fair.c 142.6 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
/*
 * Completely Fair Scheduling (CFS) Class (SCHED_NORMAL/SCHED_BATCH)
 *
 *  Copyright (C) 2007 Red Hat, Inc., Ingo Molnar <mingo@redhat.com>
 *
 *  Interactivity improvements by Mike Galbraith
 *  (C) 2007 Mike Galbraith <efault@gmx.de>
 *
 *  Various enhancements by Dmitry Adamushko.
 *  (C) 2007 Dmitry Adamushko <dmitry.adamushko@gmail.com>
 *
 *  Group scheduling enhancements by Srivatsa Vaddagiri
 *  Copyright IBM Corporation, 2007
 *  Author: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com>
 *
 *  Scaled math optimizations by Thomas Gleixner
 *  Copyright (C) 2007, Thomas Gleixner <tglx@linutronix.de>
18 19 20
 *
 *  Adaptive scheduling granularity, math enhancements by Peter Zijlstra
 *  Copyright (C) 2007 Red Hat, Inc., Peter Zijlstra <pzijlstr@redhat.com>
21 22
 */

A
Arjan van de Ven 已提交
23
#include <linux/latencytop.h>
24
#include <linux/sched.h>
25
#include <linux/cpumask.h>
26 27 28 29 30 31 32
#include <linux/slab.h>
#include <linux/profile.h>
#include <linux/interrupt.h>

#include <trace/events/sched.h>

#include "sched.h"
A
Arjan van de Ven 已提交
33

34
/*
35
 * Targeted preemption latency for CPU-bound tasks:
36
 * (default: 6ms * (1 + ilog(ncpus)), units: nanoseconds)
37
 *
38
 * NOTE: this latency value is not the same as the concept of
I
Ingo Molnar 已提交
39 40 41
 * 'timeslice length' - timeslices in CFS are of variable length
 * and have no persistent notion like in traditional, time-slice
 * based scheduling concepts.
42
 *
I
Ingo Molnar 已提交
43 44
 * (to see the precise effective timeslice length of your workload,
 *  run vmstat and monitor the context-switches (cs) field)
45
 */
46 47
unsigned int sysctl_sched_latency = 6000000ULL;
unsigned int normalized_sysctl_sched_latency = 6000000ULL;
48

49 50 51 52 53 54 55 56 57 58 59 60
/*
 * The initial- and re-scaling of tunables is configurable
 * (default SCHED_TUNABLESCALING_LOG = *(1+ilog(ncpus))
 *
 * Options are:
 * SCHED_TUNABLESCALING_NONE - unscaled, always *1
 * SCHED_TUNABLESCALING_LOG - scaled logarithmical, *1+ilog(ncpus)
 * SCHED_TUNABLESCALING_LINEAR - scaled linear, *ncpus
 */
enum sched_tunable_scaling sysctl_sched_tunable_scaling
	= SCHED_TUNABLESCALING_LOG;

61
/*
62
 * Minimal preemption granularity for CPU-bound tasks:
63
 * (default: 0.75 msec * (1 + ilog(ncpus)), units: nanoseconds)
64
 */
65 66
unsigned int sysctl_sched_min_granularity = 750000ULL;
unsigned int normalized_sysctl_sched_min_granularity = 750000ULL;
67 68

/*
69 70
 * is kept at sysctl_sched_latency / sysctl_sched_min_granularity
 */
71
static unsigned int sched_nr_latency = 8;
72 73

/*
74
 * After fork, child runs first. If set to 0 (default) then
75
 * parent will (try to) run first.
76
 */
77
unsigned int sysctl_sched_child_runs_first __read_mostly;
78 79 80

/*
 * SCHED_OTHER wake-up granularity.
81
 * (default: 1 msec * (1 + ilog(ncpus)), units: nanoseconds)
82 83 84 85 86
 *
 * This option delays the preemption effects of decoupled workloads
 * and reduces their over-scheduling. Synchronous workloads will still
 * have immediate wakeup/sleep latencies.
 */
87
unsigned int sysctl_sched_wakeup_granularity = 1000000UL;
88
unsigned int normalized_sysctl_sched_wakeup_granularity = 1000000UL;
89

90 91
const_debug unsigned int sysctl_sched_migration_cost = 500000UL;

92 93 94 95 96 97 98
/*
 * The exponential sliding  window over which load is averaged for shares
 * distribution.
 * (default: 10msec)
 */
unsigned int __read_mostly sysctl_sched_shares_window = 10000000UL;

99 100 101 102 103 104 105 106 107 108 109 110 111 112
#ifdef CONFIG_CFS_BANDWIDTH
/*
 * Amount of runtime to allocate from global (tg) to local (per-cfs_rq) pool
 * each time a cfs_rq requests quota.
 *
 * Note: in the case that the slice exceeds the runtime remaining (either due
 * to consumption or the quota being specified to be smaller than the slice)
 * we will always only issue the remaining available time.
 *
 * default: 5 msec, units: microseconds
  */
unsigned int sysctl_sched_cfs_bandwidth_slice = 5000UL;
#endif

113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216
/*
 * Increase the granularity value when there are more CPUs,
 * because with more CPUs the 'effective latency' as visible
 * to users decreases. But the relationship is not linear,
 * so pick a second-best guess by going with the log2 of the
 * number of CPUs.
 *
 * This idea comes from the SD scheduler of Con Kolivas:
 */
static int get_update_sysctl_factor(void)
{
	unsigned int cpus = min_t(int, num_online_cpus(), 8);
	unsigned int factor;

	switch (sysctl_sched_tunable_scaling) {
	case SCHED_TUNABLESCALING_NONE:
		factor = 1;
		break;
	case SCHED_TUNABLESCALING_LINEAR:
		factor = cpus;
		break;
	case SCHED_TUNABLESCALING_LOG:
	default:
		factor = 1 + ilog2(cpus);
		break;
	}

	return factor;
}

static void update_sysctl(void)
{
	unsigned int factor = get_update_sysctl_factor();

#define SET_SYSCTL(name) \
	(sysctl_##name = (factor) * normalized_sysctl_##name)
	SET_SYSCTL(sched_min_granularity);
	SET_SYSCTL(sched_latency);
	SET_SYSCTL(sched_wakeup_granularity);
#undef SET_SYSCTL
}

void sched_init_granularity(void)
{
	update_sysctl();
}

#if BITS_PER_LONG == 32
# define WMULT_CONST	(~0UL)
#else
# define WMULT_CONST	(1UL << 32)
#endif

#define WMULT_SHIFT	32

/*
 * Shift right and round:
 */
#define SRR(x, y) (((x) + (1UL << ((y) - 1))) >> (y))

/*
 * delta *= weight / lw
 */
static unsigned long
calc_delta_mine(unsigned long delta_exec, unsigned long weight,
		struct load_weight *lw)
{
	u64 tmp;

	/*
	 * weight can be less than 2^SCHED_LOAD_RESOLUTION for task group sched
	 * entities since MIN_SHARES = 2. Treat weight as 1 if less than
	 * 2^SCHED_LOAD_RESOLUTION.
	 */
	if (likely(weight > (1UL << SCHED_LOAD_RESOLUTION)))
		tmp = (u64)delta_exec * scale_load_down(weight);
	else
		tmp = (u64)delta_exec;

	if (!lw->inv_weight) {
		unsigned long w = scale_load_down(lw->weight);

		if (BITS_PER_LONG > 32 && unlikely(w >= WMULT_CONST))
			lw->inv_weight = 1;
		else if (unlikely(!w))
			lw->inv_weight = WMULT_CONST;
		else
			lw->inv_weight = WMULT_CONST / w;
	}

	/*
	 * Check whether we'd overflow the 64-bit multiplication:
	 */
	if (unlikely(tmp > WMULT_CONST))
		tmp = SRR(SRR(tmp, WMULT_SHIFT/2) * lw->inv_weight,
			WMULT_SHIFT/2);
	else
		tmp = SRR(tmp * lw->inv_weight, WMULT_SHIFT);

	return (unsigned long)min(tmp, (u64)(unsigned long)LONG_MAX);
}


const struct sched_class fair_sched_class;
217

218 219 220 221
/**************************************************************
 * CFS operations on generic schedulable entities:
 */

222
#ifdef CONFIG_FAIR_GROUP_SCHED
223

224
/* cpu runqueue to which this cfs_rq is attached */
225 226
static inline struct rq *rq_of(struct cfs_rq *cfs_rq)
{
227
	return cfs_rq->rq;
228 229
}

230 231
/* An entity is a task if it doesn't "own" a runqueue */
#define entity_is_task(se)	(!se->my_q)
232

233 234 235 236 237 238 239 240
static inline struct task_struct *task_of(struct sched_entity *se)
{
#ifdef CONFIG_SCHED_DEBUG
	WARN_ON_ONCE(!entity_is_task(se));
#endif
	return container_of(se, struct task_struct, se);
}

P
Peter Zijlstra 已提交
241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261
/* Walk up scheduling entities hierarchy */
#define for_each_sched_entity(se) \
		for (; se; se = se->parent)

static inline struct cfs_rq *task_cfs_rq(struct task_struct *p)
{
	return p->se.cfs_rq;
}

/* runqueue on which this entity is (to be) queued */
static inline struct cfs_rq *cfs_rq_of(struct sched_entity *se)
{
	return se->cfs_rq;
}

/* runqueue "owned" by this group */
static inline struct cfs_rq *group_cfs_rq(struct sched_entity *grp)
{
	return grp->my_q;
}

262 263 264
static inline void list_add_leaf_cfs_rq(struct cfs_rq *cfs_rq)
{
	if (!cfs_rq->on_list) {
265 266 267 268 269 270 271 272 273 274 275 276
		/*
		 * Ensure we either appear before our parent (if already
		 * enqueued) or force our parent to appear after us when it is
		 * enqueued.  The fact that we always enqueue bottom-up
		 * reduces this to two cases.
		 */
		if (cfs_rq->tg->parent &&
		    cfs_rq->tg->parent->cfs_rq[cpu_of(rq_of(cfs_rq))]->on_list) {
			list_add_rcu(&cfs_rq->leaf_cfs_rq_list,
				&rq_of(cfs_rq)->leaf_cfs_rq_list);
		} else {
			list_add_tail_rcu(&cfs_rq->leaf_cfs_rq_list,
277
				&rq_of(cfs_rq)->leaf_cfs_rq_list);
278
		}
279 280 281 282 283 284 285 286 287 288 289 290 291

		cfs_rq->on_list = 1;
	}
}

static inline void list_del_leaf_cfs_rq(struct cfs_rq *cfs_rq)
{
	if (cfs_rq->on_list) {
		list_del_rcu(&cfs_rq->leaf_cfs_rq_list);
		cfs_rq->on_list = 0;
	}
}

P
Peter Zijlstra 已提交
292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310
/* Iterate thr' all leaf cfs_rq's on a runqueue */
#define for_each_leaf_cfs_rq(rq, cfs_rq) \
	list_for_each_entry_rcu(cfs_rq, &rq->leaf_cfs_rq_list, leaf_cfs_rq_list)

/* Do the two (enqueued) entities belong to the same group ? */
static inline int
is_same_group(struct sched_entity *se, struct sched_entity *pse)
{
	if (se->cfs_rq == pse->cfs_rq)
		return 1;

	return 0;
}

static inline struct sched_entity *parent_entity(struct sched_entity *se)
{
	return se->parent;
}

311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353
/* return depth at which a sched entity is present in the hierarchy */
static inline int depth_se(struct sched_entity *se)
{
	int depth = 0;

	for_each_sched_entity(se)
		depth++;

	return depth;
}

static void
find_matching_se(struct sched_entity **se, struct sched_entity **pse)
{
	int se_depth, pse_depth;

	/*
	 * preemption test can be made between sibling entities who are in the
	 * same cfs_rq i.e who have a common parent. Walk up the hierarchy of
	 * both tasks until we find their ancestors who are siblings of common
	 * parent.
	 */

	/* First walk up until both entities are at same depth */
	se_depth = depth_se(*se);
	pse_depth = depth_se(*pse);

	while (se_depth > pse_depth) {
		se_depth--;
		*se = parent_entity(*se);
	}

	while (pse_depth > se_depth) {
		pse_depth--;
		*pse = parent_entity(*pse);
	}

	while (!is_same_group(*se, *pse)) {
		*se = parent_entity(*se);
		*pse = parent_entity(*pse);
	}
}

354 355 356 357 358 359
#else	/* !CONFIG_FAIR_GROUP_SCHED */

static inline struct task_struct *task_of(struct sched_entity *se)
{
	return container_of(se, struct task_struct, se);
}
360

361 362 363
static inline struct rq *rq_of(struct cfs_rq *cfs_rq)
{
	return container_of(cfs_rq, struct rq, cfs);
364 365 366 367
}

#define entity_is_task(se)	1

P
Peter Zijlstra 已提交
368 369
#define for_each_sched_entity(se) \
		for (; se; se = NULL)
370

P
Peter Zijlstra 已提交
371
static inline struct cfs_rq *task_cfs_rq(struct task_struct *p)
372
{
P
Peter Zijlstra 已提交
373
	return &task_rq(p)->cfs;
374 375
}

P
Peter Zijlstra 已提交
376 377 378 379 380 381 382 383 384 385 386 387 388 389
static inline struct cfs_rq *cfs_rq_of(struct sched_entity *se)
{
	struct task_struct *p = task_of(se);
	struct rq *rq = task_rq(p);

	return &rq->cfs;
}

/* runqueue "owned" by this group */
static inline struct cfs_rq *group_cfs_rq(struct sched_entity *grp)
{
	return NULL;
}

390 391 392 393 394 395 396 397
static inline void list_add_leaf_cfs_rq(struct cfs_rq *cfs_rq)
{
}

static inline void list_del_leaf_cfs_rq(struct cfs_rq *cfs_rq)
{
}

P
Peter Zijlstra 已提交
398 399 400 401 402 403 404 405 406 407 408 409 410 411
#define for_each_leaf_cfs_rq(rq, cfs_rq) \
		for (cfs_rq = &rq->cfs; cfs_rq; cfs_rq = NULL)

static inline int
is_same_group(struct sched_entity *se, struct sched_entity *pse)
{
	return 1;
}

static inline struct sched_entity *parent_entity(struct sched_entity *se)
{
	return NULL;
}

412 413 414 415 416
static inline void
find_matching_se(struct sched_entity **se, struct sched_entity **pse)
{
}

P
Peter Zijlstra 已提交
417 418
#endif	/* CONFIG_FAIR_GROUP_SCHED */

419 420
static void account_cfs_rq_runtime(struct cfs_rq *cfs_rq,
				   unsigned long delta_exec);
421 422 423 424 425

/**************************************************************
 * Scheduling class tree data structure manipulation methods:
 */

426
static inline u64 max_vruntime(u64 min_vruntime, u64 vruntime)
427
{
428 429
	s64 delta = (s64)(vruntime - min_vruntime);
	if (delta > 0)
430 431 432 433 434
		min_vruntime = vruntime;

	return min_vruntime;
}

435
static inline u64 min_vruntime(u64 min_vruntime, u64 vruntime)
P
Peter Zijlstra 已提交
436 437 438 439 440 441 442 443
{
	s64 delta = (s64)(vruntime - min_vruntime);
	if (delta < 0)
		min_vruntime = vruntime;

	return min_vruntime;
}

444 445 446 447 448 449
static inline int entity_before(struct sched_entity *a,
				struct sched_entity *b)
{
	return (s64)(a->vruntime - b->vruntime) < 0;
}

450 451 452 453 454 455 456 457 458 459 460 461
static void update_min_vruntime(struct cfs_rq *cfs_rq)
{
	u64 vruntime = cfs_rq->min_vruntime;

	if (cfs_rq->curr)
		vruntime = cfs_rq->curr->vruntime;

	if (cfs_rq->rb_leftmost) {
		struct sched_entity *se = rb_entry(cfs_rq->rb_leftmost,
						   struct sched_entity,
						   run_node);

P
Peter Zijlstra 已提交
462
		if (!cfs_rq->curr)
463 464 465 466 467 468
			vruntime = se->vruntime;
		else
			vruntime = min_vruntime(vruntime, se->vruntime);
	}

	cfs_rq->min_vruntime = max_vruntime(cfs_rq->min_vruntime, vruntime);
469 470 471 472
#ifndef CONFIG_64BIT
	smp_wmb();
	cfs_rq->min_vruntime_copy = cfs_rq->min_vruntime;
#endif
473 474
}

475 476 477
/*
 * Enqueue an entity into the rb-tree:
 */
478
static void __enqueue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se)
479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494
{
	struct rb_node **link = &cfs_rq->tasks_timeline.rb_node;
	struct rb_node *parent = NULL;
	struct sched_entity *entry;
	int leftmost = 1;

	/*
	 * Find the right place in the rbtree:
	 */
	while (*link) {
		parent = *link;
		entry = rb_entry(parent, struct sched_entity, run_node);
		/*
		 * We dont care about collisions. Nodes with
		 * the same key stay together.
		 */
495
		if (entity_before(se, entry)) {
496 497 498 499 500 501 502 503 504 505 506
			link = &parent->rb_left;
		} else {
			link = &parent->rb_right;
			leftmost = 0;
		}
	}

	/*
	 * Maintain a cache of leftmost tree entries (it is frequently
	 * used):
	 */
507
	if (leftmost)
I
Ingo Molnar 已提交
508
		cfs_rq->rb_leftmost = &se->run_node;
509 510 511 512 513

	rb_link_node(&se->run_node, parent, link);
	rb_insert_color(&se->run_node, &cfs_rq->tasks_timeline);
}

514
static void __dequeue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se)
515
{
P
Peter Zijlstra 已提交
516 517 518 519 520 521
	if (cfs_rq->rb_leftmost == &se->run_node) {
		struct rb_node *next_node;

		next_node = rb_next(&se->run_node);
		cfs_rq->rb_leftmost = next_node;
	}
I
Ingo Molnar 已提交
522

