kern_synch.c revision 1.177.2.19 1 1.177.2.19 rmind /* $NetBSD: kern_synch.c,v 1.177.2.19 2007/03/24 16:50:26 rmind Exp $ */
2 1.63 thorpej
3 1.63 thorpej /*-
4 1.174 ad * Copyright (c) 1999, 2000, 2004, 2006, 2007 The NetBSD Foundation, Inc.
5 1.63 thorpej * All rights reserved.
6 1.63 thorpej *
7 1.63 thorpej * This code is derived from software contributed to The NetBSD Foundation
8 1.63 thorpej * by Jason R. Thorpe of the Numerical Aerospace Simulation Facility,
9 1.177.2.5 rmind * NASA Ames Research Center, by Charles M. Hannum, Andrew Doran and
10 1.177.2.5 rmind * Daniel Sieger.
11 1.63 thorpej *
12 1.63 thorpej * Redistribution and use in source and binary forms, with or without
13 1.63 thorpej * modification, are permitted provided that the following conditions
14 1.63 thorpej * are met:
15 1.63 thorpej * 1. Redistributions of source code must retain the above copyright
16 1.63 thorpej * notice, this list of conditions and the following disclaimer.
17 1.63 thorpej * 2. Redistributions in binary form must reproduce the above copyright
18 1.63 thorpej * notice, this list of conditions and the following disclaimer in the
19 1.63 thorpej * documentation and/or other materials provided with the distribution.
20 1.63 thorpej * 3. All advertising materials mentioning features or use of this software
21 1.63 thorpej * must display the following acknowledgement:
22 1.63 thorpej * This product includes software developed by the NetBSD
23 1.63 thorpej * Foundation, Inc. and its contributors.
24 1.63 thorpej * 4. Neither the name of The NetBSD Foundation nor the names of its
25 1.63 thorpej * contributors may be used to endorse or promote products derived
26 1.63 thorpej * from this software without specific prior written permission.
27 1.63 thorpej *
28 1.63 thorpej * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
29 1.63 thorpej * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
30 1.63 thorpej * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
31 1.63 thorpej * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
32 1.63 thorpej * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
33 1.63 thorpej * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
34 1.63 thorpej * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
35 1.63 thorpej * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
36 1.63 thorpej * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
37 1.63 thorpej * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
38 1.63 thorpej * POSSIBILITY OF SUCH DAMAGE.
39 1.63 thorpej */
40 1.26 cgd
41 1.26 cgd /*-
42 1.26 cgd * Copyright (c) 1982, 1986, 1990, 1991, 1993
43 1.26 cgd * The Regents of the University of California. All rights reserved.
44 1.26 cgd * (c) UNIX System Laboratories, Inc.
45 1.26 cgd * All or some portions of this file are derived from material licensed
46 1.26 cgd * to the University of California by American Telephone and Telegraph
47 1.26 cgd * Co. or Unix System Laboratories, Inc. and are reproduced herein with
48 1.26 cgd * the permission of UNIX System Laboratories, Inc.
49 1.26 cgd *
50 1.26 cgd * Redistribution and use in source and binary forms, with or without
51 1.26 cgd * modification, are permitted provided that the following conditions
52 1.26 cgd * are met:
53 1.26 cgd * 1. Redistributions of source code must retain the above copyright
54 1.26 cgd * notice, this list of conditions and the following disclaimer.
55 1.26 cgd * 2. Redistributions in binary form must reproduce the above copyright
56 1.26 cgd * notice, this list of conditions and the following disclaimer in the
57 1.26 cgd * documentation and/or other materials provided with the distribution.
58 1.136 agc * 3. Neither the name of the University nor the names of its contributors
59 1.26 cgd * may be used to endorse or promote products derived from this software
60 1.26 cgd * without specific prior written permission.
61 1.26 cgd *
62 1.26 cgd * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
63 1.26 cgd * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
64 1.26 cgd * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
65 1.26 cgd * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
66 1.26 cgd * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
67 1.26 cgd * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
68 1.26 cgd * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
69 1.26 cgd * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
70 1.26 cgd * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
71 1.26 cgd * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
72 1.26 cgd * SUCH DAMAGE.
