xref: /freebsd/sys/kern/kern_switch.c (revision 830940567b49bb0c08dfaed40418999e76616909)
1 /*-
2  * Copyright (c) 2001 Jake Burkholder <jake@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, this list of conditions and the following disclaimer.
10  * 2. Redistributions in binary form must reproduce the above copyright
11  *    notice, this list of conditions and the following disclaimer in the
12  *    documentation and/or other materials provided with the distribution.
13  *
14  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
15  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
16  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
17  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
18  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
19  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
20  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
21  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
22  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
23  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
24  * SUCH DAMAGE.
25  */
26 
27 
28 #include <sys/cdefs.h>
29 __FBSDID("$FreeBSD$");
30 
31 #include "opt_sched.h"
32 
33 #include <sys/param.h>
34 #include <sys/systm.h>
35 #include <sys/kdb.h>
36 #include <sys/kernel.h>
37 #include <sys/ktr.h>
38 #include <sys/lock.h>
39 #include <sys/mutex.h>
40 #include <sys/proc.h>
41 #include <sys/queue.h>
42 #include <sys/sched.h>
43 #include <sys/smp.h>
44 #include <sys/sysctl.h>
45 
46 #include <machine/cpu.h>
47 
48 /* Uncomment this to enable logging of critical_enter/exit. */
49 #if 0
50 #define	KTR_CRITICAL	KTR_SCHED
51 #else
52 #define	KTR_CRITICAL	0
53 #endif
54 
55 #ifdef FULL_PREEMPTION
56 #ifndef PREEMPTION
57 #error "The FULL_PREEMPTION option requires the PREEMPTION option"
58 #endif
59 #endif
60 
61 CTASSERT((RQB_BPW * RQB_LEN) == RQ_NQS);
62 
63 /*
64  * kern.sched.preemption allows user space to determine if preemption support
65  * is compiled in or not.  It is not currently a boot or runtime flag that
66  * can be changed.
67  */
68 #ifdef PREEMPTION
69 static int kern_sched_preemption = 1;
70 #else
71 static int kern_sched_preemption = 0;
72 #endif
73 SYSCTL_INT(_kern_sched, OID_AUTO, preemption, CTLFLAG_RD,
74     &kern_sched_preemption, 0, "Kernel preemption enabled");
75 
76 /*
77  * Support for scheduler stats exported via kern.sched.stats.  All stats may
78  * be reset with kern.sched.stats.reset = 1.  Stats may be defined elsewhere
79  * with SCHED_STAT_DEFINE().
80  */
81 #ifdef SCHED_STATS
82 SYSCTL_NODE(_kern_sched, OID_AUTO, stats, CTLFLAG_RW, 0, "switch stats");
83 
84 /* Switch reasons from mi_switch(). */
85 DPCPU_DEFINE(long, sched_switch_stats[SWT_COUNT]);
86 SCHED_STAT_DEFINE_VAR(uncategorized,
87     &DPCPU_NAME(sched_switch_stats[SWT_NONE]), "");
88 SCHED_STAT_DEFINE_VAR(preempt,
89     &DPCPU_NAME(sched_switch_stats[SWT_PREEMPT]), "");
90 SCHED_STAT_DEFINE_VAR(owepreempt,