523 524 525
	rb_erase(&se->run_node, &cfs_rq->tasks_timeline);
}

526
struct sched_entity *__pick_first_entity(struct cfs_rq *cfs_rq)
527
{
528 529 530 531 532 533
	struct rb_node *left = cfs_rq->rb_leftmost;

	if (!left)
		return NULL;

	return rb_entry(left, struct sched_entity, run_node);
534 535
}

536 537 538 539 540 541 542 543 544 545 546
static struct sched_entity *__pick_next_entity(struct sched_entity *se)
{
	struct rb_node *next = rb_next(&se->run_node);

	if (!next)
		return NULL;

	return rb_entry(next, struct sched_entity, run_node);
}

#ifdef CONFIG_SCHED_DEBUG
547
struct sched_entity *__pick_last_entity(struct cfs_rq *cfs_rq)
548
{
549
	struct rb_node *last = rb_last(&cfs_rq->tasks_timeline);
550

551 552
	if (!last)
		return NULL;
553 554

	return rb_entry(last, struct sched_entity, run_node);
555 556
}

557 558 559 560
/**************************************************************
 * Scheduling class statistics methods:
 */

561
int sched_proc_update_handler(struct ctl_table *table, int write,
562
		void __user *buffer, size_t *lenp,
563 564
		loff_t *ppos)
{
565
	int ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
566
	int factor = get_update_sysctl_factor();
567 568 569 570 571 572 573

	if (ret || !write)
		return ret;

	sched_nr_latency = DIV_ROUND_UP(sysctl_sched_latency,
					sysctl_sched_min_granularity);

574 575 576 577 578 579 580
#define WRT_SYSCTL(name) \
	(normalized_sysctl_##name = sysctl_##name / (factor))
	WRT_SYSCTL(sched_min_granularity);
	WRT_SYSCTL(sched_latency);
	WRT_SYSCTL(sched_wakeup_granularity);
#undef WRT_SYSCTL

581 582 583
	return 0;
}
#endif
584

585
/*
586
 * delta /= w
587 588 589 590
 */
static inline unsigned long
calc_delta_fair(unsigned long delta, struct sched_entity *se)
{
591 592
	if (unlikely(se->load.weight != NICE_0_LOAD))
		delta = calc_delta_mine(delta, NICE_0_LOAD, &se->load);
593 594 595 596

	return delta;
}

597 598 599 600 601 602 603 604
/*
 * The idea is to set a period in which each task runs once.
 *
 * When there are too many tasks (sysctl_sched_nr_latency) we have to stretch
 * this period because otherwise the slices get too small.
 *
 * p = (nr <= nl) ? l : l*nr/nl
 */
605 606 607
static u64 __sched_period(unsigned long nr_running)
{
	u64 period = sysctl_sched_latency;
608
	unsigned long nr_latency = sched_nr_latency;
609 610

	if (unlikely(nr_running > nr_latency)) {
611
		period = sysctl_sched_min_granularity;
612 613 614 615 616 617
		period *= nr_running;
	}

	return period;
}

618 619 620 621
/*
 * We calculate the wall-time slice from the period by taking a part
 * proportional to the weight.
 *
622
 * s = p*P[w/rw]
623
 */
P
Peter Zijlstra 已提交
624
static u64 sched_slice(struct cfs_rq *cfs_rq, struct sched_entity *se)
625
{
M
Mike Galbraith 已提交
626
	u64 slice = __sched_period(cfs_rq->nr_running + !se->on_rq);
627

M
Mike Galbraith 已提交
628
	for_each_sched_entity(se) {
L
Lin Ming 已提交
629
		struct load_weight *load;
630
		struct load_weight lw;
L
Lin Ming 已提交
631 632 633

		cfs_rq = cfs_rq_of(se);
		load = &cfs_rq->load;
634

M
Mike Galbraith 已提交
635
		if (unlikely(!se->on_rq)) {
636
			lw = cfs_rq->load;
M
Mike Galbraith 已提交
637 638 639 640 641 642 643

			update_load_add(&lw, se->load.weight);
			load = &lw;
		}
		slice = calc_delta_mine(slice, se->load.weight, load);
	}
	return slice;
644 645
}

646
/*
647
 * We calculate the vruntime slice of a to be inserted task
648
 *
649
 * vs = s/w
650
 */
651
static u64 sched_vslice(struct cfs_rq *cfs_rq, struct sched_entity *se)
P
Peter Zijlstra 已提交
652
{
653
	return calc_delta_fair(sched_slice(cfs_rq, se), se);
654 655
}

656
static void update_cfs_load(struct cfs_rq *cfs_rq, int global_update);
657
static void update_cfs_shares(struct cfs_rq *cfs_rq);
658

659 660 661 662 663
/*
 * Update the current task's runtime statistics. Skip current tasks that
 * are not in our scheduling class.
 */
static inline void
I
Ingo Molnar 已提交
664 665
__update_curr(struct cfs_rq *cfs_rq, struct sched_entity *curr,
	      unsigned long delta_exec)
666
{
667
	unsigned long delta_exec_weighted;
668

669 670
	schedstat_set(curr->statistics.exec_max,
		      max((u64)delta_exec, curr->statistics.exec_max));
671 672

	curr->sum_exec_runtime += delta_exec;
673
	schedstat_add(cfs_rq, exec_clock, delta_exec);
674
	delta_exec_weighted = calc_delta_fair(delta_exec, curr);
675

I
Ingo Molnar 已提交
676
	curr->vruntime += delta_exec_weighted;
677
	update_min_vruntime(cfs_rq);
678

P
Peter Zijlstra 已提交
679
#if defined CONFIG_SMP && defined CONFIG_FAIR_GROUP_SCHED
680 681
	cfs_rq->load_unacc_exec_time += delta_exec;
#endif
682 683
}

684
static void update_curr(struct cfs_rq *cfs_rq)
685
{
686
	struct sched_entity *curr = cfs_rq->curr;
687
	u64 now = rq_of(cfs_rq)->clock_task;
688 689 690 691 692 693 694 695 696 697
	unsigned long delta_exec;

	if (unlikely(!curr))
		return;

	/*
	 * Get the amount of time the current task was running
	 * since the last time we changed load (this cannot
	 * overflow on 32 bits):
	 */
I
Ingo Molnar 已提交
698
	delta_exec = (unsigned long)(now - curr->exec_start);
P
Peter Zijlstra 已提交
699 700
	if (!delta_exec)
		return;
701

I
Ingo Molnar 已提交
702 703
	__update_curr(cfs_rq, curr, delta_exec);
	curr->exec_start = now;
704 705 706 707

	if (entity_is_task(curr)) {
		struct task_struct *curtask = task_of(curr);

708
		trace_sched_stat_runtime(curtask, delta_exec, curr->vruntime);
709
		cpuacct_charge(curtask, delta_exec);
710
		account_group_exec_runtime(curtask, delta_exec);
711
	}
712 713

	account_cfs_rq_runtime(cfs_rq, delta_exec);
714 715 716
}

static inline void
717
update_stats_wait_start(struct cfs_rq *cfs_rq, struct sched_entity *se)
718
{
719
	schedstat_set(se->statistics.wait_start, rq_of(cfs_rq)->clock);
720 721 722 723 724
}

/*
 * Task is being enqueued - update stats:
 */
725
static void update_stats_enqueue(struct cfs_rq *cfs_rq, struct sched_entity *se)
726 727 728 729 730
{
	/*
	 * Are we enqueueing a waiting task? (for current tasks
	 * a dequeue/enqueue event is a NOP)
	 */
731
	if (se != cfs_rq->curr)
732
		update_stats_wait_start(cfs_rq, se);
733 734 735
}

static void
736
update_stats_wait_end(struct cfs_rq *cfs_rq, struct sched_entity *se)
737
{
738 739 740 741 742
	schedstat_set(se->statistics.wait_max, max(se->statistics.wait_max,
			rq_of(cfs_rq)->clock - se->statistics.wait_start));
	schedstat_set(se->statistics.wait_count, se->statistics.wait_count + 1);
	schedstat_set(se->statistics.wait_sum, se->statistics.wait_sum +
			rq_of(cfs_rq)->clock - se->statistics.wait_start);
743 744 745
#ifdef CONFIG_SCHEDSTATS
	if (entity_is_task(se)) {
		trace_sched_stat_wait(task_of(se),
746
			rq_of(cfs_rq)->clock - se->statistics.wait_start);
747 748
	}
#endif
749
	schedstat_set(se->statistics.wait_start, 0);
750 751 752
}

static inline void
753
update_stats_dequeue(struct cfs_rq *cfs_rq, struct sched_entity *se)
754 755 756 757 758
{
	/*
	 * Mark the end of the wait period if dequeueing a
	 * waiting task:
	 */
759
	if (se != cfs_rq->curr)
760
		update_stats_wait_end(cfs_rq, se);
761 762 763 764 765 766
}

/*
 * We are picking a new current task - update its stats:
 */
static inline void
767
update_stats_curr_start(struct cfs_rq *cfs_rq, struct sched_entity *se)
768 769 770 771
{
	/*
	 * We are starting a new run period:
	 */
772
	se->exec_start = rq_of(cfs_rq)->clock_task;
773 774 775 776 777 778
}

/**************************************************
 * Scheduling class queueing methods:
 */

779 780 781 782 783 784 785 786 787 788 789 790 791
#if defined CONFIG_SMP && defined CONFIG_FAIR_GROUP_SCHED
static void
add_cfs_task_weight(struct cfs_rq *cfs_rq, unsigned long weight)
{
	cfs_rq->task_weight += weight;
}
#else
static inline void
add_cfs_task_weight(struct cfs_rq *cfs_rq, unsigned long weight)
{
}
#endif

792 793 794 795
static void
account_entity_enqueue(struct cfs_rq *cfs_rq, struct sched_entity *se)
{
	update_load_add(&cfs_rq->load, se->load.weight);
796
	if (!parent_entity(se))
797
		update_load_add(&rq_of(cfs_rq)->load, se->load.weight);
798
	if (entity_is_task(se)) {
799
		add_cfs_task_weight(cfs_rq, se->load.weight);
800 801
		list_add(&se->group_node, &cfs_rq->tasks);
	}
802 803 804 805 806 807 808
	cfs_rq->nr_running++;
}

static void
account_entity_dequeue(struct cfs_rq *cfs_rq, struct sched_entity *se)
{
	update_load_sub(&cfs_rq->load, se->load.weight);
809
	if (!parent_entity(se))
810
		update_load_sub(&rq_of(cfs_rq)->load, se->load.weight);
811
	if (entity_is_task(se)) {
812
		add_cfs_task_weight(cfs_rq, -se->load.weight);
813 814
		list_del_init(&se->group_node);
	}
815 816 817
	cfs_rq->nr_running--;
}

818
#ifdef CONFIG_FAIR_GROUP_SCHED
819 820
/* we need this in update_cfs_load and load-balance functions below */
static inline int throttled_hierarchy(struct cfs_rq *cfs_rq);
821
# ifdef CONFIG_SMP
822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837
static void update_cfs_rq_load_contribution(struct cfs_rq *cfs_rq,
					    int global_update)
{
	struct task_group *tg = cfs_rq->tg;
	long load_avg;

	load_avg = div64_u64(cfs_rq->load_avg, cfs_rq->load_period+1);
	load_avg -= cfs_rq->load_contribution;

	if (global_update || abs(load_avg) > cfs_rq->load_contribution / 8) {
		atomic_add(load_avg, &tg->load_weight);
		cfs_rq->load_contribution += load_avg;
	}
}

static void update_cfs_load(struct cfs_rq *cfs_rq, int global_update)
P
Peter Zijlstra 已提交
838
{
839
	u64 period = sysctl_sched_shares_window;
P
Peter Zijlstra 已提交
840
	u64 now, delta;
841
	unsigned long load = cfs_rq->load.weight;
P
Peter Zijlstra 已提交
842

843
	if (cfs_rq->tg == &root_task_group || throttled_hierarchy(cfs_rq))
P
Peter Zijlstra 已提交
844 845
		return;

846
	now = rq_of(cfs_rq)->clock_task;
P
Peter Zijlstra 已提交
847 848
	delta = now - cfs_rq->load_stamp;

849 850 851 852 853
	/* truncate load history at 4 idle periods */
	if (cfs_rq->load_stamp > cfs_rq->load_last &&
	    now - cfs_rq->load_last > 4 * period) {
		cfs_rq->load_period = 0;
		cfs_rq->load_avg = 0;
854
		delta = period - 1;
855 856
	}

P
Peter Zijlstra 已提交
857
	cfs_rq->load_stamp = now;
858
	cfs_rq->load_unacc_exec_time = 0;
P
Peter Zijlstra 已提交
859
	cfs_rq->load_period += delta;
860 861 862 863
	if (load) {
		cfs_rq->load_last = now;
		cfs_rq->load_avg += delta * load;
	}
P
Peter Zijlstra 已提交
864

865 866 867 868 869
	/* consider updating load contribution on each fold or truncate */
	if (global_update || cfs_rq->load_period > period
	    || !cfs_rq->load_period)
		update_cfs_rq_load_contribution(cfs_rq, global_update);

P
Peter Zijlstra 已提交
870 871 872 873 874 875 876 877 878 879
	while (cfs_rq->load_period > period) {
		/*
		 * Inline assembly required to prevent the compiler
		 * optimising this loop into a divmod call.
		 * See __iter_div_u64_rem() for another example of this.
		 */
		asm("" : "+rm" (cfs_rq->load_period));
		cfs_rq->load_period /= 2;
		cfs_rq->load_avg /= 2;
	}
880

881 882
	if (!cfs_rq->curr && !cfs_rq->nr_running && !cfs_rq->load_avg)
		list_del_leaf_cfs_rq(cfs_rq);
P
Peter Zijlstra 已提交
883 884
}

885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900
static inline long calc_tg_weight(struct task_group *tg, struct cfs_rq *cfs_rq)
{
	long tg_weight;

	/*
	 * Use this CPU's actual weight instead of the last load_contribution
	 * to gain a more accurate current total weight. See
	 * update_cfs_rq_load_contribution().
	 */
	tg_weight = atomic_read(&tg->load_weight);
	tg_weight -= cfs_rq->load_contribution;
	tg_weight += cfs_rq->load.weight;

	return tg_weight;
}

901
static long calc_cfs_shares(struct cfs_rq *cfs_rq, struct task_group *tg)
902
{
903
	long tg_weight, load, shares;
904

905
	tg_weight = calc_tg_weight(tg, cfs_rq);
906
	load = cfs_rq->load.weight;
907 908

	shares = (tg->shares * load);
909 910
	if (tg_weight)
		shares /= tg_weight;
911 912 913 914 915 916 917 918 919 920 921 922 923

	if (shares < MIN_SHARES)
		shares = MIN_SHARES;
	if (shares > tg->shares)
		shares = tg->shares;

	return shares;
}

static void update_entity_shares_tick(struct cfs_rq *cfs_rq)
{
	if (cfs_rq->load_unacc_exec_time > sysctl_sched_shares_window) {
		update_cfs_load(cfs_rq, 0);
924
		update_cfs_shares(cfs_rq);
925 926 927 928 929 930 931
	}
}
# else /* CONFIG_SMP */
static void update_cfs_load(struct cfs_rq *cfs_rq, int global_update)
{
}

932
static inline long calc_cfs_shares(struct cfs_rq *cfs_rq, struct task_group *tg)
933 934 935 936 937 938 939 940
{
	return tg->shares;
}

static inline void update_entity_shares_tick(struct cfs_rq *cfs_rq)
{
}
# endif /* CONFIG_SMP */
P
Peter Zijlstra 已提交
941 942 943
static void reweight_entity(struct cfs_rq *cfs_rq, struct sched_entity *se,
			    unsigned long weight)
{
944 945 946 947
	if (se->on_rq) {
		/* commit outstanding execution time */
		if (cfs_rq->curr == se)
			update_curr(cfs_rq);
P
Peter Zijlstra 已提交
948
		account_entity_dequeue(cfs_rq, se);
949
	}
P
Peter Zijlstra 已提交
950 951 952 953 954 955 956

	update_load_set(&se->load, weight);

	if (se->on_rq)
		account_entity_enqueue(cfs_rq, se);
}

957
static void update_cfs_shares(struct cfs_rq *cfs_rq)
P
Peter Zijlstra 已提交
958 959 960
{
	struct task_group *tg;
	struct sched_entity *se;
961
	long shares;
P
Peter Zijlstra 已提交
962 963 964

	tg = cfs_rq->tg;
	se = tg->se[cpu_of(rq_of(cfs_rq))];
965
	if (!se || throttled_hierarchy(cfs_rq))
P
Peter Zijlstra 已提交
966
		return;
967 968 969 970
#ifndef CONFIG_SMP
	if (likely(se->load.weight == tg->shares))
		return;
#endif
971
	shares = calc_cfs_shares(cfs_rq, tg);
P
Peter Zijlstra 已提交
972 973 974 975

	reweight_entity(cfs_rq_of(se), se, shares);
}
#else /* CONFIG_FAIR_GROUP_SCHED */
976
static void update_cfs_load(struct cfs_rq *cfs_rq, int global_update)
P
Peter Zijlstra 已提交
977 978 979
{
}

980
static inline void update_cfs_shares(struct cfs_rq *cfs_rq)
P
Peter Zijlstra 已提交
981 982
{
}
983 984 985 986

static inline void update_entity_shares_tick(struct cfs_rq *cfs_rq)
{
}
P
Peter Zijlstra 已提交
987 988
#endif /* CONFIG_FAIR_GROUP_SCHED */

989
static void enqueue_sleeper(struct cfs_rq *cfs_rq, struct sched_entity *se)
990 991
{
#ifdef CONFIG_SCHEDSTATS
992 993 994 995 996
	struct task_struct *tsk = NULL;

	if (entity_is_task(se))
		tsk = task_of(se);

997 998
	if (se->statistics.sleep_start) {
		u64 delta = rq_of(cfs_rq)->clock - se->statistics.sleep_start;
999 1000 1001 1002

		if ((s64)delta < 0)
			delta = 0;