73 1.26 cgd *
74 1.50 fvdl * @(#)kern_synch.c 8.9 (Berkeley) 5/19/95
75 1.26 cgd */
76 1.106 lukem
77 1.106 lukem #include <sys/cdefs.h>
78 1.177.2.19 rmind __KERNEL_RCSID(0, "$NetBSD: kern_synch.c,v 1.177.2.19 2007/03/24 16:50:26 rmind Exp $");
79 1.48 mrg
80 1.109 yamt #include "opt_kstack.h"
81 1.82 thorpej #include "opt_lockdebug.h"
82 1.83 thorpej #include "opt_multiprocessor.h"
83 1.110 briggs #include "opt_perfctrs.h"
84 1.26 cgd
85 1.174 ad #define __MUTEX_PRIVATE
86 1.174 ad
87 1.26 cgd #include <sys/param.h>
88 1.26 cgd #include <sys/systm.h>
89 1.26 cgd #include <sys/proc.h>
90 1.26 cgd #include <sys/kernel.h>
91 1.111 briggs #if defined(PERFCTRS)
92 1.110 briggs #include <sys/pmc.h>
93 1.111 briggs #endif
94 1.177.2.8 yamt #include <sys/cpu.h>
95 1.26 cgd #include <sys/resourcevar.h>
96 1.55 ross #include <sys/sched.h>
97 1.177.2.10 yamt #include <sys/syscall_stats.h>
98 1.174 ad #include <sys/sleepq.h>
99 1.174 ad #include <sys/lockdebug.h>
100 1.47 mrg
101 1.47 mrg #include <uvm/uvm_extern.h>
102 1.47 mrg
103 1.26 cgd int lbolt; /* once a second sleep address */
104 1.26 cgd
105 1.177.2.10 yamt static void sched_unsleep(struct lwp *);
106 1.177.2.10 yamt static void sched_changepri(struct lwp *, pri_t);
107 1.177.2.10 yamt static void sched_lendpri(struct lwp *, pri_t);
108 1.122 thorpej
109 1.174 ad syncobj_t sleep_syncobj = {
110 1.174 ad SOBJ_SLEEPQ_SORTED,
111 1.174 ad sleepq_unsleep,
112 1.177.2.10 yamt sleepq_changepri,
113 1.177.2.10 yamt sleepq_lendpri,
114 1.177.2.10 yamt syncobj_noowner,
115 1.174 ad };
116 1.174 ad
117 1.174 ad syncobj_t sched_syncobj = {
118 1.174 ad SOBJ_SLEEPQ_SORTED,
119 1.174 ad sched_unsleep,
120 1.177.2.10 yamt sched_changepri,
121 1.177.2.10 yamt sched_lendpri,
122 1.177.2.10 yamt syncobj_noowner,
123 1.174 ad };
124 1.122 thorpej
125 1.26 cgd /*
126 1.174 ad * During autoconfiguration or after a panic, a sleep will simply lower the
127 1.174 ad * priority briefly to allow interrupts, then return. The priority to be
128 1.174 ad * used (safepri) is machine-dependent, thus this value is initialized and
129 1.174 ad * maintained in the machine-dependent layers. This priority will typically
130 1.174 ad * be 0, or the lowest priority that is safe for use on the interrupt stack;
131 1.174 ad * it can be made higher to block network software interrupts after panics.
132 1.26 cgd */
133 1.174 ad int safepri;
134 1.26 cgd
135 1.26 cgd /*
136 1.174 ad * OBSOLETE INTERFACE
137 1.174 ad *
138 1.26 cgd * General sleep call. Suspends the current process until a wakeup is
139 1.26 cgd * performed on the specified identifier. The process will then be made
140 1.174 ad * runnable with the specified priority. Sleeps at most timo/hz seconds (0
141 1.174 ad * means no timeout). If pri includes PCATCH flag, signals are checked
142 1.26 cgd * before and after sleeping, else signals are not checked. Returns 0 if
143 1.26 cgd * awakened, EWOULDBLOCK if the timeout expires. If PCATCH is set and a
144 1.26 cgd * signal needs to be delivered, ERESTART is returned if the current system
145 1.26 cgd * call should be restarted if possible, and EINTR is returned if the system
146 1.26 cgd * call should be interrupted by the signal (return EINTR).
147 1.77 thorpej *
148 1.174 ad * The interlock is held until we are on a sleep queue. The interlock will
149 1.174 ad * be locked before returning back to the caller unless the PNORELOCK flag
150 1.174 ad * is specified, in which case the interlock will always be unlocked upon
151 1.174 ad * return.
152 1.26 cgd */
153 1.26 cgd int
154 1.177.2.10 yamt ltsleep(wchan_t ident, pri_t priority, const char *wmesg, int timo,
155 1.174 ad volatile struct simplelock *interlock)
156 1.26 cgd {
157 1.122 thorpej struct lwp *l = curlwp;
158 1.174 ad sleepq_t *sq;
159 1.174 ad int error, catch;
160 1.26 cgd
161 1.174 ad if (sleepq_dontsleep(l)) {
162 1.174 ad (void)sleepq_abort(NULL, 0);
163 1.174 ad if ((priority & PNORELOCK) != 0)
164 1.77 thorpej simple_unlock(interlock);
165 1.174 ad return 0;
166 1.26 cgd }
167 1.78 sommerfe
168 1.174 ad sq = sleeptab_lookup(&sleeptab, ident);
169 1.174 ad sleepq_enter(sq, l);
170 1.42 cgd
171 1.174 ad if (interlock != NULL) {
172 1.174 ad LOCK_ASSERT(simple_lock_held(interlock));
173 1.174 ad simple_unlock(interlock);
174 1.150 chs }
175 1.150 chs
176 1.174 ad catch = priority & PCATCH;
177 1.174 ad sleepq_block(sq, priority & PRIMASK, ident, wmesg, timo, catch,
178 1.174 ad &sleep_syncobj);
179 1.174 ad error = sleepq_unblock(timo, catch);
180 1.126 pk
181 1.174 ad if (interlock != NULL && (priority & PNORELOCK) == 0)
182 1.126 pk simple_lock(interlock);
183 1.174 ad
184 1.174 ad return error;
185 1.26 cgd }
186 1.26 cgd
187 1.177.2.15 rmind int
188 1.177.2.15 rmind mtsleep(wchan_t ident, pri_t priority, const char *wmesg, int timo,
189 1.177.2.15 rmind kmutex_t *mtx)
190 1.177.2.15 rmind {
191 1.177.2.15 rmind struct lwp *l = curlwp;
192 1.177.2.15 rmind sleepq_t *sq;
193 1.177.2.15 rmind int error, catch;
194 1.177.2.15 rmind
195 1.177.2.15 rmind if (sleepq_dontsleep(l)) {
196 1.177.2.15 rmind (void)sleepq_abort(mtx, (priority & PNORELOCK) != 0);
197 1.177.2.15 rmind return 0;
198 1.177.2.15 rmind }
199 1.177.2.15 rmind
200 1.177.2.15 rmind sq = sleeptab_lookup(&sleeptab, ident);
201 1.177.2.15 rmind sleepq_enter(sq, l);
202 1.177.2.15 rmind mutex_exit(mtx);
203 1.177.2.15 rmind
204 1.177.2.15 rmind catch = priority & PCATCH;
205 1.177.2.15 rmind sleepq_block(sq, priority & PRIMASK, ident, wmesg, timo, catch,
206 1.177.2.15 rmind &sleep_syncobj);
207 1.177.2.15 rmind error = sleepq_unblock(timo, catch);
208 1.177.2.15 rmind
209 1.177.2.15 rmind if ((priority & PNORELOCK) == 0)
210 1.177.2.15 rmind mutex_enter(mtx);
211 1.177.2.15 rmind
212 1.177.2.15 rmind return error;
213 1.177.2.15 rmind }
214 1.177.2.15 rmind
215 1.26 cgd /*
216 1.174 ad * General sleep call for situations where a wake-up is not expected.
