xref: /freebsd/sys/kern/sched_ule.c (revision 3d238d9e981227b3bf739a51281e5d200bff3f8c)
1 /*-
2  * Copyright (c) 2002-2003, Jeffrey Roberson <jeff@freebsd.org>
3  * All rights reserved.
4  *
5  * Redistribution and use in source and binary forms, with or without
6  * modification, are permitted provided that the following conditions
7  * are met:
8  * 1. Redistributions of source code must retain the above copyright
9  *    notice unmodified, this list of conditions, and the following
10  *    disclaimer.
11  * 2. Redistributions in binary form must reproduce the above copyright
12  *    notice, this list of conditions and the following disclaimer in the
13  *    documentation and/or other materials provided with the distribution.
14  *
15  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
16  * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
17  * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
18  * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
19  * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
20  * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
21  * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
22  * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
23  * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
24  * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
25  */
26 
27 #include <sys/cdefs.h>
28 __FBSDID("$FreeBSD$");
29 
30 #include <sys/param.h>
31 #include <sys/systm.h>
32 #include <sys/kernel.h>
33 #include <sys/ktr.h>
34 #include <sys/lock.h>
35 #include <sys/mutex.h>
36 #include <sys/proc.h>
37 #include <sys/resource.h>
38 #include <sys/sched.h>
39 #include <sys/smp.h>
40 #include <sys/sx.h>
41 #include <sys/sysctl.h>
42 #include <sys/sysproto.h>
43 #include <sys/vmmeter.h>
44 #ifdef DDB
45 #include <ddb/ddb.h>
46 #endif
47 #ifdef KTRACE
48 #include <sys/uio.h>
49 #include <sys/ktrace.h>
50 #endif
51 
52 #include <machine/cpu.h>
53 
54 #define KTR_ULE         KTR_NFS
55 
56 /* decay 95% of `p_pctcpu' in 60 seconds; see CCPU_SHIFT before changing */
57 /* XXX This is bogus compatability crap for ps */
58 static fixpt_t  ccpu = 0.95122942450071400909 * FSCALE; /* exp(-1/20) */
59 SYSCTL_INT(_kern, OID_AUTO, ccpu, CTLFLAG_RD, &ccpu, 0, "");
60 
61 static void sched_setup(void *dummy);
62 SYSINIT(sched_setup, SI_SUB_RUN_QUEUE, SI_ORDER_FIRST, sched_setup, NULL)
63 
64 static SYSCTL_NODE(_kern, OID_AUTO, sched, CTLFLAG_RW, 0, "SCHED");
65 
66 static int sched_strict;
67 SYSCTL_INT(_kern_sched, OID_AUTO, strict, CTLFLAG_RD, &sched_strict, 0, "");
68 
69 static int slice_min = 1;
70 SYSCTL_INT(_kern_sched, OID_AUTO, slice_min, CTLFLAG_RW, &slice_min, 0, "");
71 
72 static int slice_max = 10;
73 SYSCTL_INT(_kern_sched, OID_AUTO, slice_max, CTLFLAG_RW, &slice_max, 0, "");
74 
75 int realstathz;
76 int tickincr = 1;
77 
78 #ifdef SMP
79 /* Callout to handle load balancing SMP systems. */
80 static struct callout kseq_lb_callout;
81 #endif
82 
83 /*
84  * These datastructures are allocated within their parent datastructure but
85  * are scheduler specific.
86  */
87 
88 struct ke_sched {
89 	int		ske_slice;
90 	struct runq	*ske_runq;
91 	/* The following variables are only used for pctcpu calculation */
92 	int		ske_ltick;	/* Last tick that we were running on */
93 	int		ske_ftick;	/* First tick that we were running on */
94 	int		ske_ticks;	/* Tick count */
95 	/* CPU that we have affinity for. */
96 	u_char		ske_cpu;
97 };
98 #define	ke_slice	ke_sched->ske_slice
99 #define	ke_runq		ke_sched->ske_runq
100 #define	ke_ltick	ke_sched->ske_ltick
101 #define	ke_ftick	ke_sched->ske_ftick
102 #define	ke_ticks	ke_sched->ske_ticks
103 #define	ke_cpu		ke_sched->ske_cpu
104 
105 struct kg_sched {
106 	int	skg_slptime;		/* Number of ticks we vol. slept */
107 	int	skg_runtime;		/* Number of ticks we were running */
108 };
109 #define	kg_slptime	kg_sched->skg_slptime
110 #define	kg_runtime	kg_sched->skg_runtime
111 
112 struct td_sched {
113 	int	std_slptime;
114 };
115 #define	td_slptime	td_sched->std_slptime
116 
117 struct td_sched td_sched;
118 struct ke_sched ke_sched;
119 struct kg_sched kg_sched;
120 
121 struct ke_sched *kse0_sched = &ke_sched;
122 struct kg_sched *ksegrp0_sched = &kg_sched;
123 struct p_sched *proc0_sched = NULL;
124 struct td_sched *thread0_sched = &td_sched;
125 
126 /*
127  * The priority is primarily determined by the interactivity score.  Thus, we
128  * give lower(better) priorities to kse groups that use less CPU.  The nice
129  * value is then directly added to this to allow nice to have some effect
130  * on latency.
131  *
132  * PRI_RANGE:	Total priority range for timeshare threads.
133  * PRI_NRESV:	Number of nice values.
134  * PRI_BASE:	The start of the dynamic range.
135  */
136 #define	SCHED_PRI_RANGE		(PRI_MAX_TIMESHARE - PRI_MIN_TIMESHARE + 1)
137 #define	SCHED_PRI_NRESV		PRIO_TOTAL
138 #define	SCHED_PRI_NHALF		(PRIO_TOTAL / 2)
139 #define	SCHED_PRI_NTHRESH	(SCHED_PRI_NHALF - 1)
140 #define	SCHED_PRI_BASE		(PRI_MIN_TIMESHARE)
141 #define	SCHED_PRI_INTERACT(score)					\
142     ((score) * SCHED_PRI_RANGE / SCHED_INTERACT_MAX)
143 
144 /*
145  * These determine the interactivity of a process.
146  *
147  * SLP_RUN_MAX:	Maximum amount of sleep time + run time we'll accumulate
148  *		before throttling back.