1003 1004
		if (unlikely(delta > se->statistics.sleep_max))
			se->statistics.sleep_max = delta;
1005

1006 1007
		se->statistics.sleep_start = 0;
		se->statistics.sum_sleep_runtime += delta;
A
Arjan van de Ven 已提交
1008

1009
		if (tsk) {
1010
			account_scheduler_latency(tsk, delta >> 10, 1);
1011 1012
			trace_sched_stat_sleep(tsk, delta);
		}
1013
	}
1014 1015
	if (se->statistics.block_start) {
		u64 delta = rq_of(cfs_rq)->clock - se->statistics.block_start;
1016 1017 1018 1019

		if ((s64)delta < 0)
			delta = 0;

1020 1021
		if (unlikely(delta > se->statistics.block_max))
			se->statistics.block_max = delta;
1022

1023 1024
		se->statistics.block_start = 0;
		se->statistics.sum_sleep_runtime += delta;
I
Ingo Molnar 已提交
1025

1026
		if (tsk) {
1027
			if (tsk->in_iowait) {
1028 1029
				se->statistics.iowait_sum += delta;
				se->statistics.iowait_count++;
1030
				trace_sched_stat_iowait(tsk, delta);
1031 1032
			}

1033 1034
			trace_sched_stat_blocked(tsk, delta);

1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045
			/*
			 * Blocking time is in units of nanosecs, so shift by
			 * 20 to get a milliseconds-range estimation of the
			 * amount of time that the task spent sleeping:
			 */
			if (unlikely(prof_on == SLEEP_PROFILING)) {
				profile_hits(SLEEP_PROFILING,
						(void *)get_wchan(tsk),
						delta >> 20);
			}
			account_scheduler_latency(tsk, delta >> 10, 0);
I
Ingo Molnar 已提交
1046
		}
1047 1048 1049 1050
	}
#endif
}

P
Peter Zijlstra 已提交
1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063
static void check_spread(struct cfs_rq *cfs_rq, struct sched_entity *se)
{
#ifdef CONFIG_SCHED_DEBUG
	s64 d = se->vruntime - cfs_rq->min_vruntime;

	if (d < 0)
		d = -d;

	if (d > 3*sysctl_sched_latency)
		schedstat_inc(cfs_rq, nr_spread_over);
#endif
}

1064 1065 1066
static void
place_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int initial)
{
1067
	u64 vruntime = cfs_rq->min_vruntime;
P
Peter Zijlstra 已提交
1068

1069 1070 1071 1072 1073 1074
	/*
	 * The 'current' period is already promised to the current tasks,
	 * however the extra weight of the new task will slow them down a
	 * little, place the new task so that it fits in the slot that
	 * stays open at the end.
	 */
P
Peter Zijlstra 已提交
1075
	if (initial && sched_feat(START_DEBIT))
1076
		vruntime += sched_vslice(cfs_rq, se);
1077

1078
	/* sleeps up to a single latency don't count. */
1079
	if (!initial) {
1080
		unsigned long thresh = sysctl_sched_latency;
1081

1082 1083 1084 1085 1086 1087
		/*
		 * Halve their sleep time's effect, to allow
		 * for a gentler effect of sleepers:
		 */
		if (sched_feat(GENTLE_FAIR_SLEEPERS))
			thresh >>= 1;
1088

1089
		vruntime -= thresh;
1090 1091
	}

1092 1093 1094
	/* ensure we never gain time by being placed backwards. */
	vruntime = max_vruntime(se->vruntime, vruntime);

P
Peter Zijlstra 已提交
1095
	se->vruntime = vruntime;
1096 1097
}

1098 1099
static void check_enqueue_throttle(struct cfs_rq *cfs_rq);

1100
static void
1101
enqueue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int flags)
1102
{
1103 1104 1105 1106
	/*
	 * Update the normalized vruntime before updating min_vruntime
	 * through callig update_curr().
	 */
1107
	if (!(flags & ENQUEUE_WAKEUP) || (flags & ENQUEUE_WAKING))
1108 1109
		se->vruntime += cfs_rq->min_vruntime;

1110
	/*
1111
	 * Update run-time statistics of the 'current'.
1112
	 */
1113
	update_curr(cfs_rq);
1114
	update_cfs_load(cfs_rq, 0);
P
Peter Zijlstra 已提交
1115
	account_entity_enqueue(cfs_rq, se);
1116
	update_cfs_shares(cfs_rq);
1117

1118
	if (flags & ENQUEUE_WAKEUP) {
1119
		place_entity(cfs_rq, se, 0);
1120
		enqueue_sleeper(cfs_rq, se);
I
Ingo Molnar 已提交
1121
	}
1122

1123
	update_stats_enqueue(cfs_rq, se);
P
Peter Zijlstra 已提交
1124
	check_spread(cfs_rq, se);
1125 1126
	if (se != cfs_rq->curr)
		__enqueue_entity(cfs_rq, se);
P
Peter Zijlstra 已提交
1127
	se->on_rq = 1;
1128

1129
	if (cfs_rq->nr_running == 1) {
1130
		list_add_leaf_cfs_rq(cfs_rq);
1131 1132
		check_enqueue_throttle(cfs_rq);
	}
1133 1134
}

1135
static void __clear_buddies_last(struct sched_entity *se)
P
Peter Zijlstra 已提交
1136
{
1137 1138 1139 1140 1141 1142 1143 1144
	for_each_sched_entity(se) {
		struct cfs_rq *cfs_rq = cfs_rq_of(se);
		if (cfs_rq->last == se)
			cfs_rq->last = NULL;
		else
			break;
	}
}
P
Peter Zijlstra 已提交
1145

1146 1147 1148 1149 1150 1151 1152 1153 1154
static void __clear_buddies_next(struct sched_entity *se)
{
	for_each_sched_entity(se) {
		struct cfs_rq *cfs_rq = cfs_rq_of(se);
		if (cfs_rq->next == se)
			cfs_rq->next = NULL;
		else
			break;
	}
P
Peter Zijlstra 已提交
1155 1156
}

1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167
static void __clear_buddies_skip(struct sched_entity *se)
{
	for_each_sched_entity(se) {
		struct cfs_rq *cfs_rq = cfs_rq_of(se);
		if (cfs_rq->skip == se)
			cfs_rq->skip = NULL;
		else
			break;
	}
}

P
Peter Zijlstra 已提交
1168 1169
static void clear_buddies(struct cfs_rq *cfs_rq, struct sched_entity *se)
{
1170 1171 1172 1173 1174
	if (cfs_rq->last == se)
		__clear_buddies_last(se);

	if (cfs_rq->next == se)
		__clear_buddies_next(se);
1175 1176 1177

	if (cfs_rq->skip == se)
		__clear_buddies_skip(se);
P
Peter Zijlstra 已提交
1178 1179
}

1180 1181
static void return_cfs_rq_runtime(struct cfs_rq *cfs_rq);

1182
static void
1183
dequeue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int flags)
1184
{
1185 1186 1187 1188 1189
	/*
	 * Update run-time statistics of the 'current'.
	 */
	update_curr(cfs_rq);

1190
	update_stats_dequeue(cfs_rq, se);
1191
	if (flags & DEQUEUE_SLEEP) {
P
Peter Zijlstra 已提交
1192
#ifdef CONFIG_SCHEDSTATS
1193 1194 1195 1196
		if (entity_is_task(se)) {
			struct task_struct *tsk = task_of(se);

			if (tsk->state & TASK_INTERRUPTIBLE)
1197
				se->statistics.sleep_start = rq_of(cfs_rq)->clock;
1198
			if (tsk->state & TASK_UNINTERRUPTIBLE)
1199
				se->statistics.block_start = rq_of(cfs_rq)->clock;
1200
		}
1201
#endif
P
Peter Zijlstra 已提交
1202 1203
	}

P
Peter Zijlstra 已提交
1204
	clear_buddies(cfs_rq, se);
P
Peter Zijlstra 已提交
1205

1206
	if (se != cfs_rq->curr)
1207
		__dequeue_entity(cfs_rq, se);
P
Peter Zijlstra 已提交
1208
	se->on_rq = 0;
1209
	update_cfs_load(cfs_rq, 0);
1210
	account_entity_dequeue(cfs_rq, se);
1211 1212 1213 1214 1215 1216

	/*
	 * Normalize the entity after updating the min_vruntime because the
	 * update can refer to the ->curr item and we need to reflect this
	 * movement in our normalized position.
	 */
1217
	if (!(flags & DEQUEUE_SLEEP))
1218
		se->vruntime -= cfs_rq->min_vruntime;
1219

1220 1221 1222
	/* return excess runtime on last dequeue */
	return_cfs_rq_runtime(cfs_rq);

1223 1224
	update_min_vruntime(cfs_rq);
	update_cfs_shares(cfs_rq);
1225 1226 1227 1228 1229
}

/*
 * Preempt the current task with a newly woken task if needed:
 */
1230
static void
I
Ingo Molnar 已提交
1231
check_preempt_tick(struct cfs_rq *cfs_rq, struct sched_entity *curr)
1232
{
1233
	unsigned long ideal_runtime, delta_exec;
1234 1235
	struct sched_entity *se;
	s64 delta;
1236

P
Peter Zijlstra 已提交
1237
	ideal_runtime = sched_slice(cfs_rq, curr);
1238
	delta_exec = curr->sum_exec_runtime - curr->prev_sum_exec_runtime;
1239
	if (delta_exec > ideal_runtime) {
1240
		resched_task(rq_of(cfs_rq)->curr);
1241 1242 1243 1244 1245
		/*
		 * The current task ran long enough, ensure it doesn't get
		 * re-elected due to buddy favours.
		 */
		clear_buddies(cfs_rq, curr);
1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256
		return;
	}

	/*
	 * Ensure that a task that missed wakeup preemption by a
	 * narrow margin doesn't have to wait for a full slice.
	 * This also mitigates buddy induced latencies under load.
	 */
	if (delta_exec < sysctl_sched_min_granularity)
		return;

1257 1258
	se = __pick_first_entity(cfs_rq);
	delta = curr->vruntime - se->vruntime;
1259

1260 1261
	if (delta < 0)
		return;
1262

1263 1264
	if (delta > ideal_runtime)
		resched_task(rq_of(cfs_rq)->curr);
1265 1266
}

1267
static void
1268
set_next_entity(struct cfs_rq *cfs_rq, struct sched_entity *se)
1269
{
1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280
	/* 'current' is not kept within the tree. */
	if (se->on_rq) {
		/*
		 * Any task has to be enqueued before it get to execute on
		 * a CPU. So account for the time it spent waiting on the
		 * runqueue.
		 */
		update_stats_wait_end(cfs_rq, se);
		__dequeue_entity(cfs_rq, se);
	}

1281
	update_stats_curr_start(cfs_rq, se);
1282
	cfs_rq->curr = se;
I
Ingo Molnar 已提交
1283 1284 1285 1286 1287 1288
#ifdef CONFIG_SCHEDSTATS
	/*
	 * Track our maximum slice length, if the CPU's load is at
	 * least twice that of our own weight (i.e. dont track it
	 * when there are only lesser-weight tasks around):
	 */
1289
	if (rq_of(cfs_rq)->load.weight >= 2*se->load.weight) {
1290
		se->statistics.slice_max = max(se->statistics.slice_max,
I
Ingo Molnar 已提交
1291 1292 1293
			se->sum_exec_runtime - se->prev_sum_exec_runtime);
	}
#endif
1294
	se->prev_sum_exec_runtime = se->sum_exec_runtime;
1295 1296
}

1297 1298 1299
static int
wakeup_preempt_entity(struct sched_entity *curr, struct sched_entity *se);

1300 1301 1302 1303 1304 1305 1306
/*
 * Pick the next process, keeping these things in mind, in this order:
 * 1) keep things fair between processes/task groups
 * 2) pick the "next" process, since someone really wants that to run
 * 3) pick the "last" process, for cache locality
 * 4) do not run the "skip" process, if something else is available
 */
1307
static struct sched_entity *pick_next_entity(struct cfs_rq *cfs_rq)
1308
{
1309
	struct sched_entity *se = __pick_first_entity(cfs_rq);
1310
	struct sched_entity *left = se;
1311

1312 1313 1314 1315 1316 1317 1318 1319 1320
	/*
	 * Avoid running the skip buddy, if running something else can
	 * be done without getting too unfair.
	 */
	if (cfs_rq->skip == se) {
		struct sched_entity *second = __pick_next_entity(se);
		if (second && wakeup_preempt_entity(second, left) < 1)
			se = second;
	}
1321

1322 1323 1324 1325 1326 1327
	/*
	 * Prefer last buddy, try to return the CPU to a preempted task.
	 */
	if (cfs_rq->last && wakeup_preempt_entity(cfs_rq->last, left) < 1)
		se = cfs_rq->last;

1328 1329 1330 1331 1332 1333
	/*
	 * Someone really wants this to run. If it's not unfair, run it.
	 */
	if (cfs_rq->next && wakeup_preempt_entity(cfs_rq->next, left) < 1)
		se = cfs_rq->next;

1334
	clear_buddies(cfs_rq, se);
P
Peter Zijlstra 已提交
1335 1336

	return se;
1337 1338
}

1339 1340
static void check_cfs_rq_runtime(struct cfs_rq *cfs_rq);

1341
static void put_prev_entity(struct cfs_rq *cfs_rq, struct sched_entity *prev)
1342 1343 1344 1345 1346 1347
{
	/*
	 * If still on the runqueue then deactivate_task()
	 * was not called and update_curr() has to be done:
	 */
	if (prev->on_rq)
1348
		update_curr(cfs_rq);
1349

1350 1351 1352
	/* throttle cfs_rqs exceeding runtime */
	check_cfs_rq_runtime(cfs_rq);

P
Peter Zijlstra 已提交
1353
	check_spread(cfs_rq, prev);
1354
	if (prev->on_rq) {
1355
		update_stats_wait_start(cfs_rq, prev);
1356 1357 1358
		/* Put 'current' back into the tree. */
		__enqueue_entity(cfs_rq, prev);
	}
1359
	cfs_rq->curr = NULL;
1360 1361
}

P
Peter Zijlstra 已提交
1362 1363
static void
entity_tick(struct cfs_rq *cfs_rq, struct sched_entity *curr, int queued)
1364 1365
{
	/*
1366
	 * Update run-time statistics of the 'current'.
1367
	 */
1368
	update_curr(cfs_rq);
1369

1370 1371 1372 1373 1374
	/*
	 * Update share accounting for long-running entities.
	 */
	update_entity_shares_tick(cfs_rq);

P
Peter Zijlstra 已提交
1375 1376 1377 1378 1379
#ifdef CONFIG_SCHED_HRTICK
	/*
	 * queued ticks are scheduled to match the slice, so don't bother
	 * validating it and just reschedule.
	 */
1380 1381 1382 1383
	if (queued) {
		resched_task(rq_of(cfs_rq)->curr);
		return;
	}
P
Peter Zijlstra 已提交
1384 1385 1386 1387 1388 1389 1390 1391
	/*
	 * don't let the period tick interfere with the hrtick preemption
	 */
	if (!sched_feat(DOUBLE_TICK) &&
			hrtimer_active(&rq_of(cfs_rq)->hrtick_timer))
		return;
#endif

Y
Yong Zhang 已提交
1392
	if (cfs_rq->nr_running > 1)
I
Ingo Molnar 已提交
1393
		check_preempt_tick(cfs_rq, curr);
1394 1395
}

1396 1397 1398 1399 1400 1401

/**************************************************
 * CFS bandwidth control machinery
 */

#ifdef CONFIG_CFS_BANDWIDTH
1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427

#ifdef HAVE_JUMP_LABEL
static struct jump_label_key __cfs_bandwidth_used;

static inline bool cfs_bandwidth_used(void)
{
	return static_branch(&__cfs_bandwidth_used);
}

void account_cfs_bandwidth_used(int enabled, int was_enabled)
{
	/* only need to count groups transitioning between enabled/!enabled */
	if (enabled && !was_enabled)
		jump_label_inc(&__cfs_bandwidth_used);
	else if (!enabled && was_enabled)
		jump_label_dec(&__cfs_bandwidth_used);
}
#else /* HAVE_JUMP_LABEL */
static bool cfs_bandwidth_used(void)
{
	return true;
}

void account_cfs_bandwidth_used(int enabled, int was_enabled) {}
#endif /* HAVE_JUMP_LABEL */

1428 1429 1430 1431 1432 1433 1434 1435
/*
 * default period for cfs group bandwidth.
 * default: 0.1s, units: nanoseconds
 */
static inline u64 default_cfs_period(void)
{
	return 100000000ULL;
}
1436 1437 1438 1439 1440 1441

static inline u64 sched_cfs_bandwidth_slice(void)
{
	return (u64)sysctl_sched_cfs_bandwidth_slice * NSEC_PER_USEC;
}

P
Paul Turner 已提交
1442 1443 1444 1445 1446 1447 1448
/*
 * Replenish runtime according to assigned quota and update expiration time.
 * We use sched_clock_cpu directly instead of rq->clock to avoid adding
 * additional synchronization around rq->lock.
 *
 * requires cfs_b->lock
 */
1449
void __refill_cfs_bandwidth_runtime(struct cfs_bandwidth *cfs_b)
P
Paul Turner 已提交
1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460
{
	u64 now;

	if (cfs_b->quota == RUNTIME_INF)
		return;

	now = sched_clock_cpu(smp_processor_id());
	cfs_b->runtime = cfs_b->quota;
	cfs_b->runtime_expires = now + ktime_to_ns(cfs_b->period);
}

1461 1462 1463 1464 1465
static inline struct cfs_bandwidth *tg_cfs_bandwidth(struct task_group *tg)
{
	return &tg->cfs_bandwidth;
}

