need to prove two things:
    194  *	1) Given factor ** (5 * loadavg) ~= .1, prove factor == b/(b+1)
    195  *	2) Given b/(b+1) ** power ~= .1, prove power == (5 * loadavg)
    196  *
    197  * Facts:
    198  *         For x close to zero, exp(x) =~ 1 + x, since
    199  *              exp(x) = 0! + x**1/1! + x**2/2! + ... .
    200  *              therefore exp(-1/b) =~ 1 - (1/b) = (b-1)/b.
    201  *         For x close to zero, ln(1+x) =~ x, since
    202  *              ln(1+x) = x - x**2/2 + x**3/3 - ...     -1 < x < 1
    203  *              therefore ln(b/(b+1)) = ln(1 - 1/(b+1)) =~ -1/(b+1).
    204  *         ln(.1) =~ -2.30
    205  *
    206  * Proof of (1):
    207  *    Solve (factor)**(power) =~ .1 given power (5*loadav):
    208  *	solving for factor,
    209  *      ln(factor) =~ (-2.30/5*loadav), or
    210  *      factor =~ exp(-1/((5/2.30)*loadav)) =~ exp(-1/(2*loadav)) =
    211  *          exp(-1/b) =~ (b-1)/b =~ b/(b+1).                    QED
    212  *
    213  * Proof of (2):
    214  *    Solve (factor)**(power) =~ .1 given factor == (b/(b+1)):
    215  *	solving for power,
    216  *      power*ln(b/(b+1)) =~ -2.30, or
    217  *      power =~ 2.3 * (b + 1) = 4.6*loadav + 2.3 =~ 5*loadav.  QED
    218  *
    219  * Actual power values for the implemented algorithm are as follows:
    220  *      loadav: 1       2       3       4
    221  *      power:  5.68    10.32   14.94   19.55
    222  */
    223 
    224 /* calculations for digital decay to forget 90% of usage in 5*loadav sec */
    225 #define	loadfactor(loadav)	(2 * (loadav))
    226 #define	decay_cpu(loadfac, cpu)	(((loadfac) * (cpu)) / ((loadfac) + FSCALE))
    227 
    228 /* decay 95% of `p_pctcpu' in 60 seconds; see CCPU_SHIFT before changing */
    229 fixpt_t	ccpu = 0.95122942450071400909 * FSCALE;		/* exp(-1/20) */
    230 
    231 /*
    232  * If `ccpu' is not equal to `exp(-1/20)' and you still want to use the
    233  * faster/more-accurate formula, you'll have to estimate CCPU_SHIFT below
    234  * and possibly adjust FSHIFT in "param.h" so that (FSHIFT >= CCPU_SHIFT).
    235  *
    236  * To estimate CCPU_SHIFT for exp(-1/20), the following formula was used:
    237  *	1 - exp(-1/20) ~= 0.0487 ~= 0.0488 == 1 (fixed pt, *11* bits).
    238  *
    239  * If you dont want to bother with the faster/more-accurate formula, you
    240  * can set CCPU_SHIFT to (FSHIFT + 1) which will use a slower/less-accurate
    241  * (more general) method of calculating the %age of CPU used by a process.
    242  */
    243 #define	CCPU_SHIFT	11
    244 
    245 /*
    246  * Recompute process priorities, every hz ticks.
    247  */
    248 /* ARGSUSED */
    249 void
    250 schedcpu(void *arg)
    251 {
    252 	fixpt_t loadfac = loadfactor(averunnable.ldavg[0]);
    253 	struct lwp *l;
    254 	struct proc *p;
    255 	int s, minslp;
    256 	unsigned int newcpu;
    257 	int clkhz;
    258 
    259 	proclist_lock_read();
    260 	PROCLIST_FOREACH(p, &allproc) {
    261 		/*
    262 		 * Increment time in/out of memory and sleep time
    263 		 * (if sleeping).  We ignore overflow; with 16-bit int's
    264 		 * (remember them?) overflow takes 45 days.
    265 		 */
    266 		minslp = 2;
    267 		LIST_FOREACH(l, &p->p_lwps, l_sibling) {
    268 			l->l_swtime++;
    269 			if (l->l_stat == LSSLEEP || l->l_stat == LSSTOP ||
    270 			    l->l_stat == LSSUSPENDED) {
    271 				l->l_slptime++;
    272 				minslp = min(minslp, l->l_slptime);
    273 			} else
    274 				minslp = 0;
    275 		}
    276 		p->p_pctcpu = (p->p_pctcpu * ccpu) >> FSHIFT;
    277 		/*
    278 		 * If the process has slept the entire second,
    279 		 * stop recalculating its priority until it wakes up.
    280 		 */
    281 		if (minslp > 1)
    282 			continue;
    283 		s = splstatclock();	/* prevent state changes */
    284 		/*
    285 		 * p_pctcpu is only for ps.
    286 		 */
    287 		clkhz = stathz != 0 ? stathz : hz;
    288 #if	(FSHIFT >= CCPU_SHIFT)
    289 		p->p_pctcpu += (clkhz == 100)?