91     &DPCPU_NAME(sched_switch_stats[SWT_OWEPREEMPT]), "");
92 SCHED_STAT_DEFINE_VAR(turnstile,
93     &DPCPU_NAME(sched_switch_stats[SWT_TURNSTILE]), "");
94 SCHED_STAT_DEFINE_VAR(sleepq,
95     &DPCPU_NAME(sched_switch_stats[SWT_SLEEPQ]), "");
96 SCHED_STAT_DEFINE_VAR(sleepqtimo,
97     &DPCPU_NAME(sched_switch_stats[SWT_SLEEPQTIMO]), "");
98 SCHED_STAT_DEFINE_VAR(relinquish,
99     &DPCPU_NAME(sched_switch_stats[SWT_RELINQUISH]), "");
100 SCHED_STAT_DEFINE_VAR(needresched,
101     &DPCPU_NAME(sched_switch_stats[SWT_NEEDRESCHED]), "");
102 SCHED_STAT_DEFINE_VAR(idle,
103     &DPCPU_NAME(sched_switch_stats[SWT_IDLE]), "");
104 SCHED_STAT_DEFINE_VAR(iwait,
105     &DPCPU_NAME(sched_switch_stats[SWT_IWAIT]), "");
106 SCHED_STAT_DEFINE_VAR(suspend,
107     &DPCPU_NAME(sched_switch_stats[SWT_SUSPEND]), "");
108 SCHED_STAT_DEFINE_VAR(remotepreempt,
109     &DPCPU_NAME(sched_switch_stats[SWT_REMOTEPREEMPT]), "");
110 SCHED_STAT_DEFINE_VAR(remotewakeidle,
111     &DPCPU_NAME(sched_switch_stats[SWT_REMOTEWAKEIDLE]), "");
112 
113 static int
114 sysctl_stats_reset(SYSCTL_HANDLER_ARGS)
115 {
116 	struct sysctl_oid *p;
117 	uintptr_t counter;
118         int error;
119 	int val;
120 	int i;
121 
122         val = 0;
123         error = sysctl_handle_int(oidp, &val, 0, req);
124         if (error != 0 || req->newptr == NULL)
125                 return (error);
126         if (val == 0)
127                 return (0);
128 	/*
129 	 * Traverse the list of children of _kern_sched_stats and reset each
130 	 * to 0.  Skip the reset entry.
131 	 */
132 	SLIST_FOREACH(p, oidp->oid_parent, oid_link) {
133 		if (p == oidp || p->oid_arg1 == NULL)
134 			continue;
135 		counter = (uintptr_t)p->oid_arg1;
136 		for (i = 0; i <= mp_maxid; i++) {
137 			if (CPU_ABSENT(i))
138 				continue;
139 			*(long *)(dpcpu_off[i] + counter) = 0;
140 		}
141 	}
142 	return (0);
143 }
144 
145 SYSCTL_PROC(_kern_sched_stats, OID_AUTO, reset, CTLTYPE_INT | CTLFLAG_WR, NULL,
146     0, sysctl_stats_reset, "I", "Reset scheduler statistics");
147 #endif
148 
149 /************************************************************************
150  * Functions that manipulate runnability from a thread perspective.	*
151  ************************************************************************/
152 /*
153  * Select the thread that will be run next.
154  */
155 struct thread *
156 choosethread(void)
157 {
158 	struct thread *td;
159 
160 retry:
161 	td = sched_choose();
162 
163 	/*
164 	 * If we are in panic, only allow system threads,
165 	 * plus the one we are running in, to be run.
166 	 */
167 	if (panicstr && ((td->td_proc->p_flag & P_SYSTEM) == 0 &&
168 	    (td->td_flags & TDF_INPANIC) == 0)) {
169 		/* note that it is no longer on the run queue */
170 		TD_SET_CAN_RUN(td);
171 		goto retry;
172 	}
173 
174 	TD_SET_RUNNING(td);
175 	return (td);
176 }
177 
178 /*
179  * Kernel thread preemption implementation.  Critical sections mark
180  * regions of code in which preemptions are not allowed.