149  * SLP_RUN_THROTTLE:	Divisor for reducing slp/run time at fork time.
150  * INTERACT_MAX:	Maximum interactivity value.  Smaller is better.
151  * INTERACT_THRESH:	Threshhold for placement on the current runq.
152  */
153 #define	SCHED_SLP_RUN_MAX	((hz * 2) << 10)
154 #define	SCHED_SLP_RUN_THROTTLE	(100)
155 #define	SCHED_INTERACT_MAX	(100)
156 #define	SCHED_INTERACT_HALF	(SCHED_INTERACT_MAX / 2)
157 #define	SCHED_INTERACT_THRESH	(20)
158 
159 /*
160  * These parameters and macros determine the size of the time slice that is
161  * granted to each thread.
162  *
163  * SLICE_MIN:	Minimum time slice granted, in units of ticks.
164  * SLICE_MAX:	Maximum time slice granted.
165  * SLICE_RANGE:	Range of available time slices scaled by hz.
166  * SLICE_SCALE:	The number slices granted per val in the range of [0, max].
167  * SLICE_NICE:  Determine the amount of slice granted to a scaled nice.
168  */
169 #define	SCHED_SLICE_MIN			(slice_min)
170 #define	SCHED_SLICE_MAX			(slice_max)
171 #define	SCHED_SLICE_RANGE		(SCHED_SLICE_MAX - SCHED_SLICE_MIN + 1)
172 #define	SCHED_SLICE_SCALE(val, max)	(((val) * SCHED_SLICE_RANGE) / (max))
173 #define	SCHED_SLICE_NICE(nice)						\
174     (SCHED_SLICE_MAX - SCHED_SLICE_SCALE((nice), SCHED_PRI_NTHRESH))
175 
176 /*
177  * This macro determines whether or not the kse belongs on the current or
178  * next run queue.
179  *
180  * XXX nice value should effect how interactive a kg is.
181  */
182 #define	SCHED_INTERACTIVE(kg)						\
183     (sched_interact_score(kg) < SCHED_INTERACT_THRESH)
184 #define	SCHED_CURR(kg, ke)						\
185     (ke->ke_thread->td_priority < PRI_MIN_TIMESHARE || SCHED_INTERACTIVE(kg))
186 
187 /*
188  * Cpu percentage computation macros and defines.
189  *
190  * SCHED_CPU_TIME:	Number of seconds to average the cpu usage across.
191  * SCHED_CPU_TICKS:	Number of hz ticks to average the cpu usage across.
192  */
193 
194 #define	SCHED_CPU_TIME	10
195 #define	SCHED_CPU_TICKS	(hz * SCHED_CPU_TIME)
196 
197 /*
198  * kseq - per processor runqs and statistics.
199  */
200 
201 #define	KSEQ_NCLASS	(PRI_IDLE + 1)	/* Number of run classes. */
202 
203 struct kseq {
204 	struct runq	ksq_idle;		/* Queue of IDLE threads. */
205 	struct runq	ksq_timeshare[2];	/* Run queues for !IDLE. */
206 	struct runq	*ksq_next;		/* Next timeshare queue. */
207 	struct runq	*ksq_curr;		/* Current queue. */
208 	int		ksq_loads[KSEQ_NCLASS];	/* Load for each class */
209 	int		ksq_load;		/* Aggregate load. */
210 	short		ksq_nice[PRIO_TOTAL + 1]; /* KSEs in each nice bin. */
211 	short		ksq_nicemin;		/* Least nice. */
212 #ifdef SMP
213 	unsigned int	ksq_rslices;	/* Slices on run queue */
214 #endif
215 };
216 
217 /*
218  * One kse queue per processor.
219  */
220 #ifdef SMP
221 struct kseq	kseq_cpu[MAXCPU];
222 #define	KSEQ_SELF()	(&kseq_cpu[PCPU_GET(cpuid)])
223 #define	KSEQ_CPU(x)	(&kseq_cpu[(x)])
224 #else
225 struct kseq	kseq_cpu;
226 #define	KSEQ_SELF()	(&kseq_cpu)
227 #define	KSEQ_CPU(x)	(&kseq_cpu)
228 #endif
229 
230 static void sched_slice(struct kse *ke);
231 static void sched_priority(struct ksegrp *kg);
232 static int sched_interact_score(struct ksegrp *kg);
233 static void sched_interact_update(struct ksegrp *kg);
234 void sched_pctcpu_update(struct kse *ke);
235 int sched_pickcpu(void);
236 
237 /* Operations on per processor queues */
238 static struct kse * kseq_choose(struct kseq *kseq);
239 static void kseq_setup(struct kseq *kseq);
240 static void kseq_add(struct kseq *kseq, struct kse *ke);
241 static void kseq_rem(struct kseq *kseq, struct kse *ke);
242 static void kseq_nice_add(struct kseq *kseq, int nice);
243 static void kseq_nice_rem(struct kseq *kseq, int nice);
244 void kseq_print(int cpu);
245 #ifdef SMP
246 struct kseq * kseq_load_highest(void);
247 void kseq_balance(void *arg);
248 void kseq_move(struct kseq *from, int cpu);
249 #endif
250 
251 void
252 kseq_print(int cpu)
253 {
254 	struct kseq *kseq;
255 	int i;
256 
257 	kseq = KSEQ_CPU(cpu);
258 
259 	printf("kseq:\n");
260 	printf("\tload:           %d\n", kseq->ksq_load);
261 	printf("\tload ITHD:      %d\n", kseq->ksq_loads[PRI_ITHD]);
262 	printf("\tload REALTIME:  %d\n", kseq->ksq_loads[PRI_REALTIME]);
263 	printf("\tload TIMESHARE: %d\n", kseq->ksq_loads[PRI_TIMESHARE]);
264 	printf("\tload IDLE:      %d\n", kseq->ksq_loads[PRI_IDLE]);
265 	printf("\tnicemin:\t%d\n", kseq->ksq_nicemin);
266 	printf("\tnice counts:\n");
267 	for (i = 0; i < PRIO_TOTAL + 1; i++)
268 		if (kseq->ksq_nice[i])
269 			printf("\t\t%d = %d\n",
270 			    i - SCHED_PRI_NHALF, kseq->ksq_nice[i]);
271 }
272 
273 static void
274 kseq_add(struct kseq *kseq, struct kse *ke)
275 {
276 	mtx_assert(&sched_lock, MA_OWNED);
277 	kseq->ksq_loads[PRI_BASE(ke->ke_ksegrp->kg_pri_class)]++;
278 	kseq->ksq_load++;
279 	if (ke->ke_ksegrp->kg_pri_class == PRI_TIMESHARE)
280 	CTR6(KTR_ULE, "Add kse %p to %p (slice: %d, pri: %d, nice: %d(%d))",