    290 			((fixpt_t) p->p_cpticks) << (FSHIFT - CCPU_SHIFT):
    291                 	100 * (((fixpt_t) p->p_cpticks)
    292 				<< (FSHIFT - CCPU_SHIFT)) / clkhz;
    293 #else
    294 		p->p_pctcpu += ((FSCALE - ccpu) *
    295 			(p->p_cpticks * FSCALE / clkhz)) >> FSHIFT;
    296 #endif
    297 		p->p_cpticks = 0;
    298 		newcpu = (u_int)decay_cpu(loadfac, p->p_estcpu);
    299 		p->p_estcpu = newcpu;
    300 		splx(s);	/* Done with the process CPU ticks update */
    301 		SCHED_LOCK(s);
    302 		LIST_FOREACH(l, &p->p_lwps, l_sibling) {
    303 			if (l->l_slptime > 1)
    304 				continue;
    305 			resetpriority(l);
    306 			if (l->l_priority >= PUSER) {
    307 				if (l->l_stat == LSRUN &&
    308 				    (l->l_flag & L_INMEM) &&
    309 				    (l->l_priority / PPQ) != (l->l_usrpri / PPQ)) {
    310 					remrunqueue(l);
    311 					l->l_priority = l->l_usrpri;
    312 					setrunqueue(l);
    313 				} else
    314 					l->l_priority = l->l_usrpri;
    315 			}
    316 		}
    317 		SCHED_UNLOCK(s);
    318 	}
    319 	proclist_unlock_read();
    320 	uvm_meter();
    321 	wakeup((caddr_t)&lbolt);
    322 	callout_schedule(&schedcpu_ch, hz);
    323 }
    324 
    325 /*
    326  * Recalculate the priority of a process after it has slept for a while.
    327  * For all load averages >= 1 and max p_estcpu of 255, sleeping for at
    328  * least six times the loadfactor will decay p_estcpu to zero.
    329  */
    330 void
    331 updatepri(struct lwp *l)
    332 {
    333 	struct proc *p = l->l_proc;
    334 	unsigned int newcpu;
    335 	fixpt_t loadfac;
    336 
    337 	SCHED_ASSERT_LOCKED();
    338 
    339 	newcpu = p->p_estcpu;
    340 	loadfac = loadfactor(averunnable.ldavg[0]);
    341 
    342 	if (l->l_slptime > 5 * loadfac)
    343 		p->p_estcpu = 0; /* XXX NJWLWP */
    344 	else {
    345 		l->l_slptime--;	/* the first time was done in schedcpu */
    346 		while (newcpu && --l->l_slptime)
    347 			newcpu = (int) decay_cpu(loadfac, newcpu);
    348 		p->p_estcpu = newcpu;
    349 	}
    350 	resetpriority(l);
    351 }
    352 
    353 /*
    354  * During autoconfiguration or after a panic, a sleep will simply
    355  * lower the priority briefly to allow interrupts, then return.
    356  * The priority to be used (safepri) is machine-dependent, thus this
    357  * value is initialized and maintained in the machine-dependent layers.
    358  * This priority will typically be 0, or the lowest priority
    359  * that is safe for use on the interrupt stack; it can be made
    360  * higher to block network software interrupts after panics.
    361  */
    362 int safepri;
    363 
    364 /*
    365  * General sleep call.  Suspends the current process until a wakeup is
    366  * performed on the specified identifier.  The process will then be made
    367  * runnable with the specified priority.  Sleeps at most timo/hz seconds
    368  * (0 means no timeout).  If pri includes PCATCH flag, signals are checked
    369  * before and after sleeping, else signals are not checked.  Returns 0 if
    370  * awakened, EWOULDBLOCK if the timeout expires.  If PCATCH is set and a
    371  * signal needs to be delivered, ERESTART is returned if the current system
    372  * call should be restarted if possible, and EINTR is returned if the system
    373  * call should be interrupted by the signal (return EINTR).
    374  *
    375  * The interlock is held until the scheduler_slock is acquired.  The
    376  * interlock will be locked before returning back to the caller
    377  * unless the PNORELOCK flag is specified, in which case the
    378  * interlock will always be unlocked upon return.
    379  */
    380 int
    381 ltsleep(__volatile const void *ident, int priority, const char *wmesg, int timo,
    382     __volatile struct simplelock *interlock)
    383 {
    384 	struct lwp *l = curlwp;
    385 	struct proc *p = l ? l->l_proc : NULL;
    386 	struct slpque *qp;
    387 	int sig, s;
    388 	int catch = priority & PCATCH;
    389 	int relock = (priority & PNORELOCK) == 0;
    390 	int exiterr = (priority & PNOEXITERR) == 0;
    391 
    392 	/*
    393 	 * XXXSMP
    394 	 * This is probably bogus.  Figure out what the right
    395 	 * thing to do here really is.
    396 	 * Note that not sleeping if ltsleep is called with curlwp == NULL
    397 	 * in the shutdown case is disgusting but partly necessary given
    398 	 * how shutdown (barely) works.
    399 	 */
    400 	if (cold || (doing_shutdown && (panicstr || (l == NULL)))) {
    401 		/*
    402 		 * After a panic, or during autoconfiguration,
    403 		 * just give interrupts a chance, then just return;
    404 		 * don't run any other procs or panic below,
    405 		 * in case this is the idle process and already asleep.