181  */
182 void
183 critical_enter(void)
184 {
185 	struct thread *td;
186 
187 	td = curthread;
188 	td->td_critnest++;
189 	CTR4(KTR_CRITICAL, "critical_enter by thread %p (%ld, %s) to %d", td,
190 	    (long)td->td_proc->p_pid, td->td_name, td->td_critnest);
191 }
192 
193 void
194 critical_exit(void)
195 {
196 	struct thread *td;
197 	int flags;
198 
199 	td = curthread;
200 	KASSERT(td->td_critnest != 0,
201 	    ("critical_exit: td_critnest == 0"));
202 
203 	if (td->td_critnest == 1) {
204 		td->td_critnest = 0;
205 		if (td->td_owepreempt) {
206 			td->td_critnest = 1;
207 			thread_lock(td);
208 			td->td_critnest--;
209 			flags = SW_INVOL | SW_PREEMPT;
210 			if (TD_IS_IDLETHREAD(td))
211 				flags |= SWT_IDLE;
212 			else
213 				flags |= SWT_OWEPREEMPT;
214 			mi_switch(flags, NULL);
215 			thread_unlock(td);
216 		}
217 	} else
218 		td->td_critnest--;
219 
220 	CTR4(KTR_CRITICAL, "critical_exit by thread %p (%ld, %s) to %d", td,
221 	    (long)td->td_proc->p_pid, td->td_name, td->td_critnest);
222 }
223 
224 /************************************************************************
225  * SYSTEM RUN QUEUE manipulations and tests				*
226  ************************************************************************/
227 /*
228  * Initialize a run structure.
229  */
230 void
231 runq_init(struct runq *rq)
232 {
233 	int i;
234 
235 	bzero(rq, sizeof *rq);
236 	for (i = 0; i < RQ_NQS; i++)
237 		TAILQ_INIT(&rq->rq_queues[i]);
238 }
239 
240 /*
241  * Clear the status bit of the queue corresponding to priority level pri,
242  * indicating that it is empty.
243  */
244 static __inline void
245 runq_clrbit(struct runq *rq, int pri)
246 {
247 	struct rqbits *rqb;
248 
249 	rqb = &rq->rq_status;
250 	CTR4(KTR_RUNQ, "runq_clrbit: bits=%#x %#x bit=%#x word=%d",
251 	    rqb->rqb_bits[RQB_WORD(pri)],
252 	    rqb->rqb_bits[RQB_WORD(pri)] & ~RQB_BIT(pri),
253 	    RQB_BIT(pri), RQB_WORD(pri));
254 	rqb->rqb_bits[RQB_WORD(pri)] &= ~RQB_BIT(pri);
255 }
256 
257 /*
258  * Find the index of the first non-empty run queue.  This is done by
259  * scanning the status bits, a set bit indicates a non-empty queue.
260  */
261 static __inline int
262 runq_findbit(struct runq *rq)
263 {
264 	struct rqbits *rqb;
265 	int pri;
266 	int i;
267 
268 	rqb = &rq->rq_status;
269 	for (i = 0; i < RQB_LEN; i++)
270 		if (rqb->rqb_bits[i]) {
271 			pri = RQB_FFS(rqb->rqb_bits[i]) + (i << RQB_L2BPW);
272 			CTR3(KTR_RUNQ, "runq_findbit: bits=%#x i=%d pri=%d",
273 			    rqb->rqb_bits[i], i, pri);
274 			return (pri);
275 		}
276 
277 	return (-1);
278 }
279 
280 static __inline int
281 runq_findbit_from(struct runq *rq, u_char pri)
282 {
283 	struct rqbits *rqb;
284 	rqb_word_t mask;
285 	int i;
286 
287 	/*
288 	 * Set the mask for the first word so we ignore priorities before 'pri'.
289 	 */
290 	mask = (rqb_word_t)-1 << (pri & (RQB_BPW - 1));
291 	rqb = &rq->rq_status;
292 again:
293 	for (i = RQB_WORD(pri); i < RQB_LEN; mask = -1, i++) {
294 		mask = rqb->rqb_bits[i] & mask;
295 		if (mask == 0)
296 			continue;
297 		pri = RQB_FFS(mask) + (i << RQB_L2BPW);
298 		CTR3(KTR_RUNQ, "runq_findbit_from: bits=%#x i=%d pri=%d",
299 		    mask, i, pri);
300 		return (pri);
301 	}
302 	if (pri == 0)
303 		return (-1);
304 	/*
305 	 * Wrap back around to the beginning of the list just once so we
306 	 * scan the whole thing.