1466 1467
/* returns 0 on failure to allocate runtime */
static int assign_cfs_rq_runtime(struct cfs_rq *cfs_rq)
1468 1469 1470
{
	struct task_group *tg = cfs_rq->tg;
	struct cfs_bandwidth *cfs_b = tg_cfs_bandwidth(tg);
P
Paul Turner 已提交
1471
	u64 amount = 0, min_amount, expires;
1472 1473 1474 1475 1476 1477 1478

	/* note: this is a positive sum as runtime_remaining <= 0 */
	min_amount = sched_cfs_bandwidth_slice() - cfs_rq->runtime_remaining;

	raw_spin_lock(&cfs_b->lock);
	if (cfs_b->quota == RUNTIME_INF)
		amount = min_amount;
1479
	else {
P
Paul Turner 已提交
1480 1481 1482 1483 1484 1485 1486 1487
		/*
		 * If the bandwidth pool has become inactive, then at least one
		 * period must have elapsed since the last consumption.
		 * Refresh the global state and ensure bandwidth timer becomes
		 * active.
		 */
		if (!cfs_b->timer_active) {
			__refill_cfs_bandwidth_runtime(cfs_b);
1488
			__start_cfs_bandwidth(cfs_b);
P
Paul Turner 已提交
1489
		}
1490 1491 1492 1493 1494 1495

		if (cfs_b->runtime > 0) {
			amount = min(cfs_b->runtime, min_amount);
			cfs_b->runtime -= amount;
			cfs_b->idle = 0;
		}
1496
	}
P
Paul Turner 已提交
1497
	expires = cfs_b->runtime_expires;
1498 1499 1500
	raw_spin_unlock(&cfs_b->lock);

	cfs_rq->runtime_remaining += amount;
P
Paul Turner 已提交
1501 1502 1503 1504 1505 1506 1507
	/*
	 * we may have advanced our local expiration to account for allowed
	 * spread between our sched_clock and the one on which runtime was
	 * issued.
	 */
	if ((s64)(expires - cfs_rq->runtime_expires) > 0)
		cfs_rq->runtime_expires = expires;
1508 1509

	return cfs_rq->runtime_remaining > 0;
1510 1511
}

P
Paul Turner 已提交
1512 1513 1514 1515 1516
/*
 * Note: This depends on the synchronization provided by sched_clock and the
 * fact that rq->clock snapshots this value.
 */
static void expire_cfs_rq_runtime(struct cfs_rq *cfs_rq)
1517
{
P
Paul Turner 已提交
1518 1519 1520 1521 1522
	struct cfs_bandwidth *cfs_b = tg_cfs_bandwidth(cfs_rq->tg);
	struct rq *rq = rq_of(cfs_rq);

	/* if the deadline is ahead of our clock, nothing to do */
	if (likely((s64)(rq->clock - cfs_rq->runtime_expires) < 0))
1523 1524
		return;

P
Paul Turner 已提交
1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549
	if (cfs_rq->runtime_remaining < 0)
		return;

	/*
	 * If the local deadline has passed we have to consider the
	 * possibility that our sched_clock is 'fast' and the global deadline
	 * has not truly expired.
	 *
	 * Fortunately we can check determine whether this the case by checking
	 * whether the global deadline has advanced.
	 */

	if ((s64)(cfs_rq->runtime_expires - cfs_b->runtime_expires) >= 0) {
		/* extend local deadline, drift is bounded above by 2 ticks */
		cfs_rq->runtime_expires += TICK_NSEC;
	} else {
		/* global deadline is ahead, expiration has passed */
		cfs_rq->runtime_remaining = 0;
	}
}

static void __account_cfs_rq_runtime(struct cfs_rq *cfs_rq,
				     unsigned long delta_exec)
{
	/* dock delta_exec before expiring quota (as it could span periods) */
1550
	cfs_rq->runtime_remaining -= delta_exec;
P
Paul Turner 已提交
1551 1552 1553
	expire_cfs_rq_runtime(cfs_rq);

	if (likely(cfs_rq->runtime_remaining > 0))
1554 1555
		return;

1556 1557 1558 1559 1560 1561
	/*
	 * if we're unable to extend our runtime we resched so that the active
	 * hierarchy can be throttled
	 */
	if (!assign_cfs_rq_runtime(cfs_rq) && likely(cfs_rq->curr))
		resched_task(rq_of(cfs_rq)->curr);
1562 1563 1564 1565 1566
}

static __always_inline void account_cfs_rq_runtime(struct cfs_rq *cfs_rq,
						   unsigned long delta_exec)
{
1567
	if (!cfs_bandwidth_used() || !cfs_rq->runtime_enabled)
1568 1569 1570 1571 1572
		return;

	__account_cfs_rq_runtime(cfs_rq, delta_exec);
}

1573 1574
static inline int cfs_rq_throttled(struct cfs_rq *cfs_rq)
{
1575
	return cfs_bandwidth_used() && cfs_rq->throttled;
1576 1577
}

1578 1579 1580
/* check whether cfs_rq, or any parent, is throttled */
static inline int throttled_hierarchy(struct cfs_rq *cfs_rq)
{
1581
	return cfs_bandwidth_used() && cfs_rq->throttle_count;
1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636
}

/*
 * Ensure that neither of the group entities corresponding to src_cpu or
 * dest_cpu are members of a throttled hierarchy when performing group
 * load-balance operations.
 */
static inline int throttled_lb_pair(struct task_group *tg,
				    int src_cpu, int dest_cpu)
{
	struct cfs_rq *src_cfs_rq, *dest_cfs_rq;

	src_cfs_rq = tg->cfs_rq[src_cpu];
	dest_cfs_rq = tg->cfs_rq[dest_cpu];

	return throttled_hierarchy(src_cfs_rq) ||
	       throttled_hierarchy(dest_cfs_rq);
}

/* updated child weight may affect parent so we have to do this bottom up */
static int tg_unthrottle_up(struct task_group *tg, void *data)
{
	struct rq *rq = data;
	struct cfs_rq *cfs_rq = tg->cfs_rq[cpu_of(rq)];

	cfs_rq->throttle_count--;
#ifdef CONFIG_SMP
	if (!cfs_rq->throttle_count) {
		u64 delta = rq->clock_task - cfs_rq->load_stamp;

		/* leaving throttled state, advance shares averaging windows */
		cfs_rq->load_stamp += delta;
		cfs_rq->load_last += delta;

		/* update entity weight now that we are on_rq again */
		update_cfs_shares(cfs_rq);
	}
#endif

	return 0;
}

static int tg_throttle_down(struct task_group *tg, void *data)
{
	struct rq *rq = data;
	struct cfs_rq *cfs_rq = tg->cfs_rq[cpu_of(rq)];

	/* group is entering throttled state, record last load */
	if (!cfs_rq->throttle_count)
		update_cfs_load(cfs_rq, 0);
	cfs_rq->throttle_count++;

	return 0;
}

1637
static void throttle_cfs_rq(struct cfs_rq *cfs_rq)
1638 1639 1640 1641 1642 1643 1644 1645 1646
{
	struct rq *rq = rq_of(cfs_rq);
	struct cfs_bandwidth *cfs_b = tg_cfs_bandwidth(cfs_rq->tg);
	struct sched_entity *se;
	long task_delta, dequeue = 1;

	se = cfs_rq->tg->se[cpu_of(rq_of(cfs_rq))];

	/* account load preceding throttle */
1647 1648 1649
	rcu_read_lock();
	walk_tg_tree_from(cfs_rq->tg, tg_throttle_down, tg_nop, (void *)rq);
	rcu_read_unlock();
1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669

	task_delta = cfs_rq->h_nr_running;
	for_each_sched_entity(se) {
		struct cfs_rq *qcfs_rq = cfs_rq_of(se);
		/* throttled entity or throttle-on-deactivate */
		if (!se->on_rq)
			break;

		if (dequeue)
			dequeue_entity(qcfs_rq, se, DEQUEUE_SLEEP);
		qcfs_rq->h_nr_running -= task_delta;

		if (qcfs_rq->load.weight)
			dequeue = 0;
	}

	if (!se)
		rq->nr_running -= task_delta;

	cfs_rq->throttled = 1;
1670
	cfs_rq->throttled_timestamp = rq->clock;
1671 1672 1673 1674 1675
	raw_spin_lock(&cfs_b->lock);
	list_add_tail_rcu(&cfs_rq->throttled_list, &cfs_b->throttled_cfs_rq);
	raw_spin_unlock(&cfs_b->lock);
}

1676
void unthrottle_cfs_rq(struct cfs_rq *cfs_rq)
1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687
{
	struct rq *rq = rq_of(cfs_rq);
	struct cfs_bandwidth *cfs_b = tg_cfs_bandwidth(cfs_rq->tg);
	struct sched_entity *se;
	int enqueue = 1;
	long task_delta;

	se = cfs_rq->tg->se[cpu_of(rq_of(cfs_rq))];

	cfs_rq->throttled = 0;
	raw_spin_lock(&cfs_b->lock);
1688
	cfs_b->throttled_time += rq->clock - cfs_rq->throttled_timestamp;
1689 1690
	list_del_rcu(&cfs_rq->throttled_list);
	raw_spin_unlock(&cfs_b->lock);
1691
	cfs_rq->throttled_timestamp = 0;
1692

1693 1694 1695 1696
	update_rq_clock(rq);
	/* update hierarchical throttle state */
	walk_tg_tree_from(cfs_rq->tg, tg_nop, tg_unthrottle_up, (void *)rq);

1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759
	if (!cfs_rq->load.weight)
		return;

	task_delta = cfs_rq->h_nr_running;
	for_each_sched_entity(se) {
		if (se->on_rq)
			enqueue = 0;

		cfs_rq = cfs_rq_of(se);
		if (enqueue)
			enqueue_entity(cfs_rq, se, ENQUEUE_WAKEUP);
		cfs_rq->h_nr_running += task_delta;

		if (cfs_rq_throttled(cfs_rq))
			break;
	}

	if (!se)
		rq->nr_running += task_delta;

	/* determine whether we need to wake up potentially idle cpu */
	if (rq->curr == rq->idle && rq->cfs.nr_running)
		resched_task(rq->curr);
}

static u64 distribute_cfs_runtime(struct cfs_bandwidth *cfs_b,
		u64 remaining, u64 expires)
{
	struct cfs_rq *cfs_rq;
	u64 runtime = remaining;

	rcu_read_lock();
	list_for_each_entry_rcu(cfs_rq, &cfs_b->throttled_cfs_rq,
				throttled_list) {
		struct rq *rq = rq_of(cfs_rq);

		raw_spin_lock(&rq->lock);
		if (!cfs_rq_throttled(cfs_rq))
			goto next;

		runtime = -cfs_rq->runtime_remaining + 1;
		if (runtime > remaining)
			runtime = remaining;
		remaining -= runtime;

		cfs_rq->runtime_remaining += runtime;
		cfs_rq->runtime_expires = expires;

		/* we check whether we're throttled above */
		if (cfs_rq->runtime_remaining > 0)
			unthrottle_cfs_rq(cfs_rq);

next:
		raw_spin_unlock(&rq->lock);

		if (!remaining)
			break;
	}
	rcu_read_unlock();

	return remaining;
}

1760 1761 1762 1763 1764 1765 1766 1767
/*
 * Responsible for refilling a task_group's bandwidth and unthrottling its
 * cfs_rqs as appropriate. If there has been no activity within the last
 * period the timer is deactivated until scheduling resumes; cfs_b->idle is
 * used to track this state.
 */
static int do_sched_cfs_period_timer(struct cfs_bandwidth *cfs_b, int overrun)
{
1768 1769
	u64 runtime, runtime_expires;
	int idle = 1, throttled;
1770 1771 1772 1773 1774 1775

	raw_spin_lock(&cfs_b->lock);
	/* no need to continue the timer with no bandwidth constraint */
	if (cfs_b->quota == RUNTIME_INF)
		goto out_unlock;

1776 1777 1778
	throttled = !list_empty(&cfs_b->throttled_cfs_rq);
	/* idle depends on !throttled (for the case of a large deficit) */
	idle = cfs_b->idle && !throttled;
1779
	cfs_b->nr_periods += overrun;
1780

P
Paul Turner 已提交
1781 1782 1783 1784 1785 1786
	/* if we're going inactive then everything else can be deferred */
	if (idle)
		goto out_unlock;

	__refill_cfs_bandwidth_runtime(cfs_b);

1787 1788 1789 1790 1791 1792
	if (!throttled) {
		/* mark as potentially idle for the upcoming period */
		cfs_b->idle = 1;
		goto out_unlock;
	}

1793 1794 1795
	/* account preceding periods in which throttling occurred */
	cfs_b->nr_throttled += overrun;

1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819
	/*
	 * There are throttled entities so we must first use the new bandwidth
	 * to unthrottle them before making it generally available.  This
	 * ensures that all existing debts will be paid before a new cfs_rq is
	 * allowed to run.
	 */
	runtime = cfs_b->runtime;
	runtime_expires = cfs_b->runtime_expires;
	cfs_b->runtime = 0;

	/*
	 * This check is repeated as we are holding onto the new bandwidth
	 * while we unthrottle.  This can potentially race with an unthrottled
	 * group trying to acquire new bandwidth from the global pool.
	 */
	while (throttled && runtime > 0) {
		raw_spin_unlock(&cfs_b->lock);
		/* we can't nest cfs_b->lock while distributing bandwidth */
		runtime = distribute_cfs_runtime(cfs_b, runtime,
						 runtime_expires);
		raw_spin_lock(&cfs_b->lock);

		throttled = !list_empty(&cfs_b->throttled_cfs_rq);
	}
1820

1821 1822 1823 1824 1825 1826 1827 1828 1829
	/* return (any) remaining runtime */
	cfs_b->runtime = runtime;
	/*
	 * While we are ensured activity in the period following an
	 * unthrottle, this also covers the case in which the new bandwidth is
	 * insufficient to cover the existing bandwidth deficit.  (Forcing the
	 * timer to remain active while there are any throttled entities.)
	 */
	cfs_b->idle = 0;
1830 1831 1832 1833 1834 1835 1836
out_unlock:
	if (idle)
		cfs_b->timer_active = 0;
	raw_spin_unlock(&cfs_b->lock);

	return idle;
}
1837

1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901
/* a cfs_rq won't donate quota below this amount */
static const u64 min_cfs_rq_runtime = 1 * NSEC_PER_MSEC;
/* minimum remaining period time to redistribute slack quota */
static const u64 min_bandwidth_expiration = 2 * NSEC_PER_MSEC;
/* how long we wait to gather additional slack before distributing */
static const u64 cfs_bandwidth_slack_period = 5 * NSEC_PER_MSEC;

/* are we near the end of the current quota period? */
static int runtime_refresh_within(struct cfs_bandwidth *cfs_b, u64 min_expire)
{
	struct hrtimer *refresh_timer = &cfs_b->period_timer;
	u64 remaining;

	/* if the call-back is running a quota refresh is already occurring */
	if (hrtimer_callback_running(refresh_timer))
		return 1;

	/* is a quota refresh about to occur? */
	remaining = ktime_to_ns(hrtimer_expires_remaining(refresh_timer));
	if (remaining < min_expire)
		return 1;

	return 0;
}

static void start_cfs_slack_bandwidth(struct cfs_bandwidth *cfs_b)
{
	u64 min_left = cfs_bandwidth_slack_period + min_bandwidth_expiration;

	/* if there's a quota refresh soon don't bother with slack */
	if (runtime_refresh_within(cfs_b, min_left))
		return;

	start_bandwidth_timer(&cfs_b->slack_timer,
				ns_to_ktime(cfs_bandwidth_slack_period));
}

/* we know any runtime found here is valid as update_curr() precedes return */
static void __return_cfs_rq_runtime(struct cfs_rq *cfs_rq)
{
	struct cfs_bandwidth *cfs_b = tg_cfs_bandwidth(cfs_rq->tg);
	s64 slack_runtime = cfs_rq->runtime_remaining - min_cfs_rq_runtime;

	if (slack_runtime <= 0)
		return;

	raw_spin_lock(&cfs_b->lock);
	if (cfs_b->quota != RUNTIME_INF &&
	    cfs_rq->runtime_expires == cfs_b->runtime_expires) {
		cfs_b->runtime += slack_runtime;

		/* we are under rq->lock, defer unthrottling using a timer */
		if (cfs_b->runtime > sched_cfs_bandwidth_slice() &&
		    !list_empty(&cfs_b->throttled_cfs_rq))
			start_cfs_slack_bandwidth(cfs_b);
	}
	raw_spin_unlock(&cfs_b->lock);

	/* even if it's not valid for return we don't want to try again */
	cfs_rq->runtime_remaining -= slack_runtime;
}

static __always_inline void return_cfs_rq_runtime(struct cfs_rq *cfs_rq)
{
1902 1903 1904
	if (!cfs_bandwidth_used())
		return;

1905
	if (!cfs_rq->runtime_enabled || cfs_rq->nr_running)
1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942
		return;

	__return_cfs_rq_runtime(cfs_rq);
}

/*
 * This is done with a timer (instead of inline with bandwidth return) since
 * it's necessary to juggle rq->locks to unthrottle their respective cfs_rqs.
 */
static void do_sched_cfs_slack_timer(struct cfs_bandwidth *cfs_b)
{
	u64 runtime = 0, slice = sched_cfs_bandwidth_slice();
	u64 expires;

	/* confirm we're still not at a refresh boundary */
	if (runtime_refresh_within(cfs_b, min_bandwidth_expiration))
		return;

	raw_spin_lock(&cfs_b->lock);
	if (cfs_b->quota != RUNTIME_INF && cfs_b->runtime > slice) {
		runtime = cfs_b->runtime;
		cfs_b->runtime = 0;
	}
	expires = cfs_b->runtime_expires;
	raw_spin_unlock(&cfs_b->lock);

	if (!runtime)
		return;

	runtime = distribute_cfs_runtime(cfs_b, runtime, expires);

	raw_spin_lock(&cfs_b->lock);
	if (expires == cfs_b->runtime_expires)
		cfs_b->runtime = runtime;
	raw_spin_unlock(&cfs_b->lock);
}