281 	    ke, ke->ke_runq, ke->ke_slice, ke->ke_thread->td_priority,
282 	    ke->ke_ksegrp->kg_nice, kseq->ksq_nicemin);
283 	if (ke->ke_ksegrp->kg_pri_class == PRI_TIMESHARE)
284 		kseq_nice_add(kseq, ke->ke_ksegrp->kg_nice);
285 #ifdef SMP
286 	kseq->ksq_rslices += ke->ke_slice;
287 #endif
288 }
289 
290 static void
291 kseq_rem(struct kseq *kseq, struct kse *ke)
292 {
293 	mtx_assert(&sched_lock, MA_OWNED);
294 	kseq->ksq_loads[PRI_BASE(ke->ke_ksegrp->kg_pri_class)]--;
295 	kseq->ksq_load--;
296 	ke->ke_runq = NULL;
297 	if (ke->ke_ksegrp->kg_pri_class == PRI_TIMESHARE)
298 		kseq_nice_rem(kseq, ke->ke_ksegrp->kg_nice);
299 #ifdef SMP
300 	kseq->ksq_rslices -= ke->ke_slice;
301 #endif
302 }
303 
304 static void
305 kseq_nice_add(struct kseq *kseq, int nice)
306 {
307 	mtx_assert(&sched_lock, MA_OWNED);
308 	/* Normalize to zero. */
309 	kseq->ksq_nice[nice + SCHED_PRI_NHALF]++;
310 	if (nice < kseq->ksq_nicemin || kseq->ksq_loads[PRI_TIMESHARE] == 1)
311 		kseq->ksq_nicemin = nice;
312 }
313 
314 static void
315 kseq_nice_rem(struct kseq *kseq, int nice)
316 {
317 	int n;
318 
319 	mtx_assert(&sched_lock, MA_OWNED);
320 	/* Normalize to zero. */
321 	n = nice + SCHED_PRI_NHALF;
322 	kseq->ksq_nice[n]--;
323 	KASSERT(kseq->ksq_nice[n] >= 0, ("Negative nice count."));
324 
325 	/*
326 	 * If this wasn't the smallest nice value or there are more in
327 	 * this bucket we can just return.  Otherwise we have to recalculate
328 	 * the smallest nice.
329 	 */
330 	if (nice != kseq->ksq_nicemin ||
331 	    kseq->ksq_nice[n] != 0 ||
332 	    kseq->ksq_loads[PRI_TIMESHARE] == 0)
333 		return;
334 
335 	for (; n < SCHED_PRI_NRESV + 1; n++)
336 		if (kseq->ksq_nice[n]) {
337 			kseq->ksq_nicemin = n - SCHED_PRI_NHALF;
338 			return;
339 		}
340 }
341 
342 #ifdef SMP
343 /*
344  * kseq_balance is a simple CPU load balancing algorithm.  It operates by
345  * finding the least loaded and most loaded cpu and equalizing their load
346  * by migrating some processes.
347  *
348  * Dealing only with two CPUs at a time has two advantages.  Firstly, most
349  * installations will only have 2 cpus.  Secondly, load balancing too much at
350  * once can have an unpleasant effect on the system.  The scheduler rarely has
351  * enough information to make perfect decisions.  So this algorithm chooses
352  * algorithm simplicity and more gradual effects on load in larger systems.
353  *
354  * It could be improved by considering the priorities and slices assigned to
355  * each task prior to balancing them.  There are many pathological cases with
356  * any approach and so the semi random algorithm below may work as well as any.
357  *
358  */
359 void
360 kseq_balance(void *arg)
361 {
362 	struct kseq *kseq;
363 	int high_load;
364 	int low_load;
365 	int high_cpu;
366 	int low_cpu;
367 	int move;
368 	int diff;
369 	int i;
370 
371 	high_cpu = 0;
372 	low_cpu = 0;
373 	high_load = 0;
374 	low_load = -1;
375 
376 	mtx_lock_spin(&sched_lock);
377 	if (smp_started == 0)
378 		goto out;
379 
380 	for (i = 0; i < mp_maxid; i++) {
381 		if (CPU_ABSENT(i) || (i & stopped_cpus) != 0)
382 			continue;
383 		kseq = KSEQ_CPU(i);
384 		if (kseq->ksq_load > high_load) {
385 			high_load = kseq->ksq_load;
386 			high_cpu = i;
387 		}
388 		if (low_load == -1 || kseq->ksq_load < low_load) {
389 			low_load = kseq->ksq_load;
390 			low_cpu = i;
391 		}
392 	}
393 
394 	/*
395 	 * Nothing to do.
396 	 */
397 	if (high_load < 2 || low_load == high_load)
398 		goto out;
399 
400 	diff = high_load - low_load;
401 	move = diff / 2;
402 	if (diff & 0x1)
403 		move++;
404 
405 	for (i = 0; i < move; i++)
406 		kseq_move(KSEQ_CPU(high_cpu), low_cpu);
407 
408 out:
409 	mtx_unlock_spin(&sched_lock);
410 	callout_reset(&kseq_lb_callout, hz, kseq_balance, NULL);
411 
412 	return;
413 }
414 
415 struct kseq *
416 kseq_load_highest(void)
417 {
418 	struct kseq *kseq;
419 	int load;
420 	int cpu;
421 	int i;
422 
423 	mtx_assert(&sched_lock, MA_OWNED);
424 	cpu = 0;
425 	load = 0;
426 
427 	for (i = 0; i < mp_maxid; i++) {
428 		if (CPU_ABSENT(i) || (i & stopped_cpus) != 0)
429 			continue;
430 		kseq = KSEQ_CPU(i);
431 		if (kseq->ksq_load > load) {
432 			load = kseq->ksq_load;
433 			cpu = i;
434 		}
435 	}
436 	if (load > 1)
437 		return (KSEQ_CPU(cpu));
438 
439 	return (NULL);
440 }
441 
442 void
443 kseq_move(struct kseq *from, int cpu)
444 {
445 	struct kse *ke;
446 
447 	ke = kseq_choose(from);
448 	runq_remove(ke->ke_runq, ke);
449 	ke->ke_state = KES_THREAD;
450 	kseq_rem(from, ke);
451 
452 	ke->ke_cpu = cpu;
453 	sched_add(ke);
454 }
455 #endif
456 
457 struct kse *
458 kseq_choose(struct kseq *kseq)
459 {
460 	struct kse *ke;
461 	struct runq *swap;
462 
463 	mtx_assert(&sched_lock, MA_OWNED);
464 	swap = NULL;
465 
466 	for (;;) {
467 		ke = runq_choose(kseq->ksq_curr);
468 		if (ke == NULL) {
469 			/*
470 			 * We already swaped once and didn't get anywhere.
