source: mainline/kernel/generic/src/proc/scheduler.c@ b51cf2c

lfn serial ticket/834-toolchain-update topic/msim-upgrade topic/simplify-dev-export
Last change on this file since b51cf2c was 0c2d9bb, checked in by Martin Decky <martin@…>, 12 years ago

merge mainline changes
  • Property mode set to 100644
File size: 17.5 KB
Line 
1/*
2 * Copyright (c) 2010 Jakub Jermar
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 *
9 * - Redistributions of source code must retain the above copyright
10 * notice, this list of conditions and the following disclaimer.
11 * - 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 * - The name of the author may not be used to endorse or promote products
15 * derived from this software without specific prior written permission.
16 *
17 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
18 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
19 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
20 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
21 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
22 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
23 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
24 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
25 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
26 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
27 */
28
29/** @addtogroup genericproc
30 * @{
31 */
32
33/**
34 * @file
35 * @brief Scheduler and load balancing.
36 *
37 * This file contains the scheduler and kcpulb kernel thread which
38 * performs load-balancing of per-CPU run queues.
39 */
40
41#include <proc/scheduler.h>
42#include <proc/thread.h>
43#include <proc/task.h>
44#include <mm/frame.h>
45#include <mm/page.h>
46#include <mm/as.h>
47#include <time/timeout.h>
48#include <time/delay.h>
49#include <arch/asm.h>
50#include <arch/faddr.h>
51#include <arch/cycle.h>
52#include <atomic.h>
53#include <synch/spinlock.h>
54#include <config.h>
55#include <context.h>
56#include <fpu_context.h>
57#include <func.h>
58#include <arch.h>
59#include <adt/list.h>
60#include <panic.h>
61#include <cpu.h>
62#include <print.h>
63#include <debug.h>
64#include <stacktrace.h>
65
66//#define SCHEDULER_VERBOSE 1
67
68static void scheduler_separated_stack(void);
69
70atomic_t nrdy; /**< Number of ready threads in the system. */
71
72/** Carry out actions before new task runs. */
73static void before_task_runs(void)
74{
75 before_task_runs_arch();
76}
77
78/** Take actions before new thread runs.
79 *
80 * Perform actions that need to be
81 * taken before the newly selected
82 * thread is passed control.
83 *
84 * THREAD->lock is locked on entry
85 *
86 */
87static void before_thread_runs(void)
88{
89 before_thread_runs_arch();
90
91#ifdef CONFIG_FPU_LAZY
92 if (THREAD == CPU->fpu_owner)
93 fpu_enable();
94 else
95 fpu_disable();
96#elif defined CONFIG_FPU
97 fpu_enable();
98 if (THREAD->fpu_context_exists)
99 fpu_context_restore(THREAD->saved_fpu_context);
100 else {
101 fpu_init();
102 THREAD->fpu_context_exists = true;
103 }
104#endif
105
106#ifdef CONFIG_UDEBUG
107 if (THREAD->btrace) {
108 istate_t *istate = THREAD->udebug.uspace_state;
109 if (istate != NULL) {
110 printf("Thread %" PRIu64 " stack trace:\n", THREAD->tid);
111 stack_trace_istate(istate);
112 }
113
114 THREAD->btrace = false;
115 }
116#endif
117}
118
119/** Take actions after THREAD had run.
120 *
121 * Perform actions that need to be
122 * taken after the running thread
123 * had been preempted by the scheduler.
124 *
125 * THREAD->lock is locked on entry
126 *
127 */
128static void after_thread_ran(void)
129{
130 after_thread_ran_arch();
131}
132
133#ifdef CONFIG_FPU_LAZY
134void scheduler_fpu_lazy_request(void)
135{
136restart:
137 fpu_enable();
138 irq_spinlock_lock(&CPU->lock, false);
139
140 /* Save old context */
141 if (CPU->fpu_owner != NULL) {
142 irq_spinlock_lock(&CPU->fpu_owner->lock, false);
143 fpu_context_save(CPU->fpu_owner->saved_fpu_context);
144
145 /* Don't prevent migration */
146 CPU->fpu_owner->fpu_context_engaged = false;
147 irq_spinlock_unlock(&CPU->fpu_owner->lock, false);
148 CPU->fpu_owner = NULL;
149 }
150
151 irq_spinlock_lock(&THREAD->lock, false);
152 if (THREAD->fpu_context_exists) {
153 fpu_context_restore(THREAD->saved_fpu_context);
154 } else {
155 /* Allocate FPU context */
156 if (!THREAD->saved_fpu_context) {
157 /* Might sleep */
158 irq_spinlock_unlock(&THREAD->lock, false);
159 irq_spinlock_unlock(&CPU->lock, false);
160 THREAD->saved_fpu_context =
161 (fpu_context_t *) slab_alloc(fpu_context_slab, 0);
162
163 /* We may have switched CPUs during slab_alloc */
164 goto restart;
165 }
166 fpu_init();
167 THREAD->fpu_context_exists = true;
168 }
169
170 CPU->fpu_owner = THREAD;
171 THREAD->fpu_context_engaged = true;
172 irq_spinlock_unlock(&THREAD->lock, false);
173
174 irq_spinlock_unlock(&CPU->lock, false);
175}
176#endif /* CONFIG_FPU_LAZY */
177
178/** Initialize scheduler
179 *
180 * Initialize kernel scheduler.