217 1.26 cgd */
218 1.174 ad int
219 1.177.2.10 yamt kpause(const char *wmesg, bool intr, int timo, kmutex_t *mtx)
220 1.26 cgd {
221 1.174 ad struct lwp *l = curlwp;
222 1.174 ad sleepq_t *sq;
223 1.174 ad int error;
224 1.26 cgd
225 1.174 ad if (sleepq_dontsleep(l))
226 1.174 ad return sleepq_abort(NULL, 0);
227 1.26 cgd
228 1.174 ad if (mtx != NULL)
229 1.174 ad mutex_exit(mtx);
230 1.174 ad sq = sleeptab_lookup(&sleeptab, l);
231 1.174 ad sleepq_enter(sq, l);
232 1.174 ad sleepq_block(sq, sched_kpri(l), l, wmesg, timo, intr, &sleep_syncobj);
233 1.174 ad error = sleepq_unblock(timo, intr);
234 1.174 ad if (mtx != NULL)
235 1.174 ad mutex_enter(mtx);
236 1.83 thorpej
237 1.174 ad return error;
238 1.139 cl }
239 1.139 cl
240 1.26 cgd /*
241 1.174 ad * OBSOLETE INTERFACE
242 1.174 ad *
243 1.26 cgd * Make all processes sleeping on the specified identifier runnable.
244 1.26 cgd */
245 1.26 cgd void
246 1.174 ad wakeup(wchan_t ident)
247 1.26 cgd {
248 1.174 ad sleepq_t *sq;
249 1.83 thorpej
250 1.174 ad if (cold)
251 1.174 ad return;
252 1.83 thorpej
253 1.174 ad sq = sleeptab_lookup(&sleeptab, ident);
254 1.174 ad sleepq_wake(sq, ident, (u_int)-1);
255 1.63 thorpej }
256 1.63 thorpej
257 1.63 thorpej /*
258 1.174 ad * OBSOLETE INTERFACE
259 1.174 ad *
260 1.63 thorpej * Make the highest priority process first in line on the specified
261 1.63 thorpej * identifier runnable.
262 1.63 thorpej */
263 1.174 ad void
264 1.174 ad wakeup_one(wchan_t ident)
265 1.63 thorpej {
266 1.174 ad sleepq_t *sq;
267 1.63 thorpej
268 1.174 ad if (cold)
269 1.174 ad return;
270 1.174 ad
271 1.174 ad sq = sleeptab_lookup(&sleeptab, ident);
272 1.174 ad sleepq_wake(sq, ident, 1);
273 1.174 ad }
274 1.63 thorpej
275 1.117 gmcgarry
276 1.117 gmcgarry /*
277 1.117 gmcgarry * General yield call. Puts the current process back on its run queue and
278 1.117 gmcgarry * performs a voluntary context switch. Should only be called when the
279 1.117 gmcgarry * current process explicitly requests it (eg sched_yield(2) in compat code).
280 1.117 gmcgarry */
281 1.117 gmcgarry void
282 1.117 gmcgarry yield(void)
283 1.117 gmcgarry {
284 1.122 thorpej struct lwp *l = curlwp;
285 1.117 gmcgarry
286 1.174 ad KERNEL_UNLOCK_ALL(l, &l->l_biglocks);
287 1.174 ad lwp_lock(l);
288 1.174 ad if (l->l_stat == LSONPROC) {
289 1.177.2.18 rmind KASSERT(lwp_locked(l, l->l_cpu->ci_schedstate.spc_mutex));
290 1.174 ad l->l_priority = l->l_usrpri;
291 1.174 ad }
292 1.174 ad l->l_nvcsw++;
293 1.177.2.13 rmind mi_switch(l);
294 1.174 ad KERNEL_LOCK(l->l_biglocks, l);
295 1.69 thorpej }
296 1.69 thorpej
297 1.69 thorpej /*
298 1.69 thorpej * General preemption call. Puts the current process back on its run queue
299 1.156 rpaulo * and performs an involuntary context switch.