1943 1944 1945 1946 1947 1948 1949
/*
 * When a group wakes up we want to make sure that its quota is not already
 * expired/exceeded, otherwise it may be allowed to steal additional ticks of
 * runtime as update_curr() throttling can not not trigger until it's on-rq.
 */
static void check_enqueue_throttle(struct cfs_rq *cfs_rq)
{
1950 1951 1952
	if (!cfs_bandwidth_used())
		return;

1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969
	/* an active group must be handled by the update_curr()->put() path */
	if (!cfs_rq->runtime_enabled || cfs_rq->curr)
		return;

	/* ensure the group is not already throttled */
	if (cfs_rq_throttled(cfs_rq))
		return;

	/* update runtime allocation */
	account_cfs_rq_runtime(cfs_rq, 0);
	if (cfs_rq->runtime_remaining <= 0)
		throttle_cfs_rq(cfs_rq);
}

/* conditionally throttle active cfs_rq's from put_prev_entity() */
static void check_cfs_rq_runtime(struct cfs_rq *cfs_rq)
{
1970 1971 1972
	if (!cfs_bandwidth_used())
		return;

1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984
	if (likely(!cfs_rq->runtime_enabled || cfs_rq->runtime_remaining > 0))
		return;

	/*
	 * it's possible for a throttled entity to be forced into a running
	 * state (e.g. set_curr_task), in this case we're finished.
	 */
	if (cfs_rq_throttled(cfs_rq))
		return;

	throttle_cfs_rq(cfs_rq);
}
1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090

static inline u64 default_cfs_period(void);
static int do_sched_cfs_period_timer(struct cfs_bandwidth *cfs_b, int overrun);
static void do_sched_cfs_slack_timer(struct cfs_bandwidth *cfs_b);

static enum hrtimer_restart sched_cfs_slack_timer(struct hrtimer *timer)
{
	struct cfs_bandwidth *cfs_b =
		container_of(timer, struct cfs_bandwidth, slack_timer);
	do_sched_cfs_slack_timer(cfs_b);

	return HRTIMER_NORESTART;
}

static enum hrtimer_restart sched_cfs_period_timer(struct hrtimer *timer)
{
	struct cfs_bandwidth *cfs_b =
		container_of(timer, struct cfs_bandwidth, period_timer);
	ktime_t now;
	int overrun;
	int idle = 0;

	for (;;) {
		now = hrtimer_cb_get_time(timer);
		overrun = hrtimer_forward(timer, now, cfs_b->period);

		if (!overrun)
			break;

		idle = do_sched_cfs_period_timer(cfs_b, overrun);
	}

	return idle ? HRTIMER_NORESTART : HRTIMER_RESTART;
}

void init_cfs_bandwidth(struct cfs_bandwidth *cfs_b)
{
	raw_spin_lock_init(&cfs_b->lock);
	cfs_b->runtime = 0;
	cfs_b->quota = RUNTIME_INF;
	cfs_b->period = ns_to_ktime(default_cfs_period());

	INIT_LIST_HEAD(&cfs_b->throttled_cfs_rq);
	hrtimer_init(&cfs_b->period_timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
	cfs_b->period_timer.function = sched_cfs_period_timer;
	hrtimer_init(&cfs_b->slack_timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
	cfs_b->slack_timer.function = sched_cfs_slack_timer;
}

static void init_cfs_rq_runtime(struct cfs_rq *cfs_rq)
{
	cfs_rq->runtime_enabled = 0;
	INIT_LIST_HEAD(&cfs_rq->throttled_list);
}

/* requires cfs_b->lock, may release to reprogram timer */
void __start_cfs_bandwidth(struct cfs_bandwidth *cfs_b)
{
	/*
	 * The timer may be active because we're trying to set a new bandwidth
	 * period or because we're racing with the tear-down path
	 * (timer_active==0 becomes visible before the hrtimer call-back
	 * terminates).  In either case we ensure that it's re-programmed
	 */
	while (unlikely(hrtimer_active(&cfs_b->period_timer))) {
		raw_spin_unlock(&cfs_b->lock);
		/* ensure cfs_b->lock is available while we wait */
		hrtimer_cancel(&cfs_b->period_timer);

		raw_spin_lock(&cfs_b->lock);
		/* if someone else restarted the timer then we're done */
		if (cfs_b->timer_active)
			return;
	}

	cfs_b->timer_active = 1;
	start_bandwidth_timer(&cfs_b->period_timer, cfs_b->period);
}

static void destroy_cfs_bandwidth(struct cfs_bandwidth *cfs_b)
{
	hrtimer_cancel(&cfs_b->period_timer);
	hrtimer_cancel(&cfs_b->slack_timer);
}

void unthrottle_offline_cfs_rqs(struct rq *rq)
{
	struct cfs_rq *cfs_rq;

	for_each_leaf_cfs_rq(rq, cfs_rq) {
		struct cfs_bandwidth *cfs_b = tg_cfs_bandwidth(cfs_rq->tg);

		if (!cfs_rq->runtime_enabled)
			continue;

		/*
		 * clock_task is not advancing so we just need to make sure
		 * there's some valid quota amount
		 */
		cfs_rq->runtime_remaining = cfs_b->quota;
		if (cfs_rq_throttled(cfs_rq))
			unthrottle_cfs_rq(cfs_rq);
	}
}

#else /* CONFIG_CFS_BANDWIDTH */
2091 2092
static void account_cfs_rq_runtime(struct cfs_rq *cfs_rq,
				     unsigned long delta_exec) {}
2093 2094
static void check_cfs_rq_runtime(struct cfs_rq *cfs_rq) {}
static void check_enqueue_throttle(struct cfs_rq *cfs_rq) {}
2095
static void return_cfs_rq_runtime(struct cfs_rq *cfs_rq) {}
2096 2097 2098 2099 2100

static inline int cfs_rq_throttled(struct cfs_rq *cfs_rq)
{
	return 0;
}
2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111

static inline int throttled_hierarchy(struct cfs_rq *cfs_rq)
{
	return 0;
}

static inline int throttled_lb_pair(struct task_group *tg,
				    int src_cpu, int dest_cpu)
{
	return 0;
}
2112 2113 2114 2115 2116

void init_cfs_bandwidth(struct cfs_bandwidth *cfs_b) {}

#ifdef CONFIG_FAIR_GROUP_SCHED
static void init_cfs_rq_runtime(struct cfs_rq *cfs_rq) {}
2117 2118
#endif

2119 2120 2121 2122 2123 2124 2125 2126 2127
static inline struct cfs_bandwidth *tg_cfs_bandwidth(struct task_group *tg)
{
	return NULL;
}
static inline void destroy_cfs_bandwidth(struct cfs_bandwidth *cfs_b) {}
void unthrottle_offline_cfs_rqs(struct rq *rq) {}

#endif /* CONFIG_CFS_BANDWIDTH */

2128 2129 2130 2131
/**************************************************
 * CFS operations on tasks:
 */

P
Peter Zijlstra 已提交
2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154
#ifdef CONFIG_SCHED_HRTICK
static void hrtick_start_fair(struct rq *rq, struct task_struct *p)
{
	struct sched_entity *se = &p->se;
	struct cfs_rq *cfs_rq = cfs_rq_of(se);

	WARN_ON(task_rq(p) != rq);

	if (hrtick_enabled(rq) && cfs_rq->nr_running > 1) {
		u64 slice = sched_slice(cfs_rq, se);
		u64 ran = se->sum_exec_runtime - se->prev_sum_exec_runtime;
		s64 delta = slice - ran;

		if (delta < 0) {
			if (rq->curr == p)
				resched_task(p);
			return;
		}

		/*
		 * Don't schedule slices shorter than 10000ns, that just
		 * doesn't make sense. Rely on vruntime for fairness.
		 */
2155
		if (rq->curr != p)
2156
			delta = max_t(s64, 10000LL, delta);
P
Peter Zijlstra 已提交
2157

2158
		hrtick_start(rq, delta);
P
Peter Zijlstra 已提交
2159 2160
	}
}
2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176

/*
 * called from enqueue/dequeue and updates the hrtick when the
 * current task is from our class and nr_running is low enough
 * to matter.
 */
static void hrtick_update(struct rq *rq)
{
	struct task_struct *curr = rq->curr;

	if (curr->sched_class != &fair_sched_class)
		return;

	if (cfs_rq_of(&curr->se)->nr_running < sched_nr_latency)
		hrtick_start_fair(rq, curr);
}
2177
#else /* !CONFIG_SCHED_HRTICK */
P
Peter Zijlstra 已提交
2178 2179 2180 2181
static inline void
hrtick_start_fair(struct rq *rq, struct task_struct *p)
{
}
2182 2183 2184 2185

static inline void hrtick_update(struct rq *rq)
{
}
P
Peter Zijlstra 已提交
2186 2187
#endif

2188 2189 2190 2191 2192
/*
 * The enqueue_task method is called before nr_running is
 * increased. Here we update the fair scheduling stats and
 * then put the task into the rbtree:
 */
2193
static void
2194
enqueue_task_fair(struct rq *rq, struct task_struct *p, int flags)
2195 2196
{
	struct cfs_rq *cfs_rq;
2197
	struct sched_entity *se = &p->se;
2198 2199

	for_each_sched_entity(se) {
2200
		if (se->on_rq)
2201 2202
			break;
		cfs_rq = cfs_rq_of(se);
2203
		enqueue_entity(cfs_rq, se, flags);
2204 2205 2206 2207 2208 2209 2210 2211 2212

		/*
		 * end evaluation on encountering a throttled cfs_rq
		 *
		 * note: in the case of encountering a throttled cfs_rq we will
		 * post the final h_nr_running increment below.
		*/
		if (cfs_rq_throttled(cfs_rq))
			break;
2213
		cfs_rq->h_nr_running++;
2214

2215
		flags = ENQUEUE_WAKEUP;
2216
	}
P
Peter Zijlstra 已提交
2217

P
Peter Zijlstra 已提交
2218
	for_each_sched_entity(se) {
2219
		cfs_rq = cfs_rq_of(se);
2220
		cfs_rq->h_nr_running++;
P
Peter Zijlstra 已提交
2221

2222 2223 2224
		if (cfs_rq_throttled(cfs_rq))
			break;

2225
		update_cfs_load(cfs_rq, 0);
2226
		update_cfs_shares(cfs_rq);
P
Peter Zijlstra 已提交
2227 2228
	}

2229 2230
	if (!se)
		inc_nr_running(rq);
2231
	hrtick_update(rq);
2232 2233
}

2234 2235
static void set_next_buddy(struct sched_entity *se);

2236 2237 2238 2239 2240
/*
 * The dequeue_task method is called before nr_running is
 * decreased. We remove the task from the rbtree and
 * update the fair scheduling stats:
 */
2241
static void dequeue_task_fair(struct rq *rq, struct task_struct *p, int flags)
2242 2243
{
	struct cfs_rq *cfs_rq;
2244
	struct sched_entity *se = &p->se;
2245
	int task_sleep = flags & DEQUEUE_SLEEP;
2246 2247 2248

	for_each_sched_entity(se) {
		cfs_rq = cfs_rq_of(se);
2249
		dequeue_entity(cfs_rq, se, flags);
2250 2251 2252 2253 2254 2255 2256 2257 2258

		/*
		 * end evaluation on encountering a throttled cfs_rq
		 *
		 * note: in the case of encountering a throttled cfs_rq we will
		 * post the final h_nr_running decrement below.
		*/
		if (cfs_rq_throttled(cfs_rq))
			break;
2259
		cfs_rq->h_nr_running--;
P
Peter Zijlstra 已提交
2260

2261
		/* Don't dequeue parent if it has other entities besides us */
2262 2263 2264 2265 2266 2267 2268
		if (cfs_rq->load.weight) {
			/*
			 * Bias pick_next to pick a task from this cfs_rq, as
			 * p is sleeping when it is within its sched_slice.
			 */
			if (task_sleep && parent_entity(se))
				set_next_buddy(parent_entity(se));
2269 2270 2271

			/* avoid re-evaluating load for this entity */
			se = parent_entity(se);
2272
			break;
2273
		}
2274
		flags |= DEQUEUE_SLEEP;
2275
	}
P
Peter Zijlstra 已提交
2276

P
Peter Zijlstra 已提交
2277
	for_each_sched_entity(se) {
2278
		cfs_rq = cfs_rq_of(se);
2279
		cfs_rq->h_nr_running--;
P
Peter Zijlstra 已提交
2280

2281 2282 2283
		if (cfs_rq_throttled(cfs_rq))
			break;

2284
		update_cfs_load(cfs_rq, 0);
2285
		update_cfs_shares(cfs_rq);
P
Peter Zijlstra 已提交
2286 2287
	}

2288 2289
	if (!se)
		dec_nr_running(rq);
2290
	hrtick_update(rq);
2291 2292
}

2293
#ifdef CONFIG_SMP
2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348
/* Used instead of source_load when we know the type == 0 */
static unsigned long weighted_cpuload(const int cpu)
{
	return cpu_rq(cpu)->load.weight;
}

/*
 * Return a low guess at the load of a migration-source cpu weighted
 * according to the scheduling class and "nice" value.
 *
 * We want to under-estimate the load of migration sources, to
 * balance conservatively.
 */
static unsigned long source_load(int cpu, int type)
{
	struct rq *rq = cpu_rq(cpu);
	unsigned long total = weighted_cpuload(cpu);

	if (type == 0 || !sched_feat(LB_BIAS))
		return total;

	return min(rq->cpu_load[type-1], total);
}

/*
 * Return a high guess at the load of a migration-target cpu weighted
 * according to the scheduling class and "nice" value.
 */
static unsigned long target_load(int cpu, int type)
{
	struct rq *rq = cpu_rq(cpu);
	unsigned long total = weighted_cpuload(cpu);

	if (type == 0 || !sched_feat(LB_BIAS))
		return total;

	return max(rq->cpu_load[type-1], total);
}

static unsigned long power_of(int cpu)
{
	return cpu_rq(cpu)->cpu_power;
}

static unsigned long cpu_avg_load_per_task(int cpu)
{
	struct rq *rq = cpu_rq(cpu);
	unsigned long nr_running = ACCESS_ONCE(rq->nr_running);

	if (nr_running)
		return rq->load.weight / nr_running;

	return 0;
}

2349

2350
static void task_waking_fair(struct task_struct *p)
2351 2352 2353
{
	struct sched_entity *se = &p->se;
	struct cfs_rq *cfs_rq = cfs_rq_of(se);
2354 2355 2356 2357
	u64 min_vruntime;

#ifndef CONFIG_64BIT
	u64 min_vruntime_copy;
2358

2359 2360 2361 2362 2363 2364 2365 2366
	do {
		min_vruntime_copy = cfs_rq->min_vruntime_copy;
		smp_rmb();
		min_vruntime = cfs_rq->min_vruntime;
	} while (min_vruntime != min_vruntime_copy);
#else
	min_vruntime = cfs_rq->min_vruntime;
#endif
2367

2368
	se->vruntime -= min_vruntime;
2369 2370
}

2371
#ifdef CONFIG_FAIR_GROUP_SCHED
2372 2373 2374 2375 2376 2377
/*
 * effective_load() calculates the load change as seen from the root_task_group
 *
 * Adding load to a group doesn't make a group heavier, but can cause movement
 * of group shares between cpus. Assuming the shares were perfectly aligned one
 * can calculate the shift in shares.
2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420
 *
 * Calculate the effective load difference if @wl is added (subtracted) to @tg
 * on this @cpu and results in a total addition (subtraction) of @wg to the
 * total group weight.
 *
 * Given a runqueue weight distribution (rw_i) we can compute a shares
 * distribution (s_i) using:
 *
 *   s_i = rw_i / \Sum rw_j						(1)
 *
 * Suppose we have 4 CPUs and our @tg is a direct child of the root group and
 * has 7 equal weight tasks, distributed as below (rw_i), with the resulting
 * shares distribution (s_i):
 *
 *   rw_i = {   2,   4,   1,   0 }
 *   s_i  = { 2/7, 4/7, 1/7,   0 }
 *
 * As per wake_affine() we're interested in the load of two CPUs (the CPU the
 * task used to run on and the CPU the waker is running on), we need to
 * compute the effect of waking a task on either CPU and, in case of a sync
 * wakeup, compute the effect of the current task going to sleep.
 *
 * So for a change of @wl to the local @cpu with an overall group weight change
 * of @wl we can compute the new shares distribution (s'_i) using:
 *
 *   s'_i = (rw_i + @wl) / (@wg + \Sum rw_j)				(2)
 *
 * Suppose we're interested in CPUs 0 and 1, and want to compute the load
 * differences in waking a task to CPU 0. The additional task changes the
 * weight and shares distributions like:
 *
 *   rw'_i = {   3,   4,   1,   0 }
 *   s'_i  = { 3/8, 4/8, 1/8,   0 }
 *
 * We can then compute the difference in effective weight by using:
 *
 *   dw_i = S * (s'_i - s_i)						(3)
 *
 * Where 'S' is the group weight as seen by its parent.
 *
 * Therefore the effective change in loads on CPU 0 would be 5/56 (3/8 - 2/7)
 * times the weight of the group. The effect on CPU 1 would be -4/56 (4/8 -
 * 4/7) times the weight of the group.
2421
 */
P
Peter Zijlstra 已提交
2422
static long effective_load(struct task_group *tg, int cpu, long wl, long wg)
2423
{
P
Peter Zijlstra 已提交
2424
	struct sched_entity *se = tg->se[cpu];
2425

2426
	if (!tg->parent)	/* the trivial, non-cgroup case */
2427 2428
		return wl;

P
Peter Zijlstra 已提交
2429
	for_each_sched_entity(se) {
2430
		long w, W;
P
Peter Zijlstra 已提交
2431

2432
		tg = se->my_q->tg;
2433

2434 2435 2436 2437
		/*
		 * W = @wg + \Sum rw_j
		 */
		W = wg + calc_tg_weight(tg, se->my_q);
P
Peter Zijlstra 已提交
2438