    406 		 */
    407 		s = splhigh();
    408 		splx(safepri);
    409 		splx(s);
    410 		if (interlock != NULL && relock == 0)
    411 			simple_unlock(interlock);
    412 		return (0);
    413 	}
    414 
    415 	KASSERT(p != NULL);
    416 	LOCK_ASSERT(interlock == NULL || simple_lock_held(interlock));
    417 
    418 #ifdef KTRACE
    419 	if (KTRPOINT(p, KTR_CSW))
    420 		ktrcsw(p, 1, 0);
    421 #endif
    422 
    423 	SCHED_LOCK(s);
    424 
    425 #ifdef DIAGNOSTIC
    426 	if (ident == NULL)
    427 		panic("ltsleep: ident == NULL");
    428 	if (l->l_stat != LSONPROC)
    429 		panic("ltsleep: l_stat %d != LSONPROC", l->l_stat);
    430 	if (l->l_back != NULL)
    431 		panic("ltsleep: p_back != NULL");
    432 #endif
    433 
    434 	l->l_wchan = ident;
    435 	l->l_wmesg = wmesg;
    436 	l->l_slptime = 0;
    437 	l->l_priority = priority & PRIMASK;
    438 
    439 	qp = SLPQUE(ident);
    440 	if (qp->sq_head == 0)
    441 		qp->sq_head = l;
    442 	else {
    443 		*qp->sq_tailp = l;
    444 	}
    445 	*(qp->sq_tailp = &l->l_forw) = 0;
    446 
    447 	if (timo)
    448 		callout_reset(&l->l_tsleep_ch, timo, endtsleep, l);
    449 
    450 	/*
    451 	 * We can now release the interlock; the scheduler_slock
    452 	 * is held, so a thread can't get in to do wakeup() before
    453 	 * we do the switch.
    454 	 *
    455 	 * XXX We leave the code block here, after inserting ourselves
    456 	 * on the sleep queue, because we might want a more clever
    457 	 * data structure for the sleep queues at some point.
    458 	 */
    459 	if (interlock != NULL)
    460 		simple_unlock(interlock);
    461 
    462 	/*
    463 	 * We put ourselves on the sleep queue and start our timeout
    464 	 * before calling CURSIG, as we could stop there, and a wakeup
    465 	 * or a SIGCONT (or both) could occur while we were stopped.
    466 	 * A SIGCONT would cause us to be marked as SSLEEP
    467 	 * without resuming us, thus we must be ready for sleep
    468 	 * when CURSIG is called.  If the wakeup happens while we're
    469 	 * stopped, p->p_wchan will be 0 upon return from CURSIG.
    470 	 */
    471 	if (catch) {
    472 		l->l_flag |= L_SINTR;
    473 		if (((sig = CURSIG(l)) != 0) ||
    474 		    ((p->p_flag & P_WEXIT) && p->p_nlwps > 1)) {
    475 			if (l->l_wchan != NULL)
    476 				unsleep(l);
    477 			l->l_stat = LSONPROC;
    478 			SCHED_UNLOCK(s);
    479 			goto resume;
    480 		}
    481 		if (l->l_wchan == NULL) {
    482 			catch = 0;
    483 			SCHED_UNLOCK(s);
    484 			goto resume;
    485 		}
    486 	} else
    487 		sig = 0;
    488 	l->l_stat = LSSLEEP;
    489 	p->p_nrlwps--;
    490 	p->p_stats->p_ru.ru_nvcsw++;
    491 	SCHED_ASSERT_LOCKED();
    492 	if (l->l_flag & L_SA)
    493 		sa_switch(l, SA_UPCALL_BLOCKED);
    494 	else
    495 		mi_switch(l, NULL);
    496 
    497 #if	defined(DDB) && !defined(GPROF)
    498 	/* handy breakpoint location after process "wakes" */
    499 	__asm(".globl bpendtsleep\nbpendtsleep:");
    500 #endif
    501 	/*
    502 	 * p->p_nrlwps is incremented by whoever made us runnable again,
    503 	 * either setrunnable() or awaken().
    504 	 */
    505 
    506 	SCHED_ASSERT_UNLOCKED();
    507 	splx(s);
    508 
    509  resume:
    510 	KDASSERT(l->l_cpu != NULL);
    511 	KDASSERT(l->l_cpu == curcpu());
    512 	l->l_cpu->ci_schedstate.spc_curpriority = l->l_usrpri;
    513 
    514 	l->l_flag &= ~L_SINTR;
    515 	if (l->l_flag & L_TIMEOUT) {
    516 		l->l_flag &= ~(L_TIMEOUT|L_CANCELLED);
    517 		if (sig == 0) {
    518 #ifdef KTRACE
    519 			if (KTRPOINT(p, KTR_CSW))
    520 				ktrcsw(p, 0, 0);
    521 #endif
    522 			if (relock && interlock != NULL)
    523 				simple_lock(interlock);
    524 			return (EWOULDBLOCK);
    525 		}
    526 	} else if (timo)
    527 		callout_stop(&l->l_tsleep_ch);
    528 
    529 	if (catch) {
    530 		const int cancelled = l->l_flag & L_CANCELLED;
    531 		l->l_flag &= ~L_CANCELLED;
    532 		if (sig != 0 || (sig = CURSIG(l)) != 0 || cancelled) {
    533 #ifdef KTRACE
    534 			if (KTRPOINT(p, KTR_CSW))
    535 				ktrcsw(p, 0, 0);
    536 #endif
    537 			if (relock && interlock != NULL)
    538 				simple_lock(interlock);
    539 			/*
    540 			 * If this sleep was canceled, don't let the syscall
    541 			 * restart.
    542 			 */
    543 			if (cancelled ||
    544 			    (SIGACTION(p, sig).sa_flags & SA_RESTART) == 0)
    545 				return (EINTR);
    546 			return (ERESTART);
    547 		}
    548 	}
    549 
    550 #ifdef KTRACE
    551 	if (KTRPOINT(p, KTR_CSW))
    552 		ktrcsw(p, 0, 0);
    553 #endif
    554 	if (relock && interlock != NULL)
    555 		simple_lock(interlock);
    556 
    557 	/* XXXNJW this is very much a kluge.
    558 	 * revisit. a better way of preventing looping/hanging syscalls like
    559 	 * wait4() and _lwp_wait() from wedging an exiting process
    560 	 * would be preferred.
    561 	 */
    562 	if (catch && ((p->p_flag & P_WEXIT) && p->p_nlwps > 1 && exiterr))
    563 		return (EINTR);
    564 	return (0);
    565 }
    566 
    567 /*
    568  * Implement timeout for tsleep.