300 1.69 thorpej */
301 1.69 thorpej void
302 1.174 ad preempt(void)
303 1.69 thorpej {
304 1.122 thorpej struct lwp *l = curlwp;
305 1.69 thorpej
306 1.174 ad KERNEL_UNLOCK_ALL(l, &l->l_biglocks);
307 1.174 ad lwp_lock(l);
308 1.174 ad if (l->l_stat == LSONPROC) {
309 1.177.2.18 rmind KASSERT(lwp_locked(l, l->l_cpu->ci_schedstate.spc_mutex));
310 1.174 ad l->l_priority = l->l_usrpri;
311 1.174 ad }
312 1.174 ad l->l_nivcsw++;
313 1.177.2.13 rmind (void)mi_switch(l);
314 1.174 ad KERNEL_LOCK(l->l_biglocks, l);
315 1.69 thorpej }
316 1.69 thorpej
317 1.69 thorpej /*
318 1.177.2.1 yamt * sched_switch_unlock: update 'curlwp' and release old lwp.
319 1.177.2.1 yamt */
320 1.177.2.1 yamt
321 1.177.2.1 yamt void
322 1.177.2.1 yamt sched_switch_unlock(struct lwp *old, struct lwp *new)
323 1.177.2.1 yamt {
324 1.177.2.1 yamt
325 1.177.2.1 yamt KASSERT(old == NULL || old == curlwp);
326 1.177.2.12 yamt KASSERT(new != NULL);
327 1.177.2.1 yamt
328 1.177.2.1 yamt if (old != NULL) {
329 1.177.2.4 yamt LOCKDEBUG_BARRIER(old->l_mutex, 1);
330 1.177.2.17 ad lwp_unlock(old);
331 1.177.2.1 yamt } else {
332 1.177.2.1 yamt LOCKDEBUG_BARRIER(NULL, 1);
333 1.177.2.1 yamt }
334 1.177.2.1 yamt curlwp = new;
335 1.177.2.1 yamt spl0();
336 1.177.2.1 yamt }
337 1.177.2.1 yamt
338 1.177.2.1 yamt /*
339 1.177.2.3 yamt * Compute the amount of time during which the current lwp was running.
340 1.177.2.3 yamt *
341 1.177.2.3 yamt * - update l_rtime unless it's an idle lwp.
342 1.177.2.3 yamt * - update spc_runtime for the next lwp.
343 1.177.2.3 yamt */
344 1.177.2.3 yamt
345 1.177.2.3 yamt static inline void
346 1.177.2.3 yamt updatertime(struct lwp *l, struct schedstate_percpu *spc)
347 1.177.2.3 yamt {
348 1.177.2.3 yamt struct timeval tv;
349 1.177.2.3 yamt long s, u;
350 1.177.2.3 yamt
351 1.177.2.10 yamt if ((l->l_flag & LW_IDLE) != 0) {
352 1.177.2.3 yamt microtime(&spc->spc_runtime);
353 1.177.2.3 yamt return;
354 1.177.2.3 yamt }
355 1.177.2.3 yamt
356 1.177.2.3 yamt microtime(&tv);
357 1.177.2.3 yamt u = l->l_rtime.tv_usec + (tv.tv_usec - spc->spc_runtime.tv_usec);
358 1.177.2.3 yamt s = l->l_rtime.tv_sec + (tv.tv_sec - spc->spc_runtime.tv_sec);
359 1.177.2.3 yamt if (u < 0) {
360 1.177.2.3 yamt u += 1000000;
361 1.177.2.3 yamt s--;
362 1.177.2.3 yamt } else if (u >= 1000000) {
363 1.177.2.3 yamt u -= 1000000;
364 1.177.2.3 yamt s++;
365 1.177.2.3 yamt }
366 1.177.2.3 yamt l->l_rtime.tv_usec = u;
367 1.177.2.3 yamt l->l_rtime.tv_sec = s;
368 1.177.2.3 yamt
369 1.177.2.3 yamt spc->spc_runtime = tv;
370 1.177.2.3 yamt }
371 1.177.2.3 yamt
372 1.177.2.3 yamt /*
373 1.177.2.14 rmind * The machine independent parts of context switch.
374 1.130 nathanw *
375 1.122 thorpej * Returns 1 if another process was actually run.
376 1.26 cgd */
377 1.122 thorpej int
378 1.177.2.13 rmind mi_switch(struct lwp *l)
379 1.26 cgd {
380 1.76 thorpej struct schedstate_percpu *spc;
381 1.177.2.13 rmind struct lwp *newl;
382 1.174 ad int retval, oldspl;
383 1.26 cgd
384 1.174 ad LOCK_ASSERT(lwp_locked(l, NULL));
385 1.174 ad
386 1.174 ad #ifdef LOCKDEBUG
387 1.174 ad spinlock_switchcheck();
388 1.174 ad simple_lock_switchcheck();
389 1.174 ad #endif
390 1.174 ad #ifdef KSTACK_CHECK_MAGIC
391 1.174 ad kstack_check_magic(l);
392 1.174 ad #endif
393 1.83 thorpej
394 1.90 sommerfe /*
395 1.174 ad * It's safe to read the per CPU schedstate unlocked here, as all we
396 1.174 ad * are after is the run time and that's guarenteed to have been last
397 1.174 ad * updated by this CPU.