471 			 */
472 			if (swap)
473 				break;
474 			swap = kseq->ksq_curr;
475 			kseq->ksq_curr = kseq->ksq_next;
476 			kseq->ksq_next = swap;
477 			continue;
478 		}
479 		/*
480 		 * If we encounter a slice of 0 the kse is in a
481 		 * TIMESHARE kse group and its nice was too far out
482 		 * of the range that receives slices.
483 		 */
484 		if (ke->ke_slice == 0) {
485 			runq_remove(ke->ke_runq, ke);
486 			sched_slice(ke);
487 			ke->ke_runq = kseq->ksq_next;
488 			runq_add(ke->ke_runq, ke);
489 			continue;
490 		}
491 		return (ke);
492 	}
493 
494 	return (runq_choose(&kseq->ksq_idle));
495 }
496 
497 static void
498 kseq_setup(struct kseq *kseq)
499 {
500 	runq_init(&kseq->ksq_timeshare[0]);
501 	runq_init(&kseq->ksq_timeshare[1]);
502 	runq_init(&kseq->ksq_idle);
503 
504 	kseq->ksq_curr = &kseq->ksq_timeshare[0];
505 	kseq->ksq_next = &kseq->ksq_timeshare[1];
506 
507 	kseq->ksq_loads[PRI_ITHD] = 0;
508 	kseq->ksq_loads[PRI_REALTIME] = 0;
509 	kseq->ksq_loads[PRI_TIMESHARE] = 0;
510 	kseq->ksq_loads[PRI_IDLE] = 0;
511 	kseq->ksq_load = 0;
512 #ifdef SMP
513 	kseq->ksq_rslices = 0;
514 #endif
515 }
516 
517 static void
518 sched_setup(void *dummy)
519 {
520 	int i;
521 
522 	slice_min = (hz/100);	/* 10ms */
523 	slice_max = (hz/7);	/* ~140ms */
524 
525 	mtx_lock_spin(&sched_lock);
526 	/* init kseqs */
527 	for (i = 0; i < MAXCPU; i++)
528 		kseq_setup(KSEQ_CPU(i));
529 
530 	kseq_add(KSEQ_SELF(), &kse0);
531 	mtx_unlock_spin(&sched_lock);
532 #ifdef SMP
533 	callout_init(&kseq_lb_callout, 1);
534 	kseq_balance(NULL);
535 #endif
536 }
537 
538 /*
539  * Scale the scheduling priority according to the "interactivity" of this
540  * process.
541  */
542 static void
543 sched_priority(struct ksegrp *kg)
544 {
545 	int pri;
546 
547 	if (kg->kg_pri_class != PRI_TIMESHARE)
548 		return;
549 
550 	pri = SCHED_PRI_INTERACT(sched_interact_score(kg));
551 	pri += SCHED_PRI_BASE;
552 	pri += kg->kg_nice;
553 
554 	if (pri > PRI_MAX_TIMESHARE)
555 		pri = PRI_MAX_TIMESHARE;
556 	else if (pri < PRI_MIN_TIMESHARE)
557 		pri = PRI_MIN_TIMESHARE;
558 
559 	kg->kg_user_pri = pri;
560 
561 	return;
562 }
563 
564 /*
565  * Calculate a time slice based on the properties of the kseg and the runq
566  * that we're on.  This is only for PRI_TIMESHARE ksegrps.
567  */
568 static void
569 sched_slice(struct kse *ke)
570 {
571 	struct kseq *kseq;
572 	struct ksegrp *kg;
573 
574 	kg = ke->ke_ksegrp;
575 	kseq = KSEQ_CPU(ke->ke_cpu);
576 
577 	/*
578 	 * Rationale:
579 	 * KSEs in interactive ksegs get the minimum slice so that we
580 	 * quickly notice if it abuses its advantage.
581 	 *
582 	 * KSEs in non-interactive ksegs are assigned a slice that is
583 	 * based on the ksegs nice value relative to the least nice kseg
584 	 * on the run queue for this cpu.
585 	 *
586 	 * If the KSE is less nice than all others it gets the maximum
587 	 * slice and other KSEs will adjust their slice relative to
588 	 * this when they first expire.
589 	 *
590 	 * There is 20 point window that starts relative to the least
591 	 * nice kse on the run queue.  Slice size is determined by
592 	 * the kse distance from the last nice ksegrp.
593 	 *
594 	 * If you are outside of the window you will get no slice and
595 	 * you will be reevaluated each time you are selected on the
596 	 * run queue.
597 	 *
598 	 */
599 
600 	if (!SCHED_INTERACTIVE(kg)) {
601 		int nice;
602 
603 		nice = kg->kg_nice + (0 - kseq->ksq_nicemin);
604 		if (kseq->ksq_loads[PRI_TIMESHARE] == 0 ||
605 		    kg->kg_nice < kseq->ksq_nicemin)
606 			ke->ke_slice = SCHED_SLICE_MAX;
607 		else if (nice <= SCHED_PRI_NTHRESH)
608 			ke->ke_slice = SCHED_SLICE_NICE(nice);
609 		else
610 			ke->ke_slice = 0;
611 	} else
612 		ke->ke_slice = SCHED_SLICE_MIN;
613 
614 	CTR6(KTR_ULE,
615 	    "Sliced %p(%d) (nice: %d, nicemin: %d, load: %d, interactive: %d)",
616 	    ke, ke->ke_slice, kg->kg_nice, kseq->ksq_nicemin,
617 	    kseq->ksq_loads[PRI_TIMESHARE], SCHED_INTERACTIVE(kg));
618 
619 	/*
620 	 * Check to see if we need to scale back the slp and run time
621 	 * in the kg.  This will cause us to forget old interactivity
622 	 * while maintaining the current ratio.