2439 2440 2441 2442
		/*
		 * w = rw_i + @wl
		 */
		w = se->my_q->load.weight + wl;
2443

2444 2445 2446 2447 2448
		/*
		 * wl = S * s'_i; see (2)
		 */
		if (W > 0 && w < W)
			wl = (w * tg->shares) / W;
2449 2450
		else
			wl = tg->shares;
2451

2452 2453 2454 2455 2456
		/*
		 * Per the above, wl is the new se->load.weight value; since
		 * those are clipped to [MIN_SHARES, ...) do so now. See
		 * calc_cfs_shares().
		 */
2457 2458
		if (wl < MIN_SHARES)
			wl = MIN_SHARES;
2459 2460 2461 2462

		/*
		 * wl = dw_i = S * (s'_i - s_i); see (3)
		 */
2463
		wl -= se->load.weight;
2464 2465 2466 2467 2468 2469 2470 2471

		/*
		 * Recursively apply this logic to all parent groups to compute
		 * the final effective load change on the root group. Since
		 * only the @tg group gets extra weight, all parent groups can
		 * only redistribute existing shares. @wl is the shift in shares
		 * resulting from this level per the above.
		 */
P
Peter Zijlstra 已提交
2472 2473
		wg = 0;
	}
2474

P
Peter Zijlstra 已提交
2475
	return wl;
2476 2477
}
#else
P
Peter Zijlstra 已提交
2478

2479 2480
static inline unsigned long effective_load(struct task_group *tg, int cpu,
		unsigned long wl, unsigned long wg)
P
Peter Zijlstra 已提交
2481
{
2482
	return wl;
2483
}
P
Peter Zijlstra 已提交
2484

2485 2486
#endif

2487
static int wake_affine(struct sched_domain *sd, struct task_struct *p, int sync)
2488
{
2489
	s64 this_load, load;
2490
	int idx, this_cpu, prev_cpu;
2491
	unsigned long tl_per_task;
2492
	struct task_group *tg;
2493
	unsigned long weight;
2494
	int balanced;
2495

2496 2497 2498 2499 2500
	idx	  = sd->wake_idx;
	this_cpu  = smp_processor_id();
	prev_cpu  = task_cpu(p);
	load	  = source_load(prev_cpu, idx);
	this_load = target_load(this_cpu, idx);
2501

2502 2503 2504 2505 2506
	/*
	 * If sync wakeup then subtract the (maximum possible)
	 * effect of the currently running task from the load
	 * of the current CPU:
	 */
2507 2508 2509 2510
	if (sync) {
		tg = task_group(current);
		weight = current->se.load.weight;

2511
		this_load += effective_load(tg, this_cpu, -weight, -weight);
2512 2513
		load += effective_load(tg, prev_cpu, 0, -weight);
	}
2514

2515 2516
	tg = task_group(p);
	weight = p->se.load.weight;
2517

2518 2519
	/*
	 * In low-load situations, where prev_cpu is idle and this_cpu is idle
2520 2521 2522
	 * due to the sync cause above having dropped this_load to 0, we'll
	 * always have an imbalance, but there's really nothing you can do
	 * about that, so that's good too.
2523 2524 2525 2526
	 *
	 * Otherwise check if either cpus are near enough in load to allow this
	 * task to be woken on this_cpu.
	 */
2527 2528
	if (this_load > 0) {
		s64 this_eff_load, prev_eff_load;
2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541

		this_eff_load = 100;
		this_eff_load *= power_of(prev_cpu);
		this_eff_load *= this_load +
			effective_load(tg, this_cpu, weight, weight);

		prev_eff_load = 100 + (sd->imbalance_pct - 100) / 2;
		prev_eff_load *= power_of(this_cpu);
		prev_eff_load *= load + effective_load(tg, prev_cpu, 0, weight);

		balanced = this_eff_load <= prev_eff_load;
	} else
		balanced = true;
2542

2543
	/*
I
Ingo Molnar 已提交
2544 2545 2546
	 * If the currently running task will sleep within
	 * a reasonable amount of time then attract this newly
	 * woken task:
2547
	 */
2548 2549
	if (sync && balanced)
		return 1;
2550

2551
	schedstat_inc(p, se.statistics.nr_wakeups_affine_attempts);
2552 2553
	tl_per_task = cpu_avg_load_per_task(this_cpu);

2554 2555 2556
	if (balanced ||
	    (this_load <= load &&
	     this_load + target_load(prev_cpu, idx) <= tl_per_task)) {
2557 2558 2559 2560 2561
		/*
		 * This domain has SD_WAKE_AFFINE and
		 * p is cache cold in this domain, and
		 * there is no bad imbalance.
		 */
2562
		schedstat_inc(sd, ttwu_move_affine);
2563
		schedstat_inc(p, se.statistics.nr_wakeups_affine);
2564 2565 2566 2567 2568 2569

		return 1;
	}
	return 0;
}

2570 2571 2572 2573 2574
/*
 * find_idlest_group finds and returns the least busy CPU group within the
 * domain.
 */
static struct sched_group *
P
Peter Zijlstra 已提交
2575
find_idlest_group(struct sched_domain *sd, struct task_struct *p,
2576
		  int this_cpu, int load_idx)
2577
{
2578
	struct sched_group *idlest = NULL, *group = sd->groups;
2579 2580
	unsigned long min_load = ULONG_MAX, this_load = 0;
	int imbalance = 100 + (sd->imbalance_pct-100)/2;
2581

2582 2583 2584 2585
	do {
		unsigned long load, avg_load;
		int local_group;
		int i;
2586

2587 2588
		/* Skip over this group if it has no CPUs allowed */
		if (!cpumask_intersects(sched_group_cpus(group),
2589
					tsk_cpus_allowed(p)))
2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608
			continue;

		local_group = cpumask_test_cpu(this_cpu,
					       sched_group_cpus(group));

		/* Tally up the load of all CPUs in the group */
		avg_load = 0;

		for_each_cpu(i, sched_group_cpus(group)) {
			/* Bias balancing toward cpus of our domain */
			if (local_group)
				load = source_load(i, load_idx);
			else
				load = target_load(i, load_idx);

			avg_load += load;
		}

		/* Adjust by relative CPU power of the group */
2609
		avg_load = (avg_load * SCHED_POWER_SCALE) / group->sgp->power;
2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634

		if (local_group) {
			this_load = avg_load;
		} else if (avg_load < min_load) {
			min_load = avg_load;
			idlest = group;
		}
	} while (group = group->next, group != sd->groups);

	if (!idlest || 100*this_load < imbalance*min_load)
		return NULL;
	return idlest;
}

/*
 * find_idlest_cpu - find the idlest cpu among the cpus in group.
 */
static int
find_idlest_cpu(struct sched_group *group, struct task_struct *p, int this_cpu)
{
	unsigned long load, min_load = ULONG_MAX;
	int idlest = -1;
	int i;

	/* Traverse only the allowed CPUs */
2635
	for_each_cpu_and(i, sched_group_cpus(group), tsk_cpus_allowed(p)) {
2636 2637 2638 2639 2640
		load = weighted_cpuload(i);

		if (load < min_load || (load == min_load && i == this_cpu)) {
			min_load = load;
			idlest = i;
2641 2642 2643
		}
	}

2644 2645
	return idlest;
}
2646

2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668
/**
 * highest_flag_domain - Return highest sched_domain containing flag.
 * @cpu:	The cpu whose highest level of sched domain is to
 *		be returned.
 * @flag:	The flag to check for the highest sched_domain
 *		for the given cpu.
 *
 * Returns the highest sched_domain of a cpu which contains the given flag.
 */
static inline struct sched_domain *highest_flag_domain(int cpu, int flag)
{
	struct sched_domain *sd, *hsd = NULL;

	for_each_domain(cpu, sd) {
		if (!(sd->flags & flag))
			break;
		hsd = sd;
	}

	return hsd;
}

2669 2670 2671
/*
 * Try and locate an idle CPU in the sched_domain.
 */
2672
static int select_idle_sibling(struct task_struct *p, int target)
2673 2674 2675
{
	int cpu = smp_processor_id();
	int prev_cpu = task_cpu(p);
2676
	struct sched_domain *sd;
2677
	struct sched_group *sg;
2678
	int i;
2679 2680

	/*
2681 2682
	 * If the task is going to be woken-up on this cpu and if it is
	 * already idle, then it is the right target.
2683
	 */
2684 2685 2686 2687 2688 2689 2690 2691
	if (target == cpu && idle_cpu(cpu))
		return cpu;

	/*
	 * If the task is going to be woken-up on the cpu where it previously
	 * ran and if it is currently idle, then it the right target.
	 */
	if (target == prev_cpu && idle_cpu(prev_cpu))
2692
		return prev_cpu;
2693 2694

	/*
2695
	 * Otherwise, iterate the domains and find an elegible idle cpu.
2696
	 */
2697
	rcu_read_lock();
2698

2699 2700
	sd = highest_flag_domain(target, SD_SHARE_PKG_RESOURCES);
	for_each_lower_domain(sd) {
2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717
		sg = sd->groups;
		do {
			if (!cpumask_intersects(sched_group_cpus(sg),
						tsk_cpus_allowed(p)))
				goto next;

			for_each_cpu(i, sched_group_cpus(sg)) {
				if (!idle_cpu(i))
					goto next;
			}

			target = cpumask_first_and(sched_group_cpus(sg),
					tsk_cpus_allowed(p));
			goto done;
next:
			sg = sg->next;
		} while (sg != sd->groups);
2718
	}
2719
done:
2720
	rcu_read_unlock();
2721 2722 2723 2724

	return target;
}

2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735
/*
 * sched_balance_self: balance the current task (running on cpu) in domains
 * that have the 'flag' flag set. In practice, this is SD_BALANCE_FORK and
 * SD_BALANCE_EXEC.
 *
 * Balance, ie. select the least loaded group.
 *
 * Returns the target CPU number, or the same CPU if no balancing is needed.
 *
 * preempt must be disabled.
 */
2736
static int
2737
select_task_rq_fair(struct task_struct *p, int sd_flag, int wake_flags)
2738
{
2739
	struct sched_domain *tmp, *affine_sd = NULL, *sd = NULL;
2740 2741 2742
	int cpu = smp_processor_id();
	int prev_cpu = task_cpu(p);
	int new_cpu = cpu;
2743
	int want_affine = 0;
2744
	int want_sd = 1;
2745
	int sync = wake_flags & WF_SYNC;
2746

2747 2748 2749
	if (p->rt.nr_cpus_allowed == 1)
		return prev_cpu;

2750
	if (sd_flag & SD_BALANCE_WAKE) {
2751
		if (cpumask_test_cpu(cpu, tsk_cpus_allowed(p)))
2752 2753 2754
			want_affine = 1;
		new_cpu = prev_cpu;
	}
2755

2756
	rcu_read_lock();
2757
	for_each_domain(cpu, tmp) {
2758 2759 2760
		if (!(tmp->flags & SD_LOAD_BALANCE))
			continue;

2761
		/*
2762 2763
		 * If power savings logic is enabled for a domain, see if we
		 * are not overloaded, if so, don't balance wider.
2764
		 */
P
Peter Zijlstra 已提交
2765
		if (tmp->flags & (SD_POWERSAVINGS_BALANCE|SD_PREFER_LOCAL)) {
2766 2767 2768 2769 2770 2771 2772 2773 2774 2775
			unsigned long power = 0;
			unsigned long nr_running = 0;
			unsigned long capacity;
			int i;

			for_each_cpu(i, sched_domain_span(tmp)) {
				power += power_of(i);
				nr_running += cpu_rq(i)->cfs.nr_running;
			}

2776
			capacity = DIV_ROUND_CLOSEST(power, SCHED_POWER_SCALE);
2777

P
Peter Zijlstra 已提交
2778 2779 2780 2781
			if (tmp->flags & SD_POWERSAVINGS_BALANCE)
				nr_running /= 2;

			if (nr_running < capacity)
2782
				want_sd = 0;
2783
		}
2784

2785
		/*
2786 2787
		 * If both cpu and prev_cpu are part of this domain,
		 * cpu is a valid SD_WAKE_AFFINE target.
2788
		 */
2789 2790 2791 2792
		if (want_affine && (tmp->flags & SD_WAKE_AFFINE) &&
		    cpumask_test_cpu(prev_cpu, sched_domain_span(tmp))) {
			affine_sd = tmp;
			want_affine = 0;
2793 2794
		}

2795 2796 2797
		if (!want_sd && !want_affine)
			break;

2798
		if (!(tmp->flags & sd_flag))
2799 2800
			continue;

2801 2802 2803 2804
		if (want_sd)
			sd = tmp;
	}

2805
	if (affine_sd) {
2806
		if (cpu == prev_cpu || wake_affine(affine_sd, p, sync))
2807 2808 2809 2810
			prev_cpu = cpu;

		new_cpu = select_idle_sibling(p, prev_cpu);
		goto unlock;
2811
	}
2812

2813
	while (sd) {
2814
		int load_idx = sd->forkexec_idx;
2815
		struct sched_group *group;
2816
		int weight;
2817

2818
		if (!(sd->flags & sd_flag)) {
2819 2820 2821
			sd = sd->child;
			continue;
		}
2822

2823 2824
		if (sd_flag & SD_BALANCE_WAKE)
			load_idx = sd->wake_idx;
2825

2826
		group = find_idlest_group(sd, p, cpu, load_idx);
2827 2828 2829 2830
		if (!group) {
			sd = sd->child;
			continue;
		}
I
Ingo Molnar 已提交
2831

2832
		new_cpu = find_idlest_cpu(group, p, cpu);
2833 2834 2835 2836
		if (new_cpu == -1 || new_cpu == cpu) {
			/* Now try balancing at a lower domain level of cpu */
			sd = sd->child;
			continue;
2837
		}
2838 2839 2840

		/* Now try balancing at a lower domain level of new_cpu */
		cpu = new_cpu;
2841
		weight = sd->span_weight;
2842 2843
		sd = NULL;
		for_each_domain(cpu, tmp) {
2844
			if (weight <= tmp->span_weight)
2845
				break;
2846
			if (tmp->flags & sd_flag)
2847 2848 2849
				sd = tmp;
		}
		/* while loop will break here if sd == NULL */
2850
	}
2851 2852
unlock:
	rcu_read_unlock();
2853

2854
	return new_cpu;
2855 2856 2857
}
#endif /* CONFIG_SMP */

P
Peter Zijlstra 已提交
2858 2859
static unsigned long
wakeup_gran(struct sched_entity *curr, struct sched_entity *se)
2860 2861 2862 2863
{
	unsigned long gran = sysctl_sched_wakeup_granularity;

	/*
P
Peter Zijlstra 已提交
2864 2865
	 * Since its curr running now, convert the gran from real-time
	 * to virtual-time in his units.
M
Mike Galbraith 已提交
2866 2867 2868 2869 2870 2871 2872 2873 2874
	 *
	 * By using 'se' instead of 'curr' we penalize light tasks, so
	 * they get preempted easier. That is, if 'se' < 'curr' then
	 * the resulting gran will be larger, therefore penalizing the
	 * lighter, if otoh 'se' > 'curr' then the resulting gran will
	 * be smaller, again penalizing the lighter task.
	 *
	 * This is especially important for buddies when the leftmost
	 * task is higher priority than the buddy.
2875
	 */
2876
	return calc_delta_fair(gran, se);
2877 2878
}

2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900
/*
 * Should 'se' preempt 'curr'.
 *
 *             |s1
 *        |s2
 *   |s3
 *         g
 *      |<--->|c
 *
 *  w(c, s1) = -1
 *  w(c, s2) =  0
 *  w(c, s3) =  1
 *
 */
static int
wakeup_preempt_entity(struct sched_entity *curr, struct sched_entity *se)
{
	s64 gran, vdiff = curr->vruntime - se->vruntime;

	if (vdiff <= 0)
		return -1;

P
Peter Zijlstra 已提交
2901
	gran = wakeup_gran(curr, se);
2902 2903 2904 2905 2906 2907
	if (vdiff > gran)
		return 1;

	return 0;
}

2908 2909
static void set_last_buddy(struct sched_entity *se)
{
2910 2911 2912 2913 2914
	if (entity_is_task(se) && unlikely(task_of(se)->policy == SCHED_IDLE))
		return;

	for_each_sched_entity(se)
		cfs_rq_of(se)->last = se;
2915 2916 2917 2918
}

static void set_next_buddy(struct sched_entity *se)
{
2919 2920 2921 2922 2923
	if (entity_is_task(se) && unlikely(task_of(se)->policy == SCHED_IDLE))
		return;

	for_each_sched_entity(se)
		cfs_rq_of(se)->next = se;
2924 2925
}

2926 2927
static void set_skip_buddy(struct sched_entity *se)
{
2928 2929
	for_each_sched_entity(se)
		cfs_rq_of(se)->skip = se;
2930 2931
}

2932 2933 2934
/*
 * Preempt the current task with a newly woken task if needed:
 */
2935
static void check_preempt_wakeup(struct rq *rq, struct task_struct *p, int wake_flags)
2936 2937
{
	struct task_struct *curr = rq->curr;
2938
	struct sched_entity *se = &curr->se, *pse = &p->se;
2939
	struct cfs_rq *cfs_rq = task_cfs_rq(curr);
2940
	int scale = cfs_rq->nr_running >= sched_nr_latency;
2941
	int next_buddy_marked = 0;
2942

I
Ingo Molnar 已提交
2943 2944 2945
	if (unlikely(se == pse))
		return;

2946 2947 2948 2949 2950 2951 2952 2953 2954
	/*
	 * This is possible from callers such as pull_task(), in which we
	 * unconditionally check_prempt_curr() after an enqueue (which may have
	 * lead to a throttle).  This both saves work and prevents false
	 * next-buddy nomination below.
	 */
	if (unlikely(throttled_hierarchy(cfs_rq_of(pse))))
		return;