307 	 */
308 	pri = 0;
309 	goto again;
310 }
311 
312 /*
313  * Set the status bit of the queue corresponding to priority level pri,
314  * indicating that it is non-empty.
315  */
316 static __inline void
317 runq_setbit(struct runq *rq, int pri)
318 {
319 	struct rqbits *rqb;
320 
321 	rqb = &rq->rq_status;
322 	CTR4(KTR_RUNQ, "runq_setbit: bits=%#x %#x bit=%#x word=%d",
323 	    rqb->rqb_bits[RQB_WORD(pri)],
324 	    rqb->rqb_bits[RQB_WORD(pri)] | RQB_BIT(pri),
325 	    RQB_BIT(pri), RQB_WORD(pri));
326 	rqb->rqb_bits[RQB_WORD(pri)] |= RQB_BIT(pri);
327 }
328 
329 /*
330  * Add the thread to the queue specified by its priority, and set the
331  * corresponding status bit.
332  */
333 void
334 runq_add(struct runq *rq, struct thread *td, int flags)
335 {
336 	struct rqhead *rqh;
337 	int pri;
338 
339 	pri = td->td_priority / RQ_PPQ;
340 	td->td_rqindex = pri;
341 	runq_setbit(rq, pri);
342 	rqh = &rq->rq_queues[pri];
343 	CTR4(KTR_RUNQ, "runq_add: td=%p pri=%d %d rqh=%p",
344 	    td, td->td_priority, pri, rqh);
345 	if (flags & SRQ_PREEMPTED) {
346 		TAILQ_INSERT_HEAD(rqh, td, td_runq);
347 	} else {
348 		TAILQ_INSERT_TAIL(rqh, td, td_runq);
349 	}
350 }
351 
352 void
353 runq_add_pri(struct runq *rq, struct thread *td, u_char pri, int flags)
354 {
355 	struct rqhead *rqh;
356 
357 	KASSERT(pri < RQ_NQS, ("runq_add_pri: %d out of range", pri));
358 	td->td_rqindex = pri;
359 	runq_setbit(rq, pri);
360 	rqh = &rq->rq_queues[pri];
361 	CTR4(KTR_RUNQ, "runq_add_pri: td=%p pri=%d idx=%d rqh=%p",
362 	    td, td->td_priority, pri, rqh);
363 	if (flags & SRQ_PREEMPTED) {
364 		TAILQ_INSERT_HEAD(rqh, td, td_runq);
365 	} else {
366 		TAILQ_INSERT_TAIL(rqh, td, td_runq);
367 	}
368 }
369 /*
370  * Return true if there are runnable processes of any priority on the run
371  * queue, false otherwise.  Has no side effects, does not modify the run
372  * queue structure.
373  */
374 int
375 runq_check(struct runq *rq)
376 {
377 	struct rqbits *rqb;
378 	int i;
379 
380 	rqb = &rq->rq_status;
381 	for (i = 0; i < RQB_LEN; i++)
382 		if (rqb->rqb_bits[i]) {
383 			CTR2(KTR_RUNQ, "runq_check: bits=%#x i=%d",
384 			    rqb->rqb_bits[i], i);
385 			return (1);
386 		}
387 	CTR0(KTR_RUNQ, "runq_check: empty");
388 
389 	return (0);
390 }
391 
392 /*
393  * Find the highest priority process on the run queue.
394  */
395 struct thread *
396 runq_choose_fuzz(struct runq *rq, int fuzz)
397 {
398 	struct rqhead *rqh;
399 	struct thread *td;
400 	int pri;
401 
402 	while ((pri = runq_findbit(rq)) != -1) {
403 		rqh = &rq->rq_queues[pri];
404 		/* fuzz == 1 is normal.. 0 or less are ignored */
405 		if (fuzz > 1) {
406 			/*
407 			 * In the first couple of entries, check if
408 			 * there is one for our CPU as a preference.