398 1.90 sommerfe */
399 1.122 thorpej KDASSERT(l->l_cpu == curcpu());
400 1.122 thorpej spc = &l->l_cpu->ci_schedstate;
401 1.76 thorpej
402 1.177.2.10 yamt /* Count time spent in current system call */
403 1.177.2.10 yamt SYSCALL_TIME_SLEEP(l);
404 1.177.2.10 yamt
405 1.26 cgd /*
406 1.174 ad * XXXSMP If we are using h/w performance counters, save context.
407 1.69 thorpej */
408 1.174 ad #if PERFCTRS
409 1.175 christos if (PMC_ENABLED(l->l_proc)) {
410 1.175 christos pmc_save_context(l->l_proc);
411 1.174 ad }
412 1.109 yamt #endif
413 1.26 cgd
414 1.113 gmcgarry /*
415 1.177.2.16 rmind * If on the CPU and we have gotten this far, then we must yield.
416 1.177.2.16 rmind */
417 1.177.2.16 rmind KASSERT(l->l_stat != LSRUN);
418 1.177.2.16 rmind if (l->l_stat == LSONPROC) {
419 1.177.2.18 rmind KASSERT(lwp_locked(l, l->l_cpu->ci_schedstate.spc_mutex));
420 1.177.2.16 rmind l->l_stat = LSRUN;
421 1.177.2.16 rmind if ((l->l_flag & LW_IDLE) == 0) {
422 1.177.2.16 rmind sched_enqueue(l, true);
423 1.177.2.16 rmind }
424 1.177.2.16 rmind }
425 1.177.2.16 rmind
426 1.177.2.16 rmind /*
427 1.174 ad * Process is about to yield the CPU; clear the appropriate
428 1.174 ad * scheduling flags.
429 1.174 ad */
430 1.174 ad spc->spc_flags &= ~SPCF_SWITCHCLEAR;
431 1.174 ad
432 1.177.2.2 yamt LOCKDEBUG_BARRIER(l->l_mutex, 1);
433 1.174 ad
434 1.174 ad /*
435 1.177.2.1 yamt * Switch to the new LWP if necessary.
436 1.177.2.1 yamt * When we run again, we'll return back here.
437 1.174 ad */
438 1.174 ad oldspl = MUTEX_SPIN_OLDSPL(l->l_cpu);
439 1.174 ad
440 1.177.2.2 yamt /*
441 1.177.2.18 rmind * Acquire the spc_mutex if necessary.
442 1.177.2.2 yamt */
443 1.177.2.2 yamt #if defined(MULTIPROCESSOR) || defined(LOCKDEBUG)
444 1.177.2.18 rmind if (l->l_mutex != spc->spc_mutex) {
445 1.177.2.18 rmind mutex_enter(spc->spc_mutex);
446 1.177.2.2 yamt }
447 1.177.2.2 yamt #endif
448 1.177.2.14 rmind /*
449 1.177.2.16 rmind * Let sched_nextlwp() select the LWP to run the CPU next.
450 1.177.2.14 rmind * If no LWP is runnable, switch to the idle LWP.
451 1.177.2.14 rmind */
452 1.177.2.19 rmind newl = sched_nextlwp();
453 1.177.2.16 rmind if (newl) {
454 1.177.2.16 rmind sched_dequeue(newl);
455 1.177.2.16 rmind } else {
456 1.177.2.1 yamt newl = l->l_cpu->ci_data.cpu_idlelwp;
457 1.177.2.1 yamt KASSERT(newl != NULL);
458 1.177.2.1 yamt }
459 1.177.2.18 rmind KASSERT(lwp_locked(newl, spc->spc_mutex));
460 1.177.2.7 yamt newl->l_stat = LSONPROC;
461 1.177.2.7 yamt newl->l_cpu = l->l_cpu;
462 1.177.2.17 ad newl->l_flag |= LW_RUNNING;
463 1.177.2.2 yamt
464 1.177.2.2 yamt #if defined(MULTIPROCESSOR) || defined(LOCKDEBUG)
465 1.177.2.18 rmind if (l->l_mutex != spc->spc_mutex) {
466 1.177.2.18 rmind mutex_exit(spc->spc_mutex);
467 1.177.2.2 yamt }
468 1.177.2.2 yamt #endif
469 1.177.2.2 yamt
470 1.177.2.3 yamt updatertime(l, spc);
471 1.177.2.1 yamt if (l != newl) {
472 1.177.2.1 yamt struct lwp *prevlwp;
473 1.177.2.1 yamt
474 1.177.2.1 yamt uvmexp.swtch++;
475 1.177.2.1 yamt pmap_deactivate(l);
476 1.177.2.17 ad l->l_flag &= ~LW_RUNNING;
477 1.177.2.1 yamt prevlwp = cpu_switchto(l, newl);
478 1.177.2.1 yamt sched_switch_unlock(prevlwp, l);
479 1.177.2.1 yamt pmap_activate(l);
480 1.177.2.1 yamt retval = 1;
481 1.177.2.1 yamt } else {
482 1.177.2.1 yamt sched_switch_unlock(l, l);
483 1.122 thorpej retval = 0;
484 1.122 thorpej }
485 1.110 briggs
486 1.177.2.1 yamt KASSERT(l == curlwp);
487 1.177.2.1 yamt KASSERT(l->l_stat == LSONPROC);
488 1.177.2.1 yamt
489 1.110 briggs /*
490 1.174 ad * XXXSMP If we are using h/w performance counters, restore context.