2955
	if (sched_feat(NEXT_BUDDY) && scale && !(wake_flags & WF_FORK)) {
M
Mike Galbraith 已提交
2956
		set_next_buddy(pse);
2957 2958
		next_buddy_marked = 1;
	}
P
Peter Zijlstra 已提交
2959

2960 2961 2962
	/*
	 * We can come here with TIF_NEED_RESCHED already set from new task
	 * wake up path.
2963 2964 2965 2966 2967 2968
	 *
	 * Note: this also catches the edge-case of curr being in a throttled
	 * group (e.g. via set_curr_task), since update_curr() (in the
	 * enqueue of curr) will have resulted in resched being set.  This
	 * prevents us from potentially nominating it as a false LAST_BUDDY
	 * below.
2969 2970 2971 2972
	 */
	if (test_tsk_need_resched(curr))
		return;

2973 2974 2975 2976 2977
	/* Idle tasks are by definition preempted by non-idle tasks. */
	if (unlikely(curr->policy == SCHED_IDLE) &&
	    likely(p->policy != SCHED_IDLE))
		goto preempt;

2978
	/*
2979 2980
	 * Batch and idle tasks do not preempt non-idle tasks (their preemption
	 * is driven by the tick):
2981
	 */
2982
	if (unlikely(p->policy != SCHED_NORMAL))
2983
		return;
2984

2985
	find_matching_se(&se, &pse);
2986
	update_curr(cfs_rq_of(se));
2987
	BUG_ON(!pse);
2988 2989 2990 2991 2992 2993 2994
	if (wakeup_preempt_entity(se, pse) == 1) {
		/*
		 * Bias pick_next to pick the sched entity that is
		 * triggering this preemption.
		 */
		if (!next_buddy_marked)
			set_next_buddy(pse);
2995
		goto preempt;
2996
	}
2997

2998
	return;
2999

3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015
preempt:
	resched_task(curr);
	/*
	 * Only set the backward buddy when the current task is still
	 * on the rq. This can happen when a wakeup gets interleaved
	 * with schedule on the ->pre_schedule() or idle_balance()
	 * point, either of which can * drop the rq lock.
	 *
	 * Also, during early boot the idle thread is in the fair class,
	 * for obvious reasons its a bad idea to schedule back to it.
	 */
	if (unlikely(!se->on_rq || curr == rq->idle))
		return;

	if (sched_feat(LAST_BUDDY) && scale && entity_is_task(se))
		set_last_buddy(se);
3016 3017
}

3018
static struct task_struct *pick_next_task_fair(struct rq *rq)
3019
{
P
Peter Zijlstra 已提交
3020
	struct task_struct *p;
3021 3022 3023
	struct cfs_rq *cfs_rq = &rq->cfs;
	struct sched_entity *se;

3024
	if (!cfs_rq->nr_running)
3025 3026 3027
		return NULL;

	do {
3028
		se = pick_next_entity(cfs_rq);
3029
		set_next_entity(cfs_rq, se);
3030 3031 3032
		cfs_rq = group_cfs_rq(se);
	} while (cfs_rq);

P
Peter Zijlstra 已提交
3033 3034 3035 3036
	p = task_of(se);
	hrtick_start_fair(rq, p);

	return p;
3037 3038 3039 3040 3041
}

/*
 * Account for a descheduled task:
 */
3042
static void put_prev_task_fair(struct rq *rq, struct task_struct *prev)
3043 3044 3045 3046 3047 3048
{
	struct sched_entity *se = &prev->se;
	struct cfs_rq *cfs_rq;

	for_each_sched_entity(se) {
		cfs_rq = cfs_rq_of(se);
3049
		put_prev_entity(cfs_rq, se);
3050 3051 3052
	}
}

3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063 3064 3065 3066 3067 3068 3069 3070 3071 3072 3073 3074 3075 3076 3077
/*
 * sched_yield() is very simple
 *
 * The magic of dealing with the ->skip buddy is in pick_next_entity.
 */
static void yield_task_fair(struct rq *rq)
{
	struct task_struct *curr = rq->curr;
	struct cfs_rq *cfs_rq = task_cfs_rq(curr);
	struct sched_entity *se = &curr->se;

	/*
	 * Are we the only task in the tree?
	 */
	if (unlikely(rq->nr_running == 1))
		return;

	clear_buddies(cfs_rq, se);

	if (curr->policy != SCHED_BATCH) {
		update_rq_clock(rq);
		/*
		 * Update run-time statistics of the 'current'.
		 */
		update_curr(cfs_rq);
3078 3079 3080 3081 3082 3083
		/*
		 * Tell update_rq_clock() that we've just updated,
		 * so we don't do microscopic update in schedule()
		 * and double the fastpath cost.
		 */
		 rq->skip_clock_update = 1;
3084 3085 3086 3087 3088
	}

	set_skip_buddy(se);
}

3089 3090 3091 3092
static bool yield_to_task_fair(struct rq *rq, struct task_struct *p, bool preempt)
{
	struct sched_entity *se = &p->se;

3093 3094
	/* throttled hierarchies are not runnable */
	if (!se->on_rq || throttled_hierarchy(cfs_rq_of(se)))
3095 3096 3097 3098 3099 3100 3101 3102 3103 3104
		return false;

	/* Tell the scheduler that we'd really like pse to run next. */
	set_next_buddy(se);

	yield_task_fair(rq);

	return true;
}

3105
#ifdef CONFIG_SMP
3106 3107 3108 3109
/**************************************************
 * Fair scheduling class load-balancing methods:
 */

3110 3111 3112 3113 3114 3115 3116 3117 3118 3119 3120 3121 3122
/*
 * pull_task - move a task from a remote runqueue to the local runqueue.
 * Both runqueues must be locked.
 */
static void pull_task(struct rq *src_rq, struct task_struct *p,
		      struct rq *this_rq, int this_cpu)
{
	deactivate_task(src_rq, p, 0);
	set_task_cpu(p, this_cpu);
	activate_task(this_rq, p, 0);
	check_preempt_curr(this_rq, p, 0);
}

3123 3124 3125 3126 3127 3128 3129 3130 3131 3132 3133 3134 3135 3136 3137 3138 3139 3140 3141 3142 3143 3144 3145 3146 3147 3148 3149 3150 3151 3152 3153 3154
/*
 * Is this task likely cache-hot:
 */
static int
task_hot(struct task_struct *p, u64 now, struct sched_domain *sd)
{
	s64 delta;

	if (p->sched_class != &fair_sched_class)
		return 0;

	if (unlikely(p->policy == SCHED_IDLE))
		return 0;

	/*
	 * Buddy candidates are cache hot:
	 */
	if (sched_feat(CACHE_HOT_BUDDY) && this_rq()->nr_running &&
			(&p->se == cfs_rq_of(&p->se)->next ||
			 &p->se == cfs_rq_of(&p->se)->last))
		return 1;

	if (sysctl_sched_migration_cost == -1)
		return 1;
	if (sysctl_sched_migration_cost == 0)
		return 0;

	delta = now - p->se.exec_start;

	return delta < (s64)sysctl_sched_migration_cost;
}

3155 3156 3157 3158 3159 3160 3161 3162 3163 3164 3165 3166 3167 3168 3169
/*
 * can_migrate_task - may task p from runqueue rq be migrated to this_cpu?
 */
static
int can_migrate_task(struct task_struct *p, struct rq *rq, int this_cpu,
		     struct sched_domain *sd, enum cpu_idle_type idle,
		     int *all_pinned)
{
	int tsk_cache_hot = 0;
	/*
	 * We do not migrate tasks that are:
	 * 1) running (obviously), or
	 * 2) cannot be migrated to this CPU due to cpus_allowed, or
	 * 3) are cache-hot on their current CPU.
	 */
3170
	if (!cpumask_test_cpu(this_cpu, tsk_cpus_allowed(p))) {
3171
		schedstat_inc(p, se.statistics.nr_failed_migrations_affine);
3172 3173 3174 3175 3176
		return 0;
	}
	*all_pinned = 0;

	if (task_running(rq, p)) {
3177
		schedstat_inc(p, se.statistics.nr_failed_migrations_running);
3178 3179 3180 3181 3182 3183 3184 3185 3186
		return 0;
	}

	/*
	 * Aggressive migration if:
	 * 1) task is cache cold, or
	 * 2) too many balance attempts have failed.
	 */

3187
	tsk_cache_hot = task_hot(p, rq->clock_task, sd);
3188 3189 3190 3191 3192
	if (!tsk_cache_hot ||
		sd->nr_balance_failed > sd->cache_nice_tries) {
#ifdef CONFIG_SCHEDSTATS
		if (tsk_cache_hot) {
			schedstat_inc(sd, lb_hot_gained[idle]);
3193
			schedstat_inc(p, se.statistics.nr_forced_migrations);
3194 3195 3196 3197 3198 3199
		}
#endif
		return 1;
	}

	if (tsk_cache_hot) {
3200
		schedstat_inc(p, se.statistics.nr_failed_migrations_hot);
3201 3202 3203 3204 3205
		return 0;
	}
	return 1;
}

3206 3207 3208 3209 3210 3211 3212 3213 3214 3215 3216 3217 3218 3219 3220 3221 3222
/*
 * move_one_task tries to move exactly one task from busiest to this_rq, as
 * part of active balancing operations within "domain".
 * Returns 1 if successful and 0 otherwise.
 *
 * Called with both runqueues locked.
 */
static int
move_one_task(struct rq *this_rq, int this_cpu, struct rq *busiest,
	      struct sched_domain *sd, enum cpu_idle_type idle)
{
	struct task_struct *p, *n;
	struct cfs_rq *cfs_rq;
	int pinned = 0;

	for_each_leaf_cfs_rq(busiest, cfs_rq) {
		list_for_each_entry_safe(p, n, &cfs_rq->tasks, se.group_node) {
3223 3224 3225
			if (throttled_lb_pair(task_group(p),
					      busiest->cpu, this_cpu))
				break;
3226 3227 3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244

			if (!can_migrate_task(p, busiest, this_cpu,
						sd, idle, &pinned))
				continue;

			pull_task(busiest, p, this_rq, this_cpu);
			/*
			 * Right now, this is only the second place pull_task()
			 * is called, so we can safely collect pull_task()
			 * stats here rather than inside pull_task().
			 */
			schedstat_inc(sd, lb_gained[idle]);
			return 1;
		}
	}

	return 0;
}

3245 3246 3247 3248
static unsigned long
balance_tasks(struct rq *this_rq, int this_cpu, struct rq *busiest,
	      unsigned long max_load_move, struct sched_domain *sd,
	      enum cpu_idle_type idle, int *all_pinned,
3249
	      struct cfs_rq *busiest_cfs_rq)
3250
{
K
Ken Chen 已提交
3251
	int loops = 0, pulled = 0;
3252
	long rem_load_move = max_load_move;
3253
	struct task_struct *p, *n;
3254 3255 3256 3257

	if (max_load_move == 0)
		goto out;

3258 3259 3260
	list_for_each_entry_safe(p, n, &busiest_cfs_rq->tasks, se.group_node) {
		if (loops++ > sysctl_sched_nr_migrate)
			break;
3261

3262
		if ((p->se.load.weight >> 1) > rem_load_move ||
K
Ken Chen 已提交
3263 3264
		    !can_migrate_task(p, busiest, this_cpu, sd, idle,
				      all_pinned))
3265
			continue;
3266

3267 3268 3269
		pull_task(busiest, p, this_rq, this_cpu);
		pulled++;
		rem_load_move -= p->se.load.weight;
3270 3271

#ifdef CONFIG_PREEMPT
3272 3273 3274 3275 3276 3277 3278
		/*
		 * NEWIDLE balancing is a source of latency, so preemptible
		 * kernels will stop after the first task is pulled to minimize
		 * the critical section.
		 */
		if (idle == CPU_NEWLY_IDLE)
			break;
3279 3280
#endif

3281 3282 3283 3284 3285 3286
		/*
		 * We only want to steal up to the prescribed amount of
		 * weighted load.
		 */
		if (rem_load_move <= 0)
			break;
3287 3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298
	}
out:
	/*
	 * Right now, this is one of only two places pull_task() is called,
	 * so we can safely collect pull_task() stats here rather than
	 * inside pull_task().
	 */
	schedstat_add(sd, lb_gained[idle], pulled);

	return max_load_move - rem_load_move;
}

P
Peter Zijlstra 已提交
3299
#ifdef CONFIG_FAIR_GROUP_SCHED
3300 3301 3302
/*
 * update tg->load_weight by folding this cpu's load_avg
 */
3303
static int update_shares_cpu(struct task_group *tg, int cpu)
3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315 3316 3317
{
	struct cfs_rq *cfs_rq;
	unsigned long flags;
	struct rq *rq;

	if (!tg->se[cpu])
		return 0;

	rq = cpu_rq(cpu);
	cfs_rq = tg->cfs_rq[cpu];

	raw_spin_lock_irqsave(&rq->lock, flags);

	update_rq_clock(rq);
3318
	update_cfs_load(cfs_rq, 1);
3319 3320 3321 3322 3323

	/*
	 * We need to update shares after updating tg->load_weight in
	 * order to adjust the weight of groups with long running tasks.
	 */
3324
	update_cfs_shares(cfs_rq);
3325 3326 3327 3328 3329 3330 3331 3332 3333 3334 3335 3336

	raw_spin_unlock_irqrestore(&rq->lock, flags);

	return 0;
}

static void update_shares(int cpu)
{
	struct cfs_rq *cfs_rq;
	struct rq *rq = cpu_rq(cpu);

	rcu_read_lock();
3337 3338 3339 3340
	/*
	 * Iterates the task_group tree in a bottom up fashion, see
	 * list_add_leaf_cfs_rq() for details.
	 */
3341 3342 3343 3344 3345
	for_each_leaf_cfs_rq(rq, cfs_rq) {
		/* throttled entities do not contribute to load */
		if (throttled_hierarchy(cfs_rq))
			continue;

3346
		update_shares_cpu(cfs_rq->tg, cpu);
3347
	}
3348 3349 3350
	rcu_read_unlock();
}

3351 3352 3353 3354 3355 3356 3357 3358 3359 3360 3361 3362 3363 3364 3365 3366 3367 3368 3369 3370 3371 3372 3373 3374 3375 3376 3377 3378
/*
 * Compute the cpu's hierarchical load factor for each task group.
 * This needs to be done in a top-down fashion because the load of a child
 * group is a fraction of its parents load.
 */
static int tg_load_down(struct task_group *tg, void *data)
{
	unsigned long load;
	long cpu = (long)data;

	if (!tg->parent) {
		load = cpu_rq(cpu)->load.weight;
	} else {
		load = tg->parent->cfs_rq[cpu]->h_load;
		load *= tg->se[cpu]->load.weight;
		load /= tg->parent->cfs_rq[cpu]->load.weight + 1;
	}

	tg->cfs_rq[cpu]->h_load = load;

	return 0;
}

static void update_h_load(long cpu)
{
	walk_tg_tree(tg_load_down, tg_nop, (void *)cpu);
}

P
Peter Zijlstra 已提交
3379 3380 3381 3382
static unsigned long
load_balance_fair(struct rq *this_rq, int this_cpu, struct rq *busiest,
		  unsigned long max_load_move,
		  struct sched_domain *sd, enum cpu_idle_type idle,
3383
		  int *all_pinned)
P
Peter Zijlstra 已提交
3384 3385
{
	long rem_load_move = max_load_move;
3386
	struct cfs_rq *busiest_cfs_rq;
P
Peter Zijlstra 已提交
3387 3388

	rcu_read_lock();
3389
	update_h_load(cpu_of(busiest));
P
Peter Zijlstra 已提交
3390

3391
	for_each_leaf_cfs_rq(busiest, busiest_cfs_rq) {
P
Peter Zijlstra 已提交
3392 3393 3394 3395 3396
		unsigned long busiest_h_load = busiest_cfs_rq->h_load;
		unsigned long busiest_weight = busiest_cfs_rq->load.weight;
		u64 rem_load, moved_load;

		/*
3397
		 * empty group or part of a throttled hierarchy
P
Peter Zijlstra 已提交
3398
		 */
3399 3400
		if (!busiest_cfs_rq->task_weight ||
		    throttled_lb_pair(busiest_cfs_rq->tg, cpu_of(busiest), this_cpu))
P
Peter Zijlstra 已提交
3401 3402 3403 3404 3405 3406
			continue;

		rem_load = (u64)rem_load_move * busiest_weight;
		rem_load = div_u64(rem_load, busiest_h_load + 1);

		moved_load = balance_tasks(this_rq, this_cpu, busiest,
3407
				rem_load, sd, idle, all_pinned,
P
Peter Zijlstra 已提交
3408 3409 3410 3411 3412 3413 3414 3415 3416 3417 3418 3419 3420 3421 3422 3423 3424
				busiest_cfs_rq);

		if (!moved_load)
			continue;

		moved_load *= busiest_h_load;
		moved_load = div_u64(moved_load, busiest_weight + 1);

		rem_load_move -= moved_load;
		if (rem_load_move < 0)
			break;
	}
	rcu_read_unlock();

	return max_load_move - rem_load_move;
}
#else
3425 3426 3427 3428
static inline void update_shares(int cpu)
{
}

P
Peter Zijlstra 已提交
3429 3430 3431 3432
static unsigned long
load_balance_fair(struct rq *this_rq, int this_cpu, struct rq *busiest,
		  unsigned long max_load_move,
		  struct sched_domain *sd, enum cpu_idle_type idle,
3433
		  int *all_pinned)
P
Peter Zijlstra 已提交
3434 3435 3436
{
	return balance_tasks(this_rq, this_cpu, busiest,
			max_load_move, sd, idle, all_pinned,
3437
			&busiest->cfs);
P
Peter Zijlstra 已提交
3438 3439 3440
}
#endif