181 *
182 */
183void scheduler_init(void)
184{
185}
186
187/** Get thread to be scheduled
188 *
189 * Get the optimal thread to be scheduled
190 * according to thread accounting and scheduler
191 * policy.
192 *
193 * @return Thread to be scheduled.
194 *
195 */
196static thread_t *find_best_thread(void)
197{
198 ASSERT(CPU != NULL);
199
200loop:
201
202 if (atomic_get(&CPU->nrdy) == 0) {
203 /*
204 * For there was nothing to run, the CPU goes to sleep
205 * until a hardware interrupt or an IPI comes.
206 * This improves energy saving and hyperthreading.
207 */
208 irq_spinlock_lock(&CPU->lock, false);
209 CPU->idle = true;
210 irq_spinlock_unlock(&CPU->lock, false);
211 interrupts_enable();
212
213 /*
214 * An interrupt might occur right now and wake up a thread.
215 * In such case, the CPU will continue to go to sleep
216 * even though there is a runnable thread.
217 */
218 cpu_sleep();
219 interrupts_disable();
220 goto loop;
221 }
222
223 unsigned int i;
224 for (i = 0; i < RQ_COUNT; i++) {
225 irq_spinlock_lock(&(CPU->rq[i].lock), false);
226 if (CPU->rq[i].n == 0) {
227 /*
228 * If this queue is empty, try a lower-priority queue.
229 */
230 irq_spinlock_unlock(&(CPU->rq[i].lock), false);
231 continue;
232 }
233
234 atomic_dec(&CPU->nrdy);
235 atomic_dec(&nrdy);
236 CPU->rq[i].n--;
237
238 /*
239 * Take the first thread from the queue.
240 */
241 thread_t *thread = list_get_instance(
242 list_first(&CPU->rq[i].rq), thread_t, rq_link);
243 list_remove(&thread->rq_link);
244
245 irq_spinlock_pass(&(CPU->rq[i].lock), &thread->lock);
246
247 thread->cpu = CPU;
248 thread->ticks = us2ticks((i + 1) * 10000);
249 thread->priority = i; /* Correct rq index */
250
251 /*
252 * Clear the stolen flag so that it can be migrated
253 * when load balancing needs emerge.
254 */
255 thread->stolen = false;
256 irq_spinlock_unlock(&thread->lock, false);
257
258 return thread;
259 }
260
261 goto loop;
262}
263
264/** Prevent rq starvation
265 *
266 * Prevent low priority threads from starving in rq's.
267 *
268 * When the function decides to relink rq's, it reconnects
269 * respective pointers so that in result threads with 'pri'
270 * greater or equal start are moved to a higher-priority queue.
271 *
272 * @param start Threshold priority.
273 *
274 */
275static void relink_rq(int start)
276{
277 list_t list;
278
279 list_initialize(&list);
280 irq_spinlock_lock(&CPU->lock, false);
281
282 if (CPU->needs_relink > NEEDS_RELINK_MAX) {
283 int i;
284 for (i = start; i < RQ_COUNT - 1; i++) {
285 /* Remember and empty rq[i + 1] */
286
287 irq_spinlock_lock(&CPU->rq[i + 1].lock, false);
288 list_concat(&list, &CPU->rq[i + 1].rq);
289 size_t n = CPU->rq[i + 1].n;
290 CPU->rq[i + 1].n = 0;
291 irq_spinlock_unlock(&CPU->rq[i + 1].lock, false);
292
293 /* Append rq[i + 1] to rq[i] */
294
295 irq_spinlock_lock(&CPU->rq[i].lock, false);
296 list_concat(&CPU->rq[i].rq, &list);
297 CPU->rq[i].n += n;
298 irq_spinlock_unlock(&CPU->rq[i].lock, false);
299 }
300
301 CPU->needs_relink = 0;
302 }
303
304 irq_spinlock_unlock(&CPU->lock, false);
305}
306
307/** The scheduler
308 *
309 * The thread scheduling procedure.