491 1.26 cgd */
492 1.114 gmcgarry #if PERFCTRS
493 1.175 christos if (PMC_ENABLED(l->l_proc)) {
494 1.175 christos pmc_restore_context(l->l_proc);
495 1.166 christos }
496 1.114 gmcgarry #endif
497 1.110 briggs
498 1.110 briggs /*
499 1.76 thorpej * We're running again; record our new start time. We might
500 1.174 ad * be running on a new CPU now, so don't use the cached
501 1.76 thorpej * schedstate_percpu pointer.
502 1.76 thorpej */
503 1.177.2.10 yamt SYSCALL_TIME_WAKEUP(l);
504 1.122 thorpej KDASSERT(l->l_cpu == curcpu());
505 1.169 yamt
506 1.177.2.1 yamt (void)splsched();
507 1.177.2.1 yamt splx(oldspl);
508 1.122 thorpej return retval;
509 1.26 cgd }
510 1.26 cgd
511 1.26 cgd /*
512 1.174 ad * Change process state to be runnable, placing it on the run queue if it is
513 1.174 ad * in memory, and awakening the swapper if it isn't in memory.
514 1.174 ad *
515 1.174 ad * Call with the process and LWP locked. Will return with the LWP unlocked.
516 1.26 cgd */
517 1.26 cgd void
518 1.122 thorpej setrunnable(struct lwp *l)
519 1.26 cgd {
520 1.122 thorpej struct proc *p = l->l_proc;
521 1.174 ad sigset_t *ss;
522 1.26 cgd
523 1.177.2.10 yamt KASSERT((l->l_flag & LW_IDLE) == 0);
524 1.177.2.10 yamt KASSERT(mutex_owned(&p->p_smutex));
525 1.177.2.10 yamt KASSERT(lwp_locked(l, NULL));
526 1.83 thorpej
527 1.122 thorpej switch (l->l_stat) {
528 1.122 thorpej case LSSTOP:
529 1.33 mycroft /*
530 1.33 mycroft * If we're being traced (possibly because someone attached us
531 1.33 mycroft * while we were stopped), check for a signal from the debugger.
532 1.33 mycroft */
533 1.174 ad if ((p->p_slflag & PSL_TRACED) != 0 && p->p_xstat != 0) {
534 1.174 ad if ((sigprop[p->p_xstat] & SA_TOLWP) != 0)
535 1.174 ad ss = &l->l_sigpend.sp_set;
536 1.174 ad else
537 1.174 ad ss = &p->p_sigpend.sp_set;
538 1.174 ad sigaddset(ss, p->p_xstat);
539 1.174 ad signotify(l);
540 1.53 mycroft }
541 1.174 ad p->p_nrlwps++;
542 1.26 cgd break;
543 1.174 ad case LSSUSPENDED:
544 1.177.2.10 yamt l->l_flag &= ~LW_WSUSPEND;
545 1.174 ad p->p_nrlwps++;
546 1.122 thorpej break;
547 1.174 ad case LSSLEEP:
548 1.174 ad KASSERT(l->l_wchan != NULL);
549 1.26 cgd break;
550 1.174 ad default:
551 1.174 ad panic("setrunnable: lwp %p state was %d", l, l->l_stat);
552 1.26 cgd }
553 1.139 cl
554 1.174 ad /*
555 1.174 ad * If the LWP was sleeping interruptably, then it's OK to start it
556 1.174 ad * again. If not, mark it as still sleeping.
557 1.174 ad */
558 1.174 ad if (l->l_wchan != NULL) {
559 1.174 ad l->l_stat = LSSLEEP;
560 1.177.2.10 yamt /* lwp_unsleep() will release the lock. */
561 1.177.2.10 yamt lwp_unsleep(l);
562 1.174 ad return;
563 1.174 ad }
564 1.139 cl
565 1.177.2.18 rmind LOCK_ASSERT(lwp_locked(l, l->l_cpu->ci_schedstate.spc_mutex));
566 1.122 thorpej
567 1.174 ad /*
568 1.174 ad * If the LWP is still on the CPU, mark it as LSONPROC. It may be
569 1.174 ad * about to call mi_switch(), in which case it will yield.
570 1.174 ad */
571 1.177.2.17 ad if ((l->l_flag & LW_RUNNING) != 0) {
572 1.174 ad l->l_stat = LSONPROC;
573 1.174 ad l->l_slptime = 0;
574 1.174 ad lwp_unlock(l);
575 1.174 ad return;
576 1.174 ad }
577 1.122 thorpej
578 1.174 ad /*
579 1.174 ad * Set the LWP runnable. If it's swapped out, we need to wake the swapper
580 1.174 ad * to bring it back in. Otherwise, enter it into a run queue.
581 1.174 ad */
582 1.177.2.5 rmind sched_setrunnable(l);
583 1.174 ad l->l_stat = LSRUN;
584 1.122 thorpej l->l_slptime = 0;
585 1.174 ad
586 1.177.2.10 yamt if (l->l_flag & LW_INMEM) {
587 1.177.2.16 rmind sched_enqueue(l, false);
588 1.177.2.10 yamt resched_cpu(l);
589 1.174 ad lwp_unlock(l);
590 1.174 ad } else {
591 1.174 ad lwp_unlock(l);
592 1.177 ad uvm_kick_scheduler();
593 1.174 ad }
594 1.26 cgd }
595 1.26 cgd
596 1.174 ad /*
597 1.174 ad * suspendsched:
598 1.174 ad *
599 1.174 ad * Convert all non-L_SYSTEM LSSLEEP or LSRUN LWPs to LSSUSPENDED.