    569  * If process hasn't been awakened (wchan non-zero),
    570  * set timeout flag and undo the sleep.  If proc
    571  * is stopped, just unsleep so it will remain stopped.
    572  */
    573 void
    574 endtsleep(void *arg)
    575 {
    576 	struct lwp *l;
    577 	int s;
    578 
    579 	l = (struct lwp *)arg;
    580 	SCHED_LOCK(s);
    581 	if (l->l_wchan) {
    582 		if (l->l_stat == LSSLEEP)
    583 			setrunnable(l);
    584 		else
    585 			unsleep(l);
    586 		l->l_flag |= L_TIMEOUT;
    587 	}
    588 	SCHED_UNLOCK(s);
    589 }
    590 
    591 /*
    592  * Remove a process from its wait queue
    593  */
    594 void
    595 unsleep(struct lwp *l)
    596 {
    597 	struct slpque *qp;
    598 	struct lwp **hp;
    599 
    600 	SCHED_ASSERT_LOCKED();
    601 
    602 	if (l->l_wchan) {
    603 		hp = &(qp = SLPQUE(l->l_wchan))->sq_head;
    604 		while (*hp != l)
    605 			hp = &(*hp)->l_forw;
    606 		*hp = l->l_forw;
    607 		if (qp->sq_tailp == &l->l_forw)
    608 			qp->sq_tailp = hp;
    609 		l->l_wchan = 0;
    610 	}
    611 }
    612 
    613 __inline void
    614 sa_awaken(struct lwp *l)
    615 {
    616 
    617 	SCHED_ASSERT_LOCKED();
    618 
    619 	if (l == l->l_savp->savp_lwp && l->l_flag & L_SA_YIELD)
    620 		l->l_flag &= ~L_SA_IDLE;
    621 }
    622 
    623 /*
    624  * Optimized-for-wakeup() version of setrunnable().
    625  */
    626 __inline void
    627 awaken(struct lwp *l)
    628 {
    629 
    630 	SCHED_ASSERT_LOCKED();
    631 
    632 	if (l->l_proc->p_sa)
    633 		sa_awaken(l);
    634 
    635 	if (l->l_slptime > 1)
    636 		updatepri(l);
    637 	l->l_slptime = 0;
    638 	l->l_stat = LSRUN;
    639 	l->l_proc->p_nrlwps++;
    640 	/*
    641 	 * Since curpriority is a user priority, p->p_priority
    642 	 * is always better than curpriority on the last CPU on
    643 	 * which it ran.
    644 	 *
    645 	 * XXXSMP See affinity comment in resched_proc().
    646 	 */
    647 	if (l->l_flag & L_INMEM) {
    648 		setrunqueue(l);
    649 		KASSERT(l->l_cpu != NULL);
    650 		need_resched(l->l_cpu);
    651 	} else
    652 		sched_wakeup(&proc0);
    653 }
    654 
    655 #if defined(MULTIPROCESSOR) || defined(LOCKDEBUG)
    656 void
    657 sched_unlock_idle(void)
    658 {
    659 
    660 	simple_unlock(&sched_lock);
    661 }
    662 
    663 void
    664 sched_lock_idle(void)
    665 {
    666 
    667 	simple_lock(&sched_lock);
    668 }
    669 #endif /* MULTIPROCESSOR || LOCKDEBUG */
    670 
    671 /*
    672  * Make all processes sleeping on the specified identifier runnable.
    673  */
    674 
    675 void
    676 wakeup(__volatile const void *ident)
    677 {
    678 	int s;
    679 
    680 	SCHED_ASSERT_UNLOCKED();
    681 
    682 	SCHED_LOCK(s);
    683 	sched_wakeup(ident);
    684 	SCHED_UNLOCK(s);
    685 }
    686 
    687 void
    688 sched_wakeup(__volatile const void *ident)
    689 {
    690 	struct slpque *qp;
    691 	struct lwp *l, **q;
    692 
    693 	SCHED_ASSERT_LOCKED();
    694 
    695 	qp = SLPQUE(ident);
    696  restart:
    697 	for (q = &qp->sq_head; (l = *q) != NULL; ) {
    698 #ifdef DIAGNOSTIC
    699 		if (l->l_back || (l->l_stat != LSSLEEP &&
    700 		    l->l_stat != LSSTOP && l->l_stat != LSSUSPENDED))
    701 			panic("wakeup");
    702 #endif
    703 		if (l->l_wchan == ident) {
    704 			l->l_wchan = 0;
    705 			*q = l->l_forw;
    706 			if (qp->sq_tailp == &l->l_forw)
    707 				qp->sq_tailp = q;
    708 			if (l->l_stat == LSSLEEP) {
    709 				awaken(l);
    710 				goto restart;
    711 			}
    712 		} else
    713 			q = &l->l_forw;
    714 	}
    715 }
    716 
    717 /*
    718  * Make the highest priority process first in line on the specified
    719  * identifier runnable.