623 	 */
624 	sched_interact_update(kg);
625 
626 	return;
627 }
628 
629 static void
630 sched_interact_update(struct ksegrp *kg)
631 {
632 	/* XXX Fixme, use a linear algorithm and not a while loop. */
633 	while ((kg->kg_runtime + kg->kg_slptime) >  SCHED_SLP_RUN_MAX) {
634 		kg->kg_runtime = (kg->kg_runtime / 5) * 4;
635 		kg->kg_slptime = (kg->kg_slptime / 5) * 4;
636 	}
637 }
638 
639 static int
640 sched_interact_score(struct ksegrp *kg)
641 {
642 	int div;
643 
644 	if (kg->kg_runtime > kg->kg_slptime) {
645 		div = max(1, kg->kg_runtime / SCHED_INTERACT_HALF);
646 		return (SCHED_INTERACT_HALF +
647 		    (SCHED_INTERACT_HALF - (kg->kg_slptime / div)));
648 	} if (kg->kg_slptime > kg->kg_runtime) {
649 		div = max(1, kg->kg_slptime / SCHED_INTERACT_HALF);
650 		return (kg->kg_runtime / div);
651 	}
652 
653 	/*
654 	 * This can happen if slptime and runtime are 0.
655 	 */
656 	return (0);
657 
658 }
659 
660 /*
661  * This is only somewhat accurate since given many processes of the same
662  * priority they will switch when their slices run out, which will be
663  * at most SCHED_SLICE_MAX.
664  */
665 int
666 sched_rr_interval(void)
667 {
668 	return (SCHED_SLICE_MAX);
669 }
670 
671 void
672 sched_pctcpu_update(struct kse *ke)
673 {
674 	/*
675 	 * Adjust counters and watermark for pctcpu calc.
676 	 */
677 
678 	/*
679 	 * Shift the tick count out so that the divide doesn't round away
680 	 * our results.
681 	 */
682 	ke->ke_ticks <<= 10;
683 	ke->ke_ticks = (ke->ke_ticks / (ke->ke_ltick - ke->ke_ftick)) *
684 		    SCHED_CPU_TICKS;
685 	ke->ke_ticks >>= 10;
686 	ke->ke_ltick = ticks;
687 	ke->ke_ftick = ke->ke_ltick - SCHED_CPU_TICKS;
688 }
689 
690 #ifdef SMP
691 /* XXX Should be changed to kseq_load_lowest() */
692 int
693 sched_pickcpu(void)
694 {
695 	struct kseq *kseq;
696 	int load;
697 	int cpu;
698 	int i;
699 
700 	mtx_assert(&sched_lock, MA_OWNED);
701 	if (!smp_started)
702 		return (0);
703 
704 	load = 0;
705 	cpu = 0;
706 
707 	for (i = 0; i < mp_maxid; i++) {
708 		if (CPU_ABSENT(i) || (i & stopped_cpus) != 0)
709 			continue;
710 		kseq = KSEQ_CPU(i);
711 		if (kseq->ksq_load < load) {
712 			cpu = i;
713 			load = kseq->ksq_load;
714 		}
715 	}
716 
717 	CTR1(KTR_RUNQ, "sched_pickcpu: %d", cpu);
718 	return (cpu);
719 }
720 #else
721 int
722 sched_pickcpu(void)
723 {
724 	return (0);
725 }
726 #endif
727 
728 void
729 sched_prio(struct thread *td, u_char prio)
730 {
731 	struct kse *ke;
732 	struct runq *rq;
733 
734 	mtx_assert(&sched_lock, MA_OWNED);
735 	ke = td->td_kse;
736 	td->td_priority = prio;
737 
738 	if (TD_ON_RUNQ(td)) {
739 		rq = ke->ke_runq;
740 
741 		runq_remove(rq, ke);
742 		runq_add(rq, ke);
743 	}
744 }
745 
746 void
747 sched_switchout(struct thread *td)
748 {
749 	struct kse *ke;
750 
751 	mtx_assert(&sched_lock, MA_OWNED);
752 
753 	ke = td->td_kse;
754 
755 	td->td_last_kse = ke;
756         td->td_lastcpu = td->td_oncpu;
757 	td->td_oncpu = NOCPU;
758         td->td_flags &= ~TDF_NEEDRESCHED;
759 
760 	if (TD_IS_RUNNING(td)) {
761 		/*
762 		 * This queue is always correct except for idle threads which
763 		 * have a higher priority due to priority propagation.
764 		 */
765 		if (ke->ke_ksegrp->kg_pri_class == PRI_IDLE &&
766 		    ke->ke_thread->td_priority > PRI_MIN_IDLE)
767 			ke->ke_runq = KSEQ_SELF()->ksq_curr;
768 		runq_add(ke->ke_runq, ke);
769 		/* setrunqueue(td); */
770 		return;
771 	}
772 	if (ke->ke_runq)
773 		kseq_rem(KSEQ_CPU(ke->ke_cpu), ke);
774 	/*
775 	 * We will not be on the run queue. So we must be
776 	 * sleeping or similar.
777 	 */
778 	if (td->td_proc->p_flag & P_SA)
779 		kse_reassign(ke);
780 }
781 
782 void
783 sched_switchin(struct thread *td)
784 {
785 	/* struct kse *ke = td->td_kse; */
786 	mtx_assert(&sched_lock, MA_OWNED);
787 
788 	td->td_oncpu = PCPU_GET(cpuid);
789 }
790 
791 void
792 sched_nice(struct ksegrp *kg, int nice)
793 {
794 	struct kse *ke;
795 	struct thread *td;
796 	struct kseq *kseq;
797 
798 	PROC_LOCK_ASSERT(kg->kg_proc, MA_OWNED);
799 	mtx_assert(&sched_lock, MA_OWNED);
800 	/*
801 	 * We need to adjust the nice counts for running KSEs.