600 1.174 ad */
601 1.94 bouyer void
602 1.174 ad suspendsched(void)
603 1.94 bouyer {
604 1.174 ad #ifdef MULTIPROCESSOR
605 1.174 ad CPU_INFO_ITERATOR cii;
606 1.174 ad struct cpu_info *ci;
607 1.174 ad #endif
608 1.122 thorpej struct lwp *l;
609 1.174 ad struct proc *p;
610 1.94 bouyer
611 1.94 bouyer /*
612 1.174 ad * We do this by process in order not to violate the locking rules.
613 1.94 bouyer */
614 1.174 ad mutex_enter(&proclist_mutex);
615 1.174 ad PROCLIST_FOREACH(p, &allproc) {
616 1.174 ad mutex_enter(&p->p_smutex);
617 1.174 ad
618 1.177.2.10 yamt if ((p->p_flag & PK_SYSTEM) != 0) {
619 1.174 ad mutex_exit(&p->p_smutex);
620 1.94 bouyer continue;
621 1.174 ad }
622 1.174 ad
623 1.174 ad p->p_stat = SSTOP;
624 1.174 ad
625 1.174 ad LIST_FOREACH(l, &p->p_lwps, l_sibling) {
626 1.174 ad if (l == curlwp)
627 1.174 ad continue;
628 1.174 ad
629 1.174 ad lwp_lock(l);
630 1.122 thorpej
631 1.97 enami /*
632 1.174 ad * Set L_WREBOOT so that the LWP will suspend itself
633 1.174 ad * when it tries to return to user mode. We want to
634 1.174 ad * try and get to get as many LWPs as possible to
635 1.174 ad * the user / kernel boundary, so that they will
636 1.174 ad * release any locks that they hold.
637 1.97 enami */
638 1.177.2.10 yamt l->l_flag |= (LW_WREBOOT | LW_WSUSPEND);
639 1.174 ad
640 1.174 ad if (l->l_stat == LSSLEEP &&
641 1.177.2.10 yamt (l->l_flag & LW_SINTR) != 0) {
642 1.174 ad /* setrunnable() will release the lock. */
643 1.174 ad setrunnable(l);
644 1.174 ad continue;
645 1.174 ad }
646 1.174 ad
647 1.174 ad lwp_unlock(l);
648 1.94 bouyer }
649 1.174 ad
650 1.174 ad mutex_exit(&p->p_smutex);
651 1.94 bouyer }
652 1.174 ad mutex_exit(&proclist_mutex);
653 1.174 ad
654 1.174 ad /*
655 1.174 ad * Kick all CPUs to make them preempt any LWPs running in user mode.
656 1.174 ad * They'll trap into the kernel and suspend themselves in userret().
657 1.174 ad */
658 1.174 ad #ifdef MULTIPROCESSOR
659 1.174 ad for (CPU_INFO_FOREACH(cii, ci))
660 1.177.2.9 yamt cpu_need_resched(ci, 0);
661 1.174 ad #else
662 1.177.2.9 yamt cpu_need_resched(curcpu(), 0);
663 1.174 ad #endif
664 1.94 bouyer }
665 1.113 gmcgarry
666 1.113 gmcgarry /*
667 1.174 ad * sched_kpri:
668 1.174 ad *
669 1.174 ad * Scale a priority level to a kernel priority level, usually
670 1.174 ad * for an LWP that is about to sleep.
671 1.174 ad */
672 1.177.2.10 yamt pri_t
673 1.174 ad sched_kpri(struct lwp *l)
674 1.174 ad {
675 1.174 ad /*
676 1.174 ad * Scale user priorities (127 -> 50) up to kernel priorities
677 1.174 ad * in the range (49 -> 8). Reserve the top 8 kernel priorities
678 1.174 ad * for high priority kthreads. Kernel priorities passed in
679 1.174 ad * are left "as is". XXX This is somewhat arbitrary.
680 1.174 ad */
681 1.174 ad static const uint8_t kpri_tab[] = {
682 1.174 ad 0, 1, 2, 3, 4, 5, 6, 7,
683 1.174 ad 8, 9, 10, 11, 12, 13, 14, 15,
684 1.174 ad 16, 17, 18, 19, 20, 21, 22, 23,
685 1.174 ad 24, 25, 26, 27, 28, 29, 30, 31,
686 1.174 ad 32, 33, 34, 35, 36, 37, 38, 39,
687 1.174 ad 40, 41, 42, 43, 44, 45, 46, 47,
688 1.174 ad 48, 49, 8, 8, 9, 9, 10, 10,
689 1.174 ad 11, 11, 12, 12, 13, 14, 14, 15,
690 1.174 ad 15, 16, 16, 17, 17, 18, 18, 19,
691 1.174 ad 20, 20, 21, 21, 22, 22, 23, 23,
692 1.174 ad 24, 24, 25, 26, 26, 27, 27, 28,
693 1.174 ad 28, 29, 29, 30, 30, 31, 32, 32,
694 1.174 ad 33, 33, 34, 34, 35, 35, 36, 36,
695 1.174 ad 37, 38, 38, 39, 39, 40, 40, 41,
696 1.174 ad 41, 42, 42, 43, 44, 44, 45, 45,
697 1.174 ad 46, 46, 47, 47, 48, 48, 49, 49,
698 1.174 ad };
699 1.174 ad
700 1.177.2.10 yamt return (pri_t)kpri_tab[l->l_usrpri];
701 1.174 ad }
702 1.174 ad
703 1.174 ad /*
704 1.174 ad * sched_unsleep:
705 1.174 ad *
706 1.174 ad * The is called when the LWP has not been awoken normally but instead
707 1.174 ad * interrupted: for example, if the sleep timed out. Because of this,
708 1.174 ad * it's not a valid action for running or idle LWPs.