409 			 */
410 			int count = fuzz;
411 			int cpu = PCPU_GET(cpuid);
412 			struct thread *td2;
413 			td2 = td = TAILQ_FIRST(rqh);
414 
415 			while (count-- && td2) {
416 				if (td2->td_lastcpu == cpu) {
417 					td = td2;
418 					break;
419 				}
420 				td2 = TAILQ_NEXT(td2, td_runq);
421 			}
422 		} else
423 			td = TAILQ_FIRST(rqh);
424 		KASSERT(td != NULL, ("runq_choose_fuzz: no proc on busy queue"));
425 		CTR3(KTR_RUNQ,
426 		    "runq_choose_fuzz: pri=%d thread=%p rqh=%p", pri, td, rqh);
427 		return (td);
428 	}
429 	CTR1(KTR_RUNQ, "runq_choose_fuzz: idleproc pri=%d", pri);
430 
431 	return (NULL);
432 }
433 
434 /*
435  * Find the highest priority process on the run queue.
436  */
437 struct thread *
438 runq_choose(struct runq *rq)
439 {
440 	struct rqhead *rqh;
441 	struct thread *td;
442 	int pri;
443 
444 	while ((pri = runq_findbit(rq)) != -1) {
445 		rqh = &rq->rq_queues[pri];
446 		td = TAILQ_FIRST(rqh);
447 		KASSERT(td != NULL, ("runq_choose: no thread on busy queue"));
448 		CTR3(KTR_RUNQ,
449 		    "runq_choose: pri=%d thread=%p rqh=%p", pri, td, rqh);
450 		return (td);
451 	}
452 	CTR1(KTR_RUNQ, "runq_choose: idlethread pri=%d", pri);
453 
454 	return (NULL);
455 }
456 
457 struct thread *
458 runq_choose_from(struct runq *rq, u_char idx)
459 {
460 	struct rqhead *rqh;
461 	struct thread *td;
462 	int pri;
463 
464 	if ((pri = runq_findbit_from(rq, idx)) != -1) {
465 		rqh = &rq->rq_queues[pri];
466 		td = TAILQ_FIRST(rqh);
467 		KASSERT(td != NULL, ("runq_choose: no thread on busy queue"));
468 		CTR4(KTR_RUNQ,
469 		    "runq_choose_from: pri=%d thread=%p idx=%d rqh=%p",
470 		    pri, td, td->td_rqindex, rqh);
471 		return (td);
472 	}
473 	CTR1(KTR_RUNQ, "runq_choose_from: idlethread pri=%d", pri);
474 
475 	return (NULL);
476 }
477 /*
478  * Remove the thread from the queue specified by its priority, and clear the
479  * corresponding status bit if the queue becomes empty.
480  * Caller must set state afterwards.
481  */
482 void
483 runq_remove(struct runq *rq, struct thread *td)
484 {
485 
486 	runq_remove_idx(rq, td, NULL);
487 }
488 
489 void
490 runq_remove_idx(struct runq *rq, struct thread *td, u_char *idx)
491 {
492 	struct rqhead *rqh;
493 	u_char pri;
494 
495 	KASSERT(td->td_flags & TDF_INMEM,
496 		("runq_remove_idx: thread swapped out"));
497 	pri = td->td_rqindex;
498 	KASSERT(pri < RQ_NQS, ("runq_remove_idx: Invalid index %d\n", pri));
499 	rqh = &rq->rq_queues[pri];
500 	CTR4(KTR_RUNQ, "runq_remove_idx: td=%p, pri=%d %d rqh=%p",
501 	    td, td->td_priority, pri, rqh);
502 	TAILQ_REMOVE(rqh, td, td_runq);
503 	if (TAILQ_EMPTY(rqh)) {
504 		CTR0(KTR_RUNQ, "runq_remove_idx: empty");
505 		runq_clrbit(rq, pri);
506 		if (idx != NULL && *idx == pri)
507 			*idx = (pri + 1) % RQ_NQS;
508 	}
509 }
510