802 	 */
803 	if (kg->kg_pri_class == PRI_TIMESHARE)
804 		FOREACH_KSE_IN_GROUP(kg, ke) {
805 			if (ke->ke_runq == NULL)
806 				continue;
807 			kseq = KSEQ_CPU(ke->ke_cpu);
808 			kseq_nice_rem(kseq, kg->kg_nice);
809 			kseq_nice_add(kseq, nice);
810 		}
811 	kg->kg_nice = nice;
812 	sched_priority(kg);
813 	FOREACH_THREAD_IN_GROUP(kg, td)
814 		td->td_flags |= TDF_NEEDRESCHED;
815 }
816 
817 void
818 sched_sleep(struct thread *td, u_char prio)
819 {
820 	mtx_assert(&sched_lock, MA_OWNED);
821 
822 	td->td_slptime = ticks;
823 	td->td_priority = prio;
824 
825 	CTR2(KTR_ULE, "sleep kse %p (tick: %d)",
826 	    td->td_kse, td->td_slptime);
827 }
828 
829 void
830 sched_wakeup(struct thread *td)
831 {
832 	mtx_assert(&sched_lock, MA_OWNED);
833 
834 	/*
835 	 * Let the kseg know how long we slept for.  This is because process
836 	 * interactivity behavior is modeled in the kseg.
837 	 */
838 	if (td->td_slptime) {
839 		struct ksegrp *kg;
840 		int hzticks;
841 
842 		kg = td->td_ksegrp;
843 		hzticks = ticks - td->td_slptime;
844 		kg->kg_slptime += hzticks << 10;
845 		sched_interact_update(kg);
846 		sched_priority(kg);
847 		if (td->td_kse)
848 			sched_slice(td->td_kse);
849 		CTR2(KTR_ULE, "wakeup kse %p (%d ticks)",
850 		    td->td_kse, hzticks);
851 		td->td_slptime = 0;
852 	}
853 	setrunqueue(td);
854         if (td->td_priority < curthread->td_priority)
855                 curthread->td_flags |= TDF_NEEDRESCHED;
856 }
857 
858 /*
859  * Penalize the parent for creating a new child and initialize the child's
860  * priority.
861  */
862 void
863 sched_fork(struct proc *p, struct proc *p1)
864 {
865 
866 	mtx_assert(&sched_lock, MA_OWNED);
867 
868 	sched_fork_ksegrp(FIRST_KSEGRP_IN_PROC(p), FIRST_KSEGRP_IN_PROC(p1));
869 	sched_fork_kse(FIRST_KSE_IN_PROC(p), FIRST_KSE_IN_PROC(p1));
870 	sched_fork_thread(FIRST_THREAD_IN_PROC(p), FIRST_THREAD_IN_PROC(p1));
871 }
872 
873 void
874 sched_fork_kse(struct kse *ke, struct kse *child)
875 {
876 
877 	child->ke_slice = 1;	/* Attempt to quickly learn interactivity. */
878 	child->ke_cpu = ke->ke_cpu; /* sched_pickcpu(); */
879 	child->ke_runq = NULL;
880 
881 	/*
882 	 * Claim that we've been running for one second for statistical
883 	 * purposes.
884 	 */
885 	child->ke_ticks = 0;
886 	child->ke_ltick = ticks;
887 	child->ke_ftick = ticks - hz;
888 }
889 
890 void
891 sched_fork_ksegrp(struct ksegrp *kg, struct ksegrp *child)
892 {
893 
894 	PROC_LOCK_ASSERT(child->kg_proc, MA_OWNED);
895 	/* XXX Need something better here */
896 
897 	child->kg_slptime = kg->kg_slptime / SCHED_SLP_RUN_THROTTLE;
898 	child->kg_runtime = kg->kg_runtime / SCHED_SLP_RUN_THROTTLE;
899 	kg->kg_runtime += tickincr << 10;
900 	sched_interact_update(kg);
901 
902 	child->kg_user_pri = kg->kg_user_pri;
903 	child->kg_nice = kg->kg_nice;
904 }
905 
906 void
907 sched_fork_thread(struct thread *td, struct thread *child)
908 {
909 }
910 
911 void
912 sched_class(struct ksegrp *kg, int class)
913 {
914 	struct kseq *kseq;
915 	struct kse *ke;
916 
917 	mtx_assert(&sched_lock, MA_OWNED);
918 	if (kg->kg_pri_class == class)
919 		return;
920 
921 	FOREACH_KSE_IN_GROUP(kg, ke) {
922 		if (ke->ke_state != KES_ONRUNQ &&
923 		    ke->ke_state != KES_THREAD)
924 			continue;
925 		kseq = KSEQ_CPU(ke->ke_cpu);
926 
927 		kseq->ksq_loads[PRI_BASE(kg->kg_pri_class)]--;
928 		kseq->ksq_loads[PRI_BASE(class)]++;
929 
930 		if (kg->kg_pri_class == PRI_TIMESHARE)
931 			kseq_nice_rem(kseq, kg->kg_nice);
932 		else if (class == PRI_TIMESHARE)
933 			kseq_nice_add(kseq, kg->kg_nice);
934 	}
935 
936 	kg->kg_pri_class = class;
937 }
938 
939 /*
940  * Return some of the child's priority and interactivity to the parent.
941  */
942 void
943 sched_exit(struct proc *p, struct proc *child)
944 {
945 	/* XXX Need something better here */
946 	mtx_assert(&sched_lock, MA_OWNED);
947 	sched_exit_kse(FIRST_KSE_IN_PROC(p), FIRST_KSE_IN_PROC(child));
948 	sched_exit_ksegrp(FIRST_KSEGRP_IN_PROC(p), FIRST_KSEGRP_IN_PROC(child));
949 }
950 
951 void
952 sched_exit_kse(struct kse *ke, struct kse *child)
953 {
954 	kseq_rem(KSEQ_CPU(child->ke_cpu), child);
955 }
956 
957 void
958 sched_exit_ksegrp(struct ksegrp *kg, struct ksegrp *child)
959 {
960 	/* kg->kg_slptime += child->kg_slptime; */
961 	kg->kg_runtime += child->kg_runtime;
962 	sched_interact_update(kg);
963 }
964 
965 void
966 sched_exit_thread(struct thread *td, struct thread *child)
967 {
968 }
969 
970 void
971 sched_clock(struct kse *ke)
972 {
973 	struct kseq *kseq;
974 	struct ksegrp *kg;
975 	struct thread *td;
976 #if 0
977 	struct kse *nke;
978 #endif
979 
980 	/*
981 	 * sched_setup() apparently happens prior to stathz being set.  We
982 	 * need to resolve the timers earlier in the boot so we can avoid
983 	 * calculating this here.