    720  */
    721 void
    722 wakeup_one(__volatile const void *ident)
    723 {
    724 	struct slpque *qp;
    725 	struct lwp *l, **q;
    726 	struct lwp *best_sleepp, **best_sleepq;
    727 	struct lwp *best_stopp, **best_stopq;
    728 	int s;
    729 
    730 	best_sleepp = best_stopp = NULL;
    731 	best_sleepq = best_stopq = NULL;
    732 
    733 	SCHED_LOCK(s);
    734 
    735 	qp = SLPQUE(ident);
    736 
    737 	for (q = &qp->sq_head; (l = *q) != NULL; q = &l->l_forw) {
    738 #ifdef DIAGNOSTIC
    739 		if (l->l_back || (l->l_stat != LSSLEEP &&
    740 		    l->l_stat != LSSTOP && l->l_stat != LSSUSPENDED))
    741 			panic("wakeup_one");
    742 #endif
    743 		if (l->l_wchan == ident) {
    744 			if (l->l_stat == LSSLEEP) {
    745 				if (best_sleepp == NULL ||
    746 				    l->l_priority < best_sleepp->l_priority) {
    747 					best_sleepp = l;
    748 					best_sleepq = q;
    749 				}
    750 			} else {
    751 				if (best_stopp == NULL ||
    752 				    l->l_priority < best_stopp->l_priority) {
    753 				    	best_stopp = l;
    754 					best_stopq = q;
    755 				}
    756 			}
    757 		}
    758 	}
    759 
    760 	/*
    761 	 * Consider any SSLEEP process higher than the highest priority SSTOP
    762 	 * process.
    763 	 */
    764 	if (best_sleepp != NULL) {
    765 		l = best_sleepp;
    766 		q = best_sleepq;
    767 	} else {
    768 		l = best_stopp;
    769 		q = best_stopq;
    770 	}
    771 
    772 	if (l != NULL) {
    773 		l->l_wchan = NULL;
    774 		*q = l->l_forw;
    775 		if (qp->sq_tailp == &l->l_forw)
    776 			qp->sq_tailp = q;
    777 		if (l->l_stat == LSSLEEP)
    778 			awaken(l);
    779 	}
    780 	SCHED_UNLOCK(s);
    781 }
    782 
    783 /*
    784  * General yield call.  Puts the current process back on its run queue and
    785  * performs a voluntary context switch.  Should only be called when the
    786  * current process explicitly requests it (eg sched_yield(2) in compat code).
    787  */
    788 void
    789 yield(void)
    790 {
    791 	struct lwp *l = curlwp;
    792 	int s;
    793 
    794 	SCHED_LOCK(s);
    795 	l->l_priority = l->l_usrpri;
    796 	l->l_stat = LSRUN;
    797 	setrunqueue(l);
    798 	l->l_proc->p_stats->p_ru.ru_nvcsw++;
    799 	mi_switch(l, NULL);
    800 	SCHED_ASSERT_UNLOCKED();
    801 	splx(s);
    802 }
    803 
    804 /*
    805  * General preemption call.  Puts the current process back on its run queue
    806  * and performs an involuntary context switch.  If a process is supplied,
    807  * we switch to that process.  Otherwise, we use the normal process selection
    808  * criteria.
    809  */
    810 
    811 void
    812 preempt(int more)
    813 {
    814 	struct lwp *l = curlwp;
    815 	int r, s;
    816 
    817 	SCHED_LOCK(s);
    818 	l->l_priority = l->l_usrpri;
    819 	l->l_stat = LSRUN;
    820 	setrunqueue(l);
    821 	l->l_proc->p_stats->p_ru.ru_nivcsw++;
    822 	r = mi_switch(l, NULL);
    823 	SCHED_ASSERT_UNLOCKED();
    824 	splx(s);
    825 	if ((l->l_flag & L_SA) != 0 && r != 0 && more == 0)
    826 		sa_preempt(l);
    827 }
    828 
    829 /*
    830  * The machine independent parts of context switch.
    831  * Must be called at splsched() (no higher!) and with
    832  * the sched_lock held.
    833  * Switch to "new" if non-NULL, otherwise let cpu_switch choose
    834  * the next lwp.
    835  *
    836  * Returns 1 if another process was actually run.
    837  */
    838 int
    839 mi_switch(struct lwp *l, struct lwp *newl)
    840 {
    841 	struct schedstate_percpu *spc;
    842 	struct rlimit *rlim;
    843 	long s, u;
    844 	struct timeval tv;
    845 	int hold_count;
    846 	struct proc *p = l->l_proc;
    847 	int retval;
    848 
    849 	SCHED_ASSERT_LOCKED();
    850 
    851 	/*
    852 	 * Release the kernel_lock, as we are about to yield the CPU.
    853 	 * The scheduler lock is still held until cpu_switch()
    854 	 * selects a new process and removes it from the run queue.
    855 	 */
    856 	hold_count = KERNEL_LOCK_RELEASE_ALL();
    857 
    858 	KDASSERT(l->l_cpu != NULL);
    859 	KDASSERT(l->l_cpu == curcpu());
    860 
    861 	spc = &l->l_cpu->ci_schedstate;
    862 
    863 #if defined(LOCKDEBUG) || defined(DIAGNOSTIC)
    864 	spinlock_switchcheck();
    865 #endif
    866 #ifdef LOCKDEBUG
    867 	simple_lock_switchcheck();
    868 #endif
    869 
    870 	/*
    871 	 * Compute the amount of time during which the current
    872 	 * process was running.
    873 	 */
    874 	microtime(&tv);
    875 	u = p->p_rtime.tv_usec +
    876 	    (tv.tv_usec - spc->spc_runtime.tv_usec);
    877 	s = p->p_rtime.tv_sec + (tv.tv_sec - spc->spc_runtime.tv_sec);
    878 	if (u < 0) {
    879 		u += 1000000;
    880 		s--;
    881 	} else if (u >= 1000000) {
    882 		u -= 1000000;
    883 		s++;
    884 	}
    885 	p->p_rtime.tv_usec = u;
    886 	p->p_rtime.tv_sec = s;
    887 
    888 	/*
    889 	 * Check if the process exceeds its CPU resource allocation.