310 * Passes control directly to
311 * scheduler_separated_stack().
312 *
313 */
314void scheduler(void)
315{
316 volatile ipl_t ipl;
317
318 ASSERT(CPU != NULL);
319
320 ipl = interrupts_disable();
321
322 if (atomic_get(&haltstate))
323 halt();
324
325 if (THREAD) {
326 irq_spinlock_lock(&THREAD->lock, false);
327
328 /* Update thread kernel accounting */
329 THREAD->kcycles += get_cycle() - THREAD->last_cycle;
330
331#if (defined CONFIG_FPU) && (!defined CONFIG_FPU_LAZY)
332 fpu_context_save(THREAD->saved_fpu_context);
333#endif
334 if (!context_save(&THREAD->saved_context)) {
335 /*
336 * This is the place where threads leave scheduler();
337 */
338
339 /* Save current CPU cycle */
340 THREAD->last_cycle = get_cycle();
341
342 irq_spinlock_unlock(&THREAD->lock, false);
343 interrupts_restore(THREAD->saved_context.ipl);
344
345 return;
346 }
347
348 /*
349 * Interrupt priority level of preempted thread is recorded
350 * here to facilitate scheduler() invocations from
351 * interrupts_disable()'d code (e.g. waitq_sleep_timeout()).
352 *
353 */
354 THREAD->saved_context.ipl = ipl;
355 }
356
357 /*
358 * Through the 'THE' structure, we keep track of THREAD, TASK, CPU, AS
359 * and preemption counter. At this point THE could be coming either
360 * from THREAD's or CPU's stack.
361 *
362 */
363 the_copy(THE, (the_t *) CPU->stack);
364
365 /*
366 * We may not keep the old stack.
367 * Reason: If we kept the old stack and got blocked, for instance, in
368 * find_best_thread(), the old thread could get rescheduled by another
369 * CPU and overwrite the part of its own stack that was also used by
370 * the scheduler on this CPU.
371 *
372 * Moreover, we have to bypass the compiler-generated POP sequence
373 * which is fooled by SP being set to the very top of the stack.
374 * Therefore the scheduler() function continues in
375 * scheduler_separated_stack().
376 *
377 */
378 context_save(&CPU->saved_context);
379 context_set(&CPU->saved_context, FADDR(scheduler_separated_stack),
380 (uintptr_t) CPU->stack, STACK_SIZE);
381 context_restore(&CPU->saved_context);
382
383 /* Not reached */
384}
385
386/** Scheduler stack switch wrapper
387 *
388 * Second part of the scheduler() function
389 * using new stack. Handling the actual context
390 * switch to a new thread.
391 *
392 */
393void scheduler_separated_stack(void)
394{
395 DEADLOCK_PROBE_INIT(p_joinwq);
396 task_t *old_task = TASK;
397 as_t *old_as = AS;
398
399 ASSERT((!THREAD) || (irq_spinlock_locked(&THREAD->lock)));
400 ASSERT(CPU != NULL);
401
402 /*
403 * Hold the current task and the address space to prevent their
404 * possible destruction should thread_destroy() be called on this or any
405 * other processor while the scheduler is still using them.
406 */
407 if (old_task)
408 task_hold(old_task);
409
410 if (old_as)
411 as_hold(old_as);
412
413 if (THREAD) {
414 /* Must be run after the switch to scheduler stack */
415 after_thread_ran();
416
417 switch (THREAD->state) {
418 case Running:
419 irq_spinlock_unlock(&THREAD->lock, false);
420 thread_ready(THREAD);
421 break;
422
423 case Exiting:
424repeat:
425 if (THREAD->detached) {
426 thread_destroy(THREAD, false);
427 } else {
428 /*
429 * The thread structure is kept allocated until
430 * somebody calls thread_detach() on it.
431 */
432 if (!irq_spinlock_trylock(&THREAD->join_wq.lock)) {
433 /*
434 * Avoid deadlock.
435 */
436 irq_spinlock_unlock(&THREAD->lock, false);
437 delay(HZ);
438 irq_spinlock_lock(&THREAD->lock, false);
439 DEADLOCK_PROBE(p_joinwq,
440 DEADLOCK_THRESHOLD);
441 goto repeat;
442 }
443 _waitq_wakeup_unsafe(&THREAD->join_wq,
444 WAKEUP_FIRST);
445 irq_spinlock_unlock(&THREAD->join_wq.lock, false);
446
447 THREAD->state = Lingering;
448 irq_spinlock_unlock(&THREAD->lock, false);
449 }
450 break;
451
452 case Sleeping:
453 /*
454 * Prefer the thread after it's woken up.