709 1.174 ad */
710 1.177.2.10 yamt static void
711 1.174 ad sched_unsleep(struct lwp *l)
712 1.174 ad {
713 1.174 ad
714 1.174 ad lwp_unlock(l);
715 1.174 ad panic("sched_unsleep");
716 1.174 ad }
717 1.174 ad
718 1.177.2.5 rmind inline void
719 1.177.2.10 yamt resched_cpu(struct lwp *l)
720 1.174 ad {
721 1.177.2.5 rmind struct cpu_info *ci;
722 1.177.2.10 yamt const pri_t pri = lwp_eprio(l);
723 1.174 ad
724 1.177.2.5 rmind /*
725 1.177.2.5 rmind * XXXSMP
726 1.177.2.5 rmind * Since l->l_cpu persists across a context switch,
727 1.177.2.5 rmind * this gives us *very weak* processor affinity, in
728 1.177.2.5 rmind * that we notify the CPU on which the process last
729 1.177.2.5 rmind * ran that it should try to switch.
730 1.177.2.5 rmind *
731 1.177.2.5 rmind * This does not guarantee that the process will run on
732 1.177.2.5 rmind * that processor next, because another processor might
733 1.177.2.5 rmind * grab it the next time it performs a context switch.
734 1.177.2.5 rmind *
735 1.177.2.5 rmind * This also does not handle the case where its last
736 1.177.2.5 rmind * CPU is running a higher-priority process, but every
737 1.177.2.5 rmind * other CPU is running a lower-priority process. There
738 1.177.2.5 rmind * are ways to handle this situation, but they're not
739 1.177.2.5 rmind * currently very pretty, and we also need to weigh the
740 1.177.2.5 rmind * cost of moving a process from one CPU to another.
741 1.177.2.5 rmind */
742 1.177.2.5 rmind ci = (l->l_cpu != NULL) ? l->l_cpu : curcpu();
743 1.177.2.5 rmind if (pri < ci->ci_schedstate.spc_curpriority)
744 1.177.2.9 yamt cpu_need_resched(ci, 0);
745 1.177.2.1 yamt }
746 1.177.2.10 yamt
747 1.177.2.10 yamt static void
748 1.177.2.10 yamt sched_changepri(struct lwp *l, pri_t pri)
749 1.177.2.10 yamt {
750 1.177.2.10 yamt
751 1.177.2.18 rmind LOCK_ASSERT(lwp_locked(l, l->l_cpu->ci_schedstate.spc_mutex));
752 1.177.2.10 yamt
753 1.177.2.10 yamt l->l_usrpri = pri;
754 1.177.2.10 yamt if (l->l_priority < PUSER)
755 1.177.2.10 yamt return;
756 1.177.2.10 yamt
757 1.177.2.10 yamt if (l->l_stat != LSRUN || (l->l_flag & LW_INMEM) == 0) {
758 1.177.2.10 yamt l->l_priority = pri;
759 1.177.2.10 yamt return;
760 1.177.2.10 yamt }
761 1.177.2.10 yamt
762 1.177.2.11 yamt sched_dequeue(l);
763 1.177.2.10 yamt l->l_priority = pri;
764 1.177.2.16 rmind sched_enqueue(l, false);
765 1.177.2.10 yamt resched_cpu(l);
766 1.177.2.10 yamt }
767 1.177.2.10 yamt
768 1.177.2.10 yamt static void
769 1.177.2.11 yamt sched_lendpri(struct lwp *l, pri_t pri)
770 1.177.2.10 yamt {
771 1.177.2.10 yamt
772 1.177.2.18 rmind LOCK_ASSERT(lwp_locked(l, l->l_cpu->ci_schedstate.spc_mutex));
773 1.177.2.10 yamt
774 1.177.2.10 yamt if (l->l_stat != LSRUN || (l->l_flag & LW_INMEM) == 0) {
775 1.177.2.10 yamt l->l_inheritedprio = pri;
776 1.177.2.10 yamt return;
777 1.177.2.10 yamt }
778 1.177.2.10 yamt
779 1.177.2.11 yamt sched_dequeue(l);
780 1.177.2.10 yamt l->l_inheritedprio = pri;
781 1.177.2.16 rmind sched_enqueue(l, false);
782 1.177.2.10 yamt resched_cpu(l);
783 1.177.2.10 yamt }
784 1.177.2.10 yamt
785 1.177.2.10 yamt struct lwp *
786 1.177.2.10 yamt syncobj_noowner(wchan_t wchan)
787 1.177.2.10 yamt {
788 1.177.2.10 yamt
789 1.177.2.10 yamt return NULL;
790 1.177.2.10 yamt }
791