984 	 */
985 	if (realstathz == 0) {
986 		realstathz = stathz ? stathz : hz;
987 		tickincr = hz / realstathz;
988 		/*
989 		 * XXX This does not work for values of stathz that are much
990 		 * larger than hz.
991 		 */
992 		if (tickincr == 0)
993 			tickincr = 1;
994 	}
995 
996 	td = ke->ke_thread;
997 	kg = ke->ke_ksegrp;
998 
999 	mtx_assert(&sched_lock, MA_OWNED);
1000 	KASSERT((td != NULL), ("schedclock: null thread pointer"));
1001 
1002 	/* Adjust ticks for pctcpu */
1003 	ke->ke_ticks++;
1004 	ke->ke_ltick = ticks;
1005 
1006 	/* Go up to one second beyond our max and then trim back down */
1007 	if (ke->ke_ftick + SCHED_CPU_TICKS + hz < ke->ke_ltick)
1008 		sched_pctcpu_update(ke);
1009 
1010 	if (td->td_flags & TDF_IDLETD)
1011 		return;
1012 
1013 	CTR4(KTR_ULE, "Tick kse %p (slice: %d, slptime: %d, runtime: %d)",
1014 	    ke, ke->ke_slice, kg->kg_slptime >> 10, kg->kg_runtime >> 10);
1015 
1016 	/*
1017 	 * We only do slicing code for TIMESHARE ksegrps.
1018 	 */
1019 	if (kg->kg_pri_class != PRI_TIMESHARE)
1020 		return;
1021 	/*
1022 	 * Check for a higher priority task on the run queue.  This can happen
1023 	 * on SMP if another processor woke up a process on our runq.
1024 	 */
1025 	kseq = KSEQ_SELF();
1026 #if 0
1027 	if (kseq->ksq_load > 1 && (nke = kseq_choose(kseq)) != NULL) {
1028 		if (sched_strict &&
1029 		    nke->ke_thread->td_priority < td->td_priority)
1030 			td->td_flags |= TDF_NEEDRESCHED;
1031 		else if (nke->ke_thread->td_priority <
1032 		    td->td_priority SCHED_PRIO_SLOP)
1033 
1034 		if (nke->ke_thread->td_priority < td->td_priority)
1035 			td->td_flags |= TDF_NEEDRESCHED;
1036 	}
1037 #endif
1038 	/*
1039 	 * We used a tick charge it to the ksegrp so that we can compute our
1040 	 * interactivity.
1041 	 */
1042 	kg->kg_runtime += tickincr << 10;
1043 	sched_interact_update(kg);
1044 
1045 	/*
1046 	 * We used up one time slice.
1047 	 */
1048 	ke->ke_slice--;
1049 #ifdef SMP
1050 	kseq->ksq_rslices--;
1051 #endif
1052 
1053 	if (ke->ke_slice > 0)
1054 		return;
1055 	/*
1056 	 * We're out of time, recompute priorities and requeue.
1057 	 */
1058 	kseq_rem(kseq, ke);
1059 	sched_priority(kg);
1060 	sched_slice(ke);
1061 	if (SCHED_CURR(kg, ke))
1062 		ke->ke_runq = kseq->ksq_curr;
1063 	else
1064 		ke->ke_runq = kseq->ksq_next;
1065 	kseq_add(kseq, ke);
1066 	td->td_flags |= TDF_NEEDRESCHED;
1067 }
1068 
1069 int
1070 sched_runnable(void)
1071 {
1072 	struct kseq *kseq;
1073 	int load;
1074 
1075 	load = 1;
1076 
1077 	mtx_lock_spin(&sched_lock);
1078 	kseq = KSEQ_SELF();
1079 
1080 	if (kseq->ksq_load)
1081 		goto out;
1082 #ifdef SMP
1083 	/*
1084 	 * For SMP we may steal other processor's KSEs.  Just search until we
1085 	 * verify that at least on other cpu has a runnable task.
1086 	 */
1087 	if (smp_started) {
1088 		int i;
1089 
1090 		for (i = 0; i < mp_maxid; i++) {
1091 			if (CPU_ABSENT(i) || (i & stopped_cpus) != 0)
1092 				continue;
1093 			kseq = KSEQ_CPU(i);
1094 			if (kseq->ksq_load > 1)
1095 				goto out;
1096 		}
1097 	}
1098 #endif
1099 	load = 0;
1100 out:
1101 	mtx_unlock_spin(&sched_lock);
1102 	return (load);
1103 }
1104 
1105 void
1106 sched_userret(struct thread *td)
1107 {
1108 	struct ksegrp *kg;
1109 	struct kseq *kseq;
1110 	struct kse *ke;
1111 
1112 	kg = td->td_ksegrp;
1113 
1114 	if (td->td_priority != kg->kg_user_pri) {
1115 		mtx_lock_spin(&sched_lock);
1116 		td->td_priority = kg->kg_user_pri;
1117 		kseq = KSEQ_SELF();
1118 		if (td->td_ksegrp->kg_pri_class == PRI_TIMESHARE &&
1119 		    kseq->ksq_load > 1 &&
1120 		    (ke = kseq_choose(kseq)) != NULL &&
1121 		    ke->ke_thread->td_priority < td->td_priority)
1122 			curthread->td_flags |= TDF_NEEDRESCHED;
1123 		mtx_unlock_spin(&sched_lock);
1124 	}
1125 }
1126 
1127 struct kse *
1128 sched_choose(void)
1129 {
1130 	struct kseq *kseq;
1131 	struct kse *ke;
1132 
1133 	mtx_assert(&sched_lock, MA_OWNED);
1134 #ifdef SMP
1135 retry:
1136 #endif
1137 	kseq = KSEQ_SELF();
1138 	ke = kseq_choose(kseq);
1139 	if (ke) {
1140 		runq_remove(ke->ke_runq, ke);
1141 		ke->ke_state = KES_THREAD;
1142 
1143 		if (ke->ke_ksegrp->kg_pri_class == PRI_TIMESHARE) {
1144 			CTR4(KTR_ULE, "Run kse %p from %p (slice: %d, pri: %d)",
1145 			    ke, ke->ke_runq, ke->ke_slice,
1146 			    ke->ke_thread->td_priority);
1147 		}
1148 		return (ke);
1149 	}
1150 
1151 #ifdef SMP
1152 	if (smp_started) {
1153 		/*
1154 		 * Find the cpu with the highest load and steal one proc.