455 */
456 THREAD->priority = -1;
457
458 /*
459 * We need to release wq->lock which we locked in
460 * waitq_sleep(). Address of wq->lock is kept in
461 * THREAD->sleep_queue.
462 */
463 irq_spinlock_unlock(&THREAD->sleep_queue->lock, false);
464
465 irq_spinlock_unlock(&THREAD->lock, false);
466 break;
467
468 default:
469 /*
470 * Entering state is unexpected.
471 */
472 panic("tid%" PRIu64 ": unexpected state %s.",
473 THREAD->tid, thread_states[THREAD->state]);
474 break;
475 }
476
477 THREAD = NULL;
478 }
479
480 THREAD = find_best_thread();
481
482 irq_spinlock_lock(&THREAD->lock, false);
483 int priority = THREAD->priority;
484 irq_spinlock_unlock(&THREAD->lock, false);
485
486 relink_rq(priority);
487
488 /*
489 * If both the old and the new task are the same,
490 * lots of work is avoided.
491 */
492 if (TASK != THREAD->task) {
493 as_t *new_as = THREAD->task->as;
494
495 /*
496 * Note that it is possible for two tasks
497 * to share one address space.
498 */
499 if (old_as != new_as) {
500 /*
501 * Both tasks and address spaces are different.
502 * Replace the old one with the new one.
503 */
504 as_switch(old_as, new_as);
505 }
506
507 TASK = THREAD->task;
508 before_task_runs();
509 }
510
511 if (old_task)
512 task_release(old_task);
513
514 if (old_as)
515 as_release(old_as);
516
517 irq_spinlock_lock(&THREAD->lock, false);
518 THREAD->state = Running;
519
520#ifdef SCHEDULER_VERBOSE
521 printf("cpu%u: tid %" PRIu64 " (priority=%d, ticks=%" PRIu64
522 ", nrdy=%d)\n", CPU->id, THREAD->tid, THREAD->priority,
523 THREAD->ticks, atomic_get(&CPU->nrdy));
524#endif
525
526 /*
527 * Some architectures provide late kernel PA2KA(identity)
528 * mapping in a page fault handler. However, the page fault
529 * handler uses the kernel stack of the running thread and
530 * therefore cannot be used to map it. The kernel stack, if
531 * necessary, is to be mapped in before_thread_runs(). This
532 * function must be executed before the switch to the new stack.
533 */
534 before_thread_runs();
535
536 /*
537 * Copy the knowledge of CPU, TASK, THREAD and preemption counter to
538 * thread's stack.
539 */
540 the_copy(THE, (the_t *) THREAD->kstack);
541
542 context_restore(&THREAD->saved_context);
543
544 /* Not reached */
545}
546
547#ifdef CONFIG_SMP
548/** Load balancing thread
549 *
550 * SMP load balancing thread, supervising thread supplies
551 * for the CPU it's wired to.
552 *
553 * @param arg Generic thread argument (unused).
554 *
555 */
556void kcpulb(void *arg)
557{
558 atomic_count_t average;
559 atomic_count_t rdy;
560
561 /*
562 * Detach kcpulb as nobody will call thread_join_timeout() on it.
563 */
564 thread_detach(THREAD);
565
566loop:
567 /*
568 * Work in 1s intervals.
569 */
570 thread_sleep(1);
571
572not_satisfied:
573 /*
574 * Calculate the number of threads that will be migrated/stolen from
575 * other CPU's. Note that situation can have changed between two
576 * passes. Each time get the most up to date counts.
577 *
578 */
579 average = atomic_get(&nrdy) / config.cpu_active + 1;
580 rdy = atomic_get(&CPU->nrdy);
581
582 if (average <= rdy)
583 goto satisfied;
584
585 atomic_count_t count = average - rdy;
586
587 /*
588 * Searching least priority queues on all CPU's first and most priority
589 * queues on all CPU's last.
590 */
591 size_t acpu;
592 size_t acpu_bias = 0;
593 int rq;
594
595 for (rq = RQ_COUNT - 1; rq >= 0; rq--) {
596 for (acpu = 0; acpu < config.cpu_active; acpu++) {
597 cpu_t *cpu = &cpus[(acpu + acpu_bias) % config.cpu_active];
598
599 /*
600 * Not interested in ourselves.
601 * Doesn't require interrupt disabling for kcpulb has
602 * THREAD_FLAG_WIRED.