    890 	 * If over max, kill it.  In any case, if it has run for more
    891 	 * than 10 minutes, reduce priority to give others a chance.
    892 	 */
    893 	rlim = &p->p_rlimit[RLIMIT_CPU];
    894 	if (s >= rlim->rlim_cur) {
    895 		/*
    896 		 * XXXSMP: we're inside the scheduler lock perimeter;
    897 		 * use sched_psignal.
    898 		 */
    899 		if (s >= rlim->rlim_max)
    900 			sched_psignal(p, SIGKILL);
    901 		else {
    902 			sched_psignal(p, SIGXCPU);
    903 			if (rlim->rlim_cur < rlim->rlim_max)
    904 				rlim->rlim_cur += 5;
    905 		}
    906 	}
    907 	if (autonicetime && s > autonicetime && p->p_ucred->cr_uid &&
    908 	    p->p_nice == NZERO) {
    909 		p->p_nice = autoniceval + NZERO;
    910 		resetpriority(l);
    911 	}
    912 
    913 	/*
    914 	 * Process is about to yield the CPU; clear the appropriate
    915 	 * scheduling flags.
    916 	 */
    917 	spc->spc_flags &= ~SPCF_SWITCHCLEAR;
    918 
    919 #ifdef KSTACK_CHECK_MAGIC
    920 	kstack_check_magic(l);
    921 #endif
    922 
    923 	/*
    924 	 * If we are using h/w performance counters, save context.
    925 	 */
    926 #if PERFCTRS
    927 	if (PMC_ENABLED(p))
    928 		pmc_save_context(p);
    929 #endif
    930 
    931 	/*
    932 	 * Switch to the new current process.  When we
    933 	 * run again, we'll return back here.
    934 	 */
    935 	uvmexp.swtch++;
    936 	if (newl == NULL) {
    937 		retval = cpu_switch(l, NULL);
    938 	} else {
    939 		remrunqueue(newl);
    940 		cpu_switchto(l, newl);
    941 		retval = 0;
    942 	}
    943 
    944 	/*
    945 	 * If we are using h/w performance counters, restore context.
    946 	 */
    947 #if PERFCTRS
    948 	if (PMC_ENABLED(p))
    949 		pmc_restore_context(p);
    950 #endif
    951 
    952 	/*
    953 	 * Make sure that MD code released the scheduler lock before
    954 	 * resuming us.
    955 	 */
    956 	SCHED_ASSERT_UNLOCKED();
    957 
    958 	/*
    959 	 * We're running again; record our new start time.  We might
    960 	 * be running on a new CPU now, so don't use the cache'd
    961 	 * schedstate_percpu pointer.
    962 	 */
    963 	KDASSERT(l->l_cpu != NULL);
    964 	KDASSERT(l->l_cpu == curcpu());
    965 	microtime(&l->l_cpu->ci_schedstate.spc_runtime);
    966 
    967 	/*
    968 	 * Reacquire the kernel_lock now.  We do this after we've
    969 	 * released the scheduler lock to avoid deadlock, and before
    970 	 * we reacquire the interlock.
    971 	 */
    972 	KERNEL_LOCK_ACQUIRE_COUNT(hold_count);
    973 
    974 	return retval;
    975 }
    976 
    977 /*
    978  * Initialize the (doubly-linked) run queues
    979  * to be empty.
    980  */
    981 void
    982 rqinit()
    983 {
    984 	int i;
    985 
    986 	for (i = 0; i < RUNQUE_NQS; i++)
    987 		sched_qs[i].ph_link = sched_qs[i].ph_rlink =
    988 		    (struct lwp *)&sched_qs[i];
    989 }
    990 
    991 static __inline void
    992 resched_proc(struct lwp *l, u_char pri)
    993 {
    994 	struct cpu_info *ci;
    995 
    996 	/*
    997 	 * XXXSMP
    998 	 * Since l->l_cpu persists across a context switch,
    999 	 * this gives us *very weak* processor affinity, in
   1000 	 * that we notify the CPU on which the process last
   1001 	 * ran that it should try to switch.
   1002 	 *
   1003 	 * This does not guarantee that the process will run on
   1004 	 * that processor next, because another processor might
   1005 	 * grab it the next time it performs a context switch.
   1006 	 *
   1007 	 * This also does not handle the case where its last
   1008 	 * CPU is running a higher-priority process, but every
   1009 	 * other CPU is running a lower-priority process.  There
   1010 	 * are ways to handle this situation, but they're not
   1011 	 * currently very pretty, and we also need to weigh the
   1012 	 * cost of moving a process from one CPU to another.
   1013 	 *
   1014 	 * XXXSMP
   1015 	 * There is also the issue of locking the other CPU's
   1016 	 * sched state, which we currently do not do.
   1017 	 */
   1018 	ci = (l->l_cpu != NULL) ? l->l_cpu : curcpu();
   1019 	if (pri < ci->ci_schedstate.spc_curpriority)
   1020 		need_resched(ci);
   1021 }
   1022 
   1023 /*
   1024  * Change process state to be runnable,
   1025  * placing it on the run queue if it is in memory,
   1026  * and awakening the swapper if it isn't in memory.
   1027  */
   1028 void
   1029 setrunnable(struct lwp *l)
   1030 {
   1031 	struct proc *p = l->l_proc;
   1032 
   1033 	SCHED_ASSERT_LOCKED();
   1034 
   1035 	switch (l->l_stat) {
   1036 	case 0:
   1037 	case LSRUN:
   1038 	case LSONPROC:
   1039 	case LSZOMB:
   1040 	case LSDEAD:
   1041 	default:
   1042 		panic("setrunnable: lwp %p state was %d", l, l->l_stat);
   1043 	case LSSTOP:
   1044 		/*
   1045 		 * If we're being traced (possibly because someone attached us
   1046 		 * while we were stopped), check for a signal from the debugger.