1155 		 */
1156 		if ((kseq = kseq_load_highest()) == NULL)
1157 			return (NULL);
1158 
1159 		/*
1160 		 * Remove this kse from this kseq and runq and then requeue
1161 		 * on the current processor.  Then we will dequeue it
1162 		 * normally above.
1163 		 */
1164 		kseq_move(kseq, PCPU_GET(cpuid));
1165 		goto retry;
1166 	}
1167 #endif
1168 
1169 	return (NULL);
1170 }
1171 
1172 void
1173 sched_add(struct kse *ke)
1174 {
1175 	struct kseq *kseq;
1176 	struct ksegrp *kg;
1177 
1178 	mtx_assert(&sched_lock, MA_OWNED);
1179 	KASSERT((ke->ke_thread != NULL), ("sched_add: No thread on KSE"));
1180 	KASSERT((ke->ke_thread->td_kse != NULL),
1181 	    ("sched_add: No KSE on thread"));
1182 	KASSERT(ke->ke_state != KES_ONRUNQ,
1183 	    ("sched_add: kse %p (%s) already in run queue", ke,
1184 	    ke->ke_proc->p_comm));
1185 	KASSERT(ke->ke_proc->p_sflag & PS_INMEM,
1186 	    ("sched_add: process swapped out"));
1187 	KASSERT(ke->ke_runq == NULL,
1188 	    ("sched_add: KSE %p is still assigned to a run queue", ke));
1189 
1190 	kg = ke->ke_ksegrp;
1191 
1192 	switch (PRI_BASE(kg->kg_pri_class)) {
1193 	case PRI_ITHD:
1194 	case PRI_REALTIME:
1195 		kseq = KSEQ_SELF();
1196 		ke->ke_runq = kseq->ksq_curr;
1197 		ke->ke_slice = SCHED_SLICE_MAX;
1198 		ke->ke_cpu = PCPU_GET(cpuid);
1199 		break;
1200 	case PRI_TIMESHARE:
1201 		kseq = KSEQ_CPU(ke->ke_cpu);
1202 		if (SCHED_CURR(kg, ke))
1203 			ke->ke_runq = kseq->ksq_curr;
1204 		else
1205 			ke->ke_runq = kseq->ksq_next;
1206 		break;
1207 	case PRI_IDLE:
1208 		kseq = KSEQ_CPU(ke->ke_cpu);
1209 		/*
1210 		 * This is for priority prop.
1211 		 */
1212 		if (ke->ke_thread->td_priority > PRI_MIN_IDLE)
1213 			ke->ke_runq = kseq->ksq_curr;
1214 		else
1215 			ke->ke_runq = &kseq->ksq_idle;
1216 		ke->ke_slice = SCHED_SLICE_MIN;
1217 		break;
1218 	default:
1219 		panic("Unknown pri class.\n");
1220 		break;
1221 	}
1222 
1223 	ke->ke_ksegrp->kg_runq_kses++;
1224 	ke->ke_state = KES_ONRUNQ;
1225 
1226 	runq_add(ke->ke_runq, ke);
1227 	kseq_add(kseq, ke);
1228 }
1229 
1230 void
1231 sched_rem(struct kse *ke)
1232 {
1233 	struct kseq *kseq;
1234 
1235 	mtx_assert(&sched_lock, MA_OWNED);
1236 	KASSERT((ke->ke_state == KES_ONRUNQ), ("KSE not on run queue"));
1237 
1238 	ke->ke_state = KES_THREAD;
1239 	ke->ke_ksegrp->kg_runq_kses--;
1240 	kseq = KSEQ_CPU(ke->ke_cpu);
1241 	runq_remove(ke->ke_runq, ke);
1242 	kseq_rem(kseq, ke);
1243 }
1244 
1245 fixpt_t
1246 sched_pctcpu(struct kse *ke)
1247 {
1248 	fixpt_t pctcpu;
1249 
1250 	pctcpu = 0;
1251 
1252 	mtx_lock_spin(&sched_lock);
1253 	if (ke->ke_ticks) {
1254 		int rtick;
1255 
1256 		/*
1257 		 * Don't update more frequently than twice a second.  Allowing
1258 		 * this causes the cpu usage to decay away too quickly due to
1259 		 * rounding errors.
1260 		 */
1261 		if (ke->ke_ltick < (ticks - (hz / 2)))
1262 			sched_pctcpu_update(ke);
1263 
1264 		/* How many rtick per second ? */
1265 		rtick = min(ke->ke_ticks / SCHED_CPU_TIME, SCHED_CPU_TICKS);
1266 		pctcpu = (FSCALE * ((FSCALE * rtick)/realstathz)) >> FSHIFT;
1267 	}
1268 
1269 	ke->ke_proc->p_swtime = ke->ke_ltick - ke->ke_ftick;
1270 	mtx_unlock_spin(&sched_lock);
1271 
1272 	return (pctcpu);
1273 }
1274 
1275 int
1276 sched_sizeof_kse(void)
1277 {
1278 	return (sizeof(struct kse) + sizeof(struct ke_sched));
1279 }
1280 
1281 int
1282 sched_sizeof_ksegrp(void)
1283 {
1284 	return (sizeof(struct ksegrp) + sizeof(struct kg_sched));
1285 }
1286 
1287 int
1288 sched_sizeof_proc(void)
1289 {
1290 	return (sizeof(struct proc));
1291 }
1292 
1293 int
1294 sched_sizeof_thread(void)
1295 {
1296 	return (sizeof(struct thread) + sizeof(struct td_sched));
1297 }
1298