603 *
604 */
605 if (CPU == cpu)
606 continue;
607
608 if (atomic_get(&cpu->nrdy) <= average)
609 continue;
610
611 irq_spinlock_lock(&(cpu->rq[rq].lock), true);
612 if (cpu->rq[rq].n == 0) {
613 irq_spinlock_unlock(&(cpu->rq[rq].lock), true);
614 continue;
615 }
616
617 thread_t *thread = NULL;
618
619 /* Search rq from the back */
620 link_t *link = cpu->rq[rq].rq.head.prev;
621
622 while (link != &(cpu->rq[rq].rq.head)) {
623 thread = (thread_t *) list_get_instance(link,
624 thread_t, rq_link);
625
626 /*
627 * Do not steal CPU-wired threads, threads
628 * already stolen, threads for which migration
629 * was temporarily disabled or threads whose
630 * FPU context is still in the CPU.
631 */
632 irq_spinlock_lock(&thread->lock, false);
633
634 if ((!thread->wired) && (!thread->stolen) &&
635 (!thread->nomigrate) &&
636 (!thread->fpu_context_engaged)) {
637 /*
638 * Remove thread from ready queue.
639 */
640 irq_spinlock_unlock(&thread->lock,
641 false);
642
643 atomic_dec(&cpu->nrdy);
644 atomic_dec(&nrdy);
645
646 cpu->rq[rq].n--;
647 list_remove(&thread->rq_link);
648
649 break;
650 }
651
652 irq_spinlock_unlock(&thread->lock, false);
653
654 link = link->prev;
655 thread = NULL;
656 }
657
658 if (thread) {
659 /*
660 * Ready thread on local CPU
661 */
662
663 irq_spinlock_pass(&(cpu->rq[rq].lock),
664 &thread->lock);
665
666#ifdef KCPULB_VERBOSE
667 printf("kcpulb%u: TID %" PRIu64 " -> cpu%u, "
668 "nrdy=%ld, avg=%ld\n", CPU->id, t->tid,
669 CPU->id, atomic_get(&CPU->nrdy),
670 atomic_get(&nrdy) / config.cpu_active);
671#endif
672
673 thread->stolen = true;
674 thread->state = Entering;
675
676 irq_spinlock_unlock(&thread->lock, true);
677 thread_ready(thread);
678
679 if (--count == 0)
680 goto satisfied;
681
682 /*
683 * We are not satisfied yet, focus on another
684 * CPU next time.
685 *
686 */
687 acpu_bias++;
688
689 continue;
690 } else
691 irq_spinlock_unlock(&(cpu->rq[rq].lock), true);
692
693 }
694 }
695
696 if (atomic_get(&CPU->nrdy)) {
697 /*
698 * Be a little bit light-weight and let migrated threads run.
699 *
700 */
701 scheduler();
702 } else {
703 /*
704 * We failed to migrate a single thread.
705 * Give up this turn.
706 *
707 */
708 goto loop;
709 }
710
711 goto not_satisfied;
712
713satisfied:
714 goto loop;
715}
716#endif /* CONFIG_SMP */
717
718/** Print information about threads & scheduler queues
719 *
720 */
721void sched_print_list(void)
722{
723 size_t cpu;
724 for (cpu = 0; cpu < config.cpu_count; cpu++) {
725 if (!cpus[cpu].active)
726 continue;
727
728 irq_spinlock_lock(&cpus[cpu].lock, true);
729
730 printf("cpu%u: address=%p, nrdy=%" PRIua ", needs_relink=%zu\n",
731 cpus[cpu].id, &cpus[cpu], atomic_get(&cpus[cpu].nrdy),
732 cpus[cpu].needs_relink);
733
734 unsigned int i;
735 for (i = 0; i < RQ_COUNT; i++) {
736 irq_spinlock_lock(&(cpus[cpu].rq[i].lock), false);
737 if (cpus[cpu].rq[i].n == 0) {
738 irq_spinlock_unlock(&(cpus[cpu].rq[i].lock), false);
739 continue;
740 }
741
742 printf("\trq[%u]: ", i);
743 list_foreach(cpus[cpu].rq[i].rq, rq_link, thread_t,
744 thread) {
745 printf("%" PRIu64 "(%s) ", thread->tid,
746 thread_states[thread->state]);
747 }
748 printf("\n");
749
750 irq_spinlock_unlock(&(cpus[cpu].rq[i].lock), false);
751 }
752
753 irq_spinlock_unlock(&cpus[cpu].lock, true);
754 }
755}
756
757/** @}
758 */
Note: See TracBrowser for help on using the repository browser.