   1047 		 */
   1048 		if ((p->p_flag & P_TRACED) != 0 && p->p_xstat != 0) {
   1049 			sigaddset(&p->p_sigctx.ps_siglist, p->p_xstat);
   1050 			CHECKSIGS(p);
   1051 		}
   1052 	case LSSLEEP:
   1053 		unsleep(l);		/* e.g. when sending signals */
   1054 		break;
   1055 
   1056 	case LSIDL:
   1057 		break;
   1058 	case LSSUSPENDED:
   1059 		break;
   1060 	}
   1061 
   1062 	if (l->l_proc->p_sa)
   1063 		sa_awaken(l);
   1064 
   1065 	l->l_stat = LSRUN;
   1066 	p->p_nrlwps++;
   1067 
   1068 	if (l->l_flag & L_INMEM)
   1069 		setrunqueue(l);
   1070 
   1071 	if (l->l_slptime > 1)
   1072 		updatepri(l);
   1073 	l->l_slptime = 0;
   1074 	if ((l->l_flag & L_INMEM) == 0)
   1075 		sched_wakeup((caddr_t)&proc0);
   1076 	else
   1077 		resched_proc(l, l->l_priority);
   1078 }
   1079 
   1080 /*
   1081  * Compute the priority of a process when running in user mode.
   1082  * Arrange to reschedule if the resulting priority is better
   1083  * than that of the current process.
   1084  */
   1085 void
   1086 resetpriority(struct lwp *l)
   1087 {
   1088 	unsigned int newpriority;
   1089 	struct proc *p = l->l_proc;
   1090 
   1091 	SCHED_ASSERT_LOCKED();
   1092 
   1093 	newpriority = PUSER + p->p_estcpu +
   1094 			NICE_WEIGHT * (p->p_nice - NZERO);
   1095 	newpriority = min(newpriority, MAXPRI);
   1096 	l->l_usrpri = newpriority;
   1097 	resched_proc(l, l->l_usrpri);
   1098 }
   1099 
   1100 /*
   1101  * Recompute priority for all LWPs in a process.
   1102  */
   1103 void
   1104 resetprocpriority(struct proc *p)
   1105 {
   1106 	struct lwp *l;
   1107 
   1108 	LIST_FOREACH(l, &p->p_lwps, l_sibling)
   1109 	    resetpriority(l);
   1110 }
   1111 
   1112 /*
   1113  * We adjust the priority of the current process.  The priority of a process
   1114  * gets worse as it accumulates CPU time.  The CPU usage estimator (p_estcpu)
   1115  * is increased here.  The formula for computing priorities (in kern_synch.c)
   1116  * will compute a different value each time p_estcpu increases. This can
   1117  * cause a switch, but unless the priority crosses a PPQ boundary the actual
   1118  * queue will not change.  The CPU usage estimator ramps up quite quickly
   1119  * when the process is running (linearly), and decays away exponentially, at
   1120  * a rate which is proportionally slower when the system is busy.  The basic
   1121  * principle is that the system will 90% forget that the process used a lot
   1122  * of CPU time in 5 * loadav seconds.  This causes the system to favor
   1123  * processes which haven't run much recently, and to round-robin among other
   1124  * processes.
   1125  */
   1126 
   1127 void
   1128 schedclock(struct lwp *l)
   1129 {
   1130 	struct proc *p = l->l_proc;
   1131 	int s;
   1132 
   1133 	p->p_estcpu = ESTCPULIM(p->p_estcpu + 1);
   1134 	SCHED_LOCK(s);
   1135 	resetpriority(l);
   1136 	SCHED_UNLOCK(s);
   1137 
   1138 	if (l->l_priority >= PUSER)
   1139 		l->l_priority = l->l_usrpri;
   1140 }
   1141 
   1142 void
   1143 suspendsched()
   1144 {
   1145 	struct lwp *l;
   1146 	int s;
   1147 
   1148 	/*
   1149 	 * Convert all non-P_SYSTEM LSSLEEP or LSRUN processes to
   1150 	 * LSSUSPENDED.
   1151 	 */
   1152 	proclist_lock_read();
   1153 	SCHED_LOCK(s);
   1154 	LIST_FOREACH(l, &alllwp, l_list) {
   1155 		if ((l->l_proc->p_flag & P_SYSTEM) != 0)
   1156 			continue;
   1157 
   1158 		switch (l->l_stat) {
   1159 		case LSRUN:
   1160 			l->l_proc->p_nrlwps--;
   1161 			if ((l->l_flag & L_INMEM) != 0)
   1162 				remrunqueue(l);
   1163 			/* FALLTHROUGH */
   1164 		case LSSLEEP:
   1165 			l->l_stat = LSSUSPENDED;
   1166 			break;
   1167 		case LSONPROC:
   1168 			/*
   1169 			 * XXX SMP: we need to deal with processes on
   1170 			 * others CPU !
   1171 			 */
   1172 			break;
   1173 		default:
   1174 			break;
   1175 		}
   1176 	}
   1177 	SCHED_UNLOCK(s);
   1178 	proclist_unlock_read();
   1179 }
   1180 
   1181 /*
   1182  * Low-level routines to access the run queue.  Optimised assembler
   1183  * routines can override these.
   1184  */
   1185 
   1186 #ifndef __HAVE_MD_RUNQUEUE
   1187 
   1188 /*
   1189  * On some architectures, it's faster to use a MSB ordering for the priorites
   1190  * than the traditional LSB ordering.
   1191  */
   1192 #ifdef __HAVE_BIGENDIAN_BITOPS
   1193 #define	RQMASK(n) (0x80000000 >> (n))
   1194 #else
   1195 #define	RQMASK(n) (0x00000001 << (n))
   1196 #endif
   1197 
   1198 /*
   1199  * The primitives that manipulate the run queues.  whichqs tells which
   1200  * of the 32 queues qs have processes in them.  Setrunqueue puts processes
   1201  * into queues, remrunqueue removes them from queues.  The running process is
   1202  * on no queue, other processes are on a queue related to p->p_priority,
   1203  * divided by 4 actually to shrink the 0-127 range of priorities into the 32
   1204  * available queues.
   1205  */
   1206 
   1207 #ifdef RQDEBUG
   1208 static void
   1209 checkrunqueue(int whichq, struct lwp *l)
   1210 {
   1211 	const struct prochd * const rq = &sched_qs[whichq];
   1212 	struct lwp *l2;
   1213 	int found = 0;
   1214 	int die = 0;
   1215 	int empty = 1;
   1216 	for (l2 = rq->ph_link; l2 != (void*) rq; l2 = l2->l_forw) {
   1217 		if (l2->l_stat != LSRUN) {
   1218 			printf("checkrunqueue[%d]: lwp %p state (%d) "
   1219 			    " != LSRUN\n", whichq, l2, l2->l_stat);
   1220 		}
   1221 		if (l2->l_back->l_forw != l2) {
   1222 			printf("checkrunqueue[%d]: lwp %p back-qptr (%p) "
   1223 			    "corrupt %p\n", whichq, l2, l2->l_back,
   1224 			    l2->l_back->l_forw);
   1225 			die = 1;
   1226 		}
   1227 		if (l2->l_forw->l_back != l2) {
   1228 			printf("checkrunqueue[%d]: lwp %p forw-qptr (%p) "
   1229 			    "corrupt %p\n", whichq, l2, l2->l_forw,
   1230 			    l2->l_forw->l_back);
   1231 			die = 1;
   1232 		}
   1233 		if (l2 == l)
   1234 			found = 1;
   1235 		empty = 0;
   1236 	}
   1237 	if (empty && (sched_whichqs & RQMASK(whichq)) != 0) {
   1238 		printf("checkrunqueue[%d]: bit set for empty run-queue %p\n",
   1239 		    whichq, rq);
   1240 		die = 1;
   1241 	} else if (!empty && (sched_whichqs & RQMASK(whichq)) == 0) {
   1242 		printf("checkrunqueue[%d]: bit clear for non-empty "
   1243 		    "run-queue %p\n", whichq, rq);
   1244 		die = 1;
   1245 	}
   1246 	if (l != NULL && (sched_whichqs & RQMASK(whichq)) == 0) {
   1247 		printf("checkrunqueue[%d]: bit clear for active lwp %p\n",
   1248 		    whichq, l);
   1249 		die = 1;
   1250 	}
   1251 	if (l != NULL && empty) {
   1252 		printf("checkrunqueue[%d]: empty run-queue %p with "
   1253 		    "active lwp %p\n", whichq, rq, l);
   1254 		die = 1;
   1255 	}
   1256 	if (l != NULL && !found) {
   1257 		printf("checkrunqueue[%d]: lwp %p not in runqueue %p!",
   1258 		    whichq, l, rq);
   1259 		die = 1;
   1260 	}
   1261 	if (die)
   1262 		panic("checkrunqueue: inconsistency found");
   1263 }
   1264 #endif /* RQDEBUG */
   1265 
   1266 void
   1267 setrunqueue(struct lwp *l)
   1268 {
   1269 	struct prochd *rq;
   1270 	struct lwp *prev;
   1271 	const int whichq = l->l_priority / 4;
   1272 
   1273 #ifdef RQDEBUG
   1274 	checkrunqueue(whichq, NULL);
   1275 #endif
   1276 #ifdef DIAGNOSTIC
   1277 	if (l->l_back != NULL || l->l_wchan != NULL || l->l_stat != LSRUN)
   1278 		panic("setrunqueue");
   1279 #endif
   1280 	sched_whichqs |= RQMASK(whichq);
   1281 	rq = &sched_qs[whichq];
   1282 	prev = rq->ph_rlink;
   1283 	l->l_forw = (struct lwp *)rq;
   1284 	rq->ph_rlink = l;
   1285 	prev->l_forw = l;
   1286 	l->l_back = prev;
   1287 #ifdef RQDEBUG
   1288 	checkrunqueue(whichq, l);
   1289 #endif
   1290 }
   1291 
   1292 void
   1293 remrunqueue(struct lwp *l)
   1294 {
   1295 	struct lwp *prev, *next;
   1296 	const int whichq = l->l_priority / 4;
   1297 #ifdef RQDEBUG
   1298 	checkrunqueue(whichq, l);
   1299 #endif
   1300 #ifdef DIAGNOSTIC
   1301 	if (((sched_whichqs & RQMASK(whichq)) == 0))
   1302 		panic("remrunqueue: bit %d not set", whichq);
   1303 #endif
   1304 	prev = l->l_back;
   1305 	l->l_back = NULL;
   1306 	next = l->l_forw;
   1307 	prev->l_forw = next;
   1308 	next->l_back = prev;
   1309 	if (prev == next)
   1310 		sched_whichqs &= ~RQMASK(whichq);
   1311 #ifdef RQDEBUG
   1312 	checkrunqueue(whichq, NULL);
   1313 #endif
   1314 }
   1315 
   1316 #undef RQMASK
   1317 #endif /* !defined(__HAVE_MD_RUNQUEUE) */
   1318