3441 3442 3443 3444 3445 3446 3447 3448 3449 3450 3451 3452
/*
 * move_tasks tries to move up to max_load_move weighted load from busiest to
 * this_rq, as part of a balancing operation within domain "sd".
 * Returns 1 if successful and 0 otherwise.
 *
 * Called with both runqueues locked.
 */
static int move_tasks(struct rq *this_rq, int this_cpu, struct rq *busiest,
		      unsigned long max_load_move,
		      struct sched_domain *sd, enum cpu_idle_type idle,
		      int *all_pinned)
{
3453
	unsigned long total_load_moved = 0, load_moved;
3454 3455

	do {
3456
		load_moved = load_balance_fair(this_rq, this_cpu, busiest,
3457
				max_load_move - total_load_moved,
3458
				sd, idle, all_pinned);
3459 3460

		total_load_moved += load_moved;
3461 3462 3463 3464 3465 3466 3467 3468 3469

#ifdef CONFIG_PREEMPT
		/*
		 * NEWIDLE balancing is a source of latency, so preemptible
		 * kernels will stop after the first task is pulled to minimize
		 * the critical section.
		 */
		if (idle == CPU_NEWLY_IDLE && this_rq->nr_running)
			break;
3470 3471 3472 3473

		if (raw_spin_is_contended(&this_rq->lock) ||
				raw_spin_is_contended(&busiest->lock))
			break;
3474
#endif
3475
	} while (load_moved && max_load_move > total_load_moved);
3476 3477 3478 3479 3480 3481 3482 3483 3484 3485 3486 3487 3488 3489 3490 3491 3492 3493 3494 3495

	return total_load_moved > 0;
}

/********** Helpers for find_busiest_group ************************/
/*
 * sd_lb_stats - Structure to store the statistics of a sched_domain
 * 		during load balancing.
 */
struct sd_lb_stats {
	struct sched_group *busiest; /* Busiest group in this sd */
	struct sched_group *this;  /* Local group in this sd */
	unsigned long total_load;  /* Total load of all groups in sd */
	unsigned long total_pwr;   /*	Total power of all groups in sd */
	unsigned long avg_load;	   /* Average load across all groups in sd */

	/** Statistics of this group */
	unsigned long this_load;
	unsigned long this_load_per_task;
	unsigned long this_nr_running;
3496
	unsigned long this_has_capacity;
3497
	unsigned int  this_idle_cpus;
3498 3499

	/* Statistics of the busiest group */
3500
	unsigned int  busiest_idle_cpus;
3501 3502 3503
	unsigned long max_load;
	unsigned long busiest_load_per_task;
	unsigned long busiest_nr_running;
3504
	unsigned long busiest_group_capacity;
3505
	unsigned long busiest_has_capacity;
3506
	unsigned int  busiest_group_weight;
3507 3508 3509 3510 3511 3512 3513 3514 3515 3516 3517 3518 3519 3520 3521 3522 3523 3524 3525 3526 3527

	int group_imb; /* Is there imbalance in this sd */
#if defined(CONFIG_SCHED_MC) || defined(CONFIG_SCHED_SMT)
	int power_savings_balance; /* Is powersave balance needed for this sd */
	struct sched_group *group_min; /* Least loaded group in sd */
	struct sched_group *group_leader; /* Group which relieves group_min */
	unsigned long min_load_per_task; /* load_per_task in group_min */
	unsigned long leader_nr_running; /* Nr running of group_leader */
	unsigned long min_nr_running; /* Nr running of group_min */
#endif
};

/*
 * sg_lb_stats - stats of a sched_group required for load_balancing
 */
struct sg_lb_stats {
	unsigned long avg_load; /*Avg load across the CPUs of the group */
	unsigned long group_load; /* Total load over the CPUs of the group */
	unsigned long sum_nr_running; /* Nr tasks running in the group */
	unsigned long sum_weighted_load; /* Weighted load of group's tasks */
	unsigned long group_capacity;
3528 3529
	unsigned long idle_cpus;
	unsigned long group_weight;
3530
	int group_imb; /* Is there an imbalance in the group ? */
3531
	int group_has_capacity; /* Is there extra capacity in the group? */
3532 3533 3534 3535 3536 3537 3538 3539 3540 3541 3542 3543 3544 3545 3546 3547 3548 3549 3550 3551 3552 3553 3554 3555 3556 3557 3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574 3575 3576 3577 3578 3579 3580 3581 3582 3583 3584 3585 3586 3587 3588 3589 3590 3591 3592 3593 3594 3595 3596 3597 3598 3599 3600 3601 3602 3603 3604 3605 3606 3607 3608 3609 3610 3611 3612 3613 3614 3615 3616 3617 3618 3619 3620 3621 3622 3623 3624 3625 3626 3627 3628 3629 3630 3631 3632 3633 3634 3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 3645 3646 3647 3648 3649 3650 3651 3652 3653 3654 3655 3656 3657 3658 3659 3660 3661 3662 3663 3664 3665 3666 3667 3668 3669 3670 3671 3672 3673 3674 3675 3676 3677 3678 3679 3680 3681 3682 3683 3684 3685 3686 3687 3688 3689 3690 3691 3692 3693 3694 3695 3696 3697 3698 3699 3700 3701 3702 3703
};

/**
 * get_sd_load_idx - Obtain the load index for a given sched domain.
 * @sd: The sched_domain whose load_idx is to be obtained.
 * @idle: The Idle status of the CPU for whose sd load_icx is obtained.
 */
static inline int get_sd_load_idx(struct sched_domain *sd,
					enum cpu_idle_type idle)
{
	int load_idx;

	switch (idle) {
	case CPU_NOT_IDLE:
		load_idx = sd->busy_idx;
		break;

	case CPU_NEWLY_IDLE:
		load_idx = sd->newidle_idx;
		break;
	default:
		load_idx = sd->idle_idx;
		break;
	}

	return load_idx;
}


#if defined(CONFIG_SCHED_MC) || defined(CONFIG_SCHED_SMT)
/**
 * init_sd_power_savings_stats - Initialize power savings statistics for
 * the given sched_domain, during load balancing.
 *
 * @sd: Sched domain whose power-savings statistics are to be initialized.
 * @sds: Variable containing the statistics for sd.
 * @idle: Idle status of the CPU at which we're performing load-balancing.
 */
static inline void init_sd_power_savings_stats(struct sched_domain *sd,
	struct sd_lb_stats *sds, enum cpu_idle_type idle)
{
	/*
	 * Busy processors will not participate in power savings
	 * balance.
	 */
	if (idle == CPU_NOT_IDLE || !(sd->flags & SD_POWERSAVINGS_BALANCE))
		sds->power_savings_balance = 0;
	else {
		sds->power_savings_balance = 1;
		sds->min_nr_running = ULONG_MAX;
		sds->leader_nr_running = 0;
	}
}

/**
 * update_sd_power_savings_stats - Update the power saving stats for a
 * sched_domain while performing load balancing.
 *
 * @group: sched_group belonging to the sched_domain under consideration.
 * @sds: Variable containing the statistics of the sched_domain
 * @local_group: Does group contain the CPU for which we're performing
 * 		load balancing ?
 * @sgs: Variable containing the statistics of the group.
 */
static inline void update_sd_power_savings_stats(struct sched_group *group,
	struct sd_lb_stats *sds, int local_group, struct sg_lb_stats *sgs)
{

	if (!sds->power_savings_balance)
		return;

	/*
	 * If the local group is idle or completely loaded
	 * no need to do power savings balance at this domain
	 */
	if (local_group && (sds->this_nr_running >= sgs->group_capacity ||
				!sds->this_nr_running))
		sds->power_savings_balance = 0;

	/*
	 * If a group is already running at full capacity or idle,
	 * don't include that group in power savings calculations
	 */
	if (!sds->power_savings_balance ||
		sgs->sum_nr_running >= sgs->group_capacity ||
		!sgs->sum_nr_running)
		return;

	/*
	 * Calculate the group which has the least non-idle load.
	 * This is the group from where we need to pick up the load
	 * for saving power
	 */
	if ((sgs->sum_nr_running < sds->min_nr_running) ||
	    (sgs->sum_nr_running == sds->min_nr_running &&
	     group_first_cpu(group) > group_first_cpu(sds->group_min))) {
		sds->group_min = group;
		sds->min_nr_running = sgs->sum_nr_running;
		sds->min_load_per_task = sgs->sum_weighted_load /
						sgs->sum_nr_running;
	}

	/*
	 * Calculate the group which is almost near its
	 * capacity but still has some space to pick up some load
	 * from other group and save more power
	 */
	if (sgs->sum_nr_running + 1 > sgs->group_capacity)
		return;

	if (sgs->sum_nr_running > sds->leader_nr_running ||
	    (sgs->sum_nr_running == sds->leader_nr_running &&
	     group_first_cpu(group) < group_first_cpu(sds->group_leader))) {
		sds->group_leader = group;
		sds->leader_nr_running = sgs->sum_nr_running;
	}
}

/**
 * check_power_save_busiest_group - see if there is potential for some power-savings balance
 * @sds: Variable containing the statistics of the sched_domain
 *	under consideration.
 * @this_cpu: Cpu at which we're currently performing load-balancing.
 * @imbalance: Variable to store the imbalance.
 *
 * Description:
 * Check if we have potential to perform some power-savings balance.
 * If yes, set the busiest group to be the least loaded group in the
 * sched_domain, so that it's CPUs can be put to idle.
 *
 * Returns 1 if there is potential to perform power-savings balance.
 * Else returns 0.
 */
static inline int check_power_save_busiest_group(struct sd_lb_stats *sds,
					int this_cpu, unsigned long *imbalance)
{
	if (!sds->power_savings_balance)
		return 0;

	if (sds->this != sds->group_leader ||
			sds->group_leader == sds->group_min)
		return 0;

	*imbalance = sds->min_load_per_task;
	sds->busiest = sds->group_min;

	return 1;

}
#else /* CONFIG_SCHED_MC || CONFIG_SCHED_SMT */
static inline void init_sd_power_savings_stats(struct sched_domain *sd,
	struct sd_lb_stats *sds, enum cpu_idle_type idle)
{
	return;
}

static inline void update_sd_power_savings_stats(struct sched_group *group,
	struct sd_lb_stats *sds, int local_group, struct sg_lb_stats *sgs)
{
	return;
}

static inline int check_power_save_busiest_group(struct sd_lb_stats *sds,
					int this_cpu, unsigned long *imbalance)
{
	return 0;
}
#endif /* CONFIG_SCHED_MC || CONFIG_SCHED_SMT */


unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu)
{
3704
	return SCHED_POWER_SCALE;
3705 3706 3707 3708 3709 3710 3711 3712 3713
}

unsigned long __weak arch_scale_freq_power(struct sched_domain *sd, int cpu)
{
	return default_scale_freq_power(sd, cpu);
}

unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu)
{
3714
	unsigned long weight = sd->span_weight;
3715 3716 3717 3718 3719 3720 3721 3722 3723 3724 3725 3726 3727 3728 3729 3730 3731 3732
	unsigned long smt_gain = sd->smt_gain;

	smt_gain /= weight;

	return smt_gain;
}

unsigned long __weak arch_scale_smt_power(struct sched_domain *sd, int cpu)
{
	return default_scale_smt_power(sd, cpu);
}

unsigned long scale_rt_power(int cpu)
{
	struct rq *rq = cpu_rq(cpu);
	u64 total, available;

	total = sched_avg_period() + (rq->clock - rq->age_stamp);
3733 3734 3735 3736 3737 3738 3739

	if (unlikely(total < rq->rt_avg)) {
		/* Ensures that power won't end up being negative */
		available = 0;
	} else {
		available = total - rq->rt_avg;
	}
3740

3741 3742
	if (unlikely((s64)total < SCHED_POWER_SCALE))
		total = SCHED_POWER_SCALE;
3743

3744
	total >>= SCHED_POWER_SHIFT;
3745 3746 3747 3748 3749 3750

	return div_u64(available, total);
}

static void update_cpu_power(struct sched_domain *sd, int cpu)
{
3751
	unsigned long weight = sd->span_weight;
3752
	unsigned long power = SCHED_POWER_SCALE;
3753 3754 3755 3756 3757 3758 3759 3760
	struct sched_group *sdg = sd->groups;

	if ((sd->flags & SD_SHARE_CPUPOWER) && weight > 1) {
		if (sched_feat(ARCH_POWER))
			power *= arch_scale_smt_power(sd, cpu);
		else
			power *= default_scale_smt_power(sd, cpu);

3761
		power >>= SCHED_POWER_SHIFT;
3762 3763
	}

3764
	sdg->sgp->power_orig = power;
3765 3766 3767 3768 3769 3770

	if (sched_feat(ARCH_POWER))
		power *= arch_scale_freq_power(sd, cpu);
	else
		power *= default_scale_freq_power(sd, cpu);

3771
	power >>= SCHED_POWER_SHIFT;
3772

3773
	power *= scale_rt_power(cpu);
3774
	power >>= SCHED_POWER_SHIFT;
3775 3776 3777 3778

	if (!power)
		power = 1;

3779
	cpu_rq(cpu)->cpu_power = power;
3780
	sdg->sgp->power = power;
3781 3782
}

3783
void update_group_power(struct sched_domain *sd, int cpu)
3784 3785 3786 3787 3788 3789 3790 3791 3792 3793 3794 3795 3796 3797
{
	struct sched_domain *child = sd->child;
	struct sched_group *group, *sdg = sd->groups;
	unsigned long power;

	if (!child) {
		update_cpu_power(sd, cpu);
		return;
	}

	power = 0;

	group = child->groups;
	do {
3798
		power += group->sgp->power;
3799 3800 3801
		group = group->next;
	} while (group != child->groups);

3802
	sdg->sgp->power = power;
3803 3804
}

3805 3806 3807 3808 3809 3810 3811 3812 3813 3814 3815
/*
 * Try and fix up capacity for tiny siblings, this is needed when
 * things like SD_ASYM_PACKING need f_b_g to select another sibling
 * which on its own isn't powerful enough.
 *
 * See update_sd_pick_busiest() and check_asym_packing().
 */
static inline int
fix_small_capacity(struct sched_domain *sd, struct sched_group *group)
{
	/*
3816
	 * Only siblings can have significantly less than SCHED_POWER_SCALE
3817
	 */
P
Peter Zijlstra 已提交
3818
	if (!(sd->flags & SD_SHARE_CPUPOWER))
3819 3820 3821 3822 3823
		return 0;

	/*
	 * If ~90% of the cpu_power is still there, we're good.
	 */
3824
	if (group->sgp->power * 32 > group->sgp->power_orig * 29)
3825 3826 3827 3828 3829
		return 1;

	return 0;
}

3830 3831 3832 3833 3834 3835 3836 3837 3838 3839 3840 3841 3842 3843
/**
 * update_sg_lb_stats - Update sched_group's statistics for load balancing.
 * @sd: The sched_domain whose statistics are to be updated.
 * @group: sched_group whose statistics are to be updated.
 * @this_cpu: Cpu for which load balance is currently performed.
 * @idle: Idle status of this_cpu
 * @load_idx: Load index of sched_domain of this_cpu for load calc.
 * @local_group: Does group contain this_cpu.
 * @cpus: Set of cpus considered for load balancing.
 * @balance: Should we balance.
 * @sgs: variable to hold the statistics for this group.
 */
static inline void update_sg_lb_stats(struct sched_domain *sd,
			struct sched_group *group, int this_cpu,
3844
			enum cpu_idle_type idle, int load_idx,
3845 3846 3847
			int local_group, const struct cpumask *cpus,
			int *balance, struct sg_lb_stats *sgs)
{
3848
	unsigned long load, max_cpu_load, min_cpu_load, max_nr_running;
3849 3850
	int i;
	unsigned int balance_cpu = -1, first_idle_cpu = 0;
3851
	unsigned long avg_load_per_task = 0;
3852

3853
	if (local_group)
3854 3855 3856 3857 3858
		balance_cpu = group_first_cpu(group);

	/* Tally up the load of all CPUs in the group */
	max_cpu_load = 0;
	min_cpu_load = ~0UL;
3859
	max_nr_running = 0;
3860 3861 3862 3863 3864 3865 3866 3867 3868 3869 3870 3871 3872 3873

	for_each_cpu_and(i, sched_group_cpus(group), cpus) {
		struct rq *rq = cpu_rq(i);

		/* Bias balancing toward cpus of our domain */
		if (local_group) {
			if (idle_cpu(i) && !first_idle_cpu) {
				first_idle_cpu = 1;
				balance_cpu = i;
			}

			load = target_load(i, load_idx);
		} else {
			load = source_load(i, load_idx);
3874
			if (load > max_cpu_load) {
3875
				max_cpu_load = load;
3876 3877
				max_nr_running = rq->nr_running;
			}
3878 3879 3880 3881 3882 3883 3884
			if (min_cpu_load > load)
				min_cpu_load = load;
		}

		sgs->group_load += load;
		sgs->sum_nr_running += rq->nr_running;
		sgs->sum_weighted_load += weighted_cpuload(i);
3885 3886
		if (idle_cpu(i))
			sgs->idle_cpus++;
3887 3888 3889 3890 3891 3892 3893 3894
	}

	/*
	 * First idle cpu or the first cpu(busiest) in this sched group
	 * is eligible for doing load balancing at this and above
	 * domains. In the newly idle case, we will allow all the cpu's
	 * to do the newly idle load balance.
	 */
3895 3896 3897 3898 3899 3900
	if (idle != CPU_NEWLY_IDLE && local_group) {
		if (balance_cpu != this_cpu) {
			*balance = 0;
			return;
		}
		update_group_power(sd, this_cpu);
3901 3902 3903
	}

	/* Adjust by relative CPU power of the group */
3904
	sgs->avg_load = (sgs->group_load*SCHED_POWER_SCALE) / group->sgp->power;
3905 3906 3907

	/*
	 * Consider the group unbalanced when the imbalance is larger
P
Peter Zijlstra 已提交
3908
	 * than the average weight of a task.
3909 3910 3911 3912 3913 3914
	 *
	 * APZ: with cgroup the avg task weight can vary wildly and
	 *      might not be a suitable number - should we keep a
	 *      normalized nr_running number somewhere that negates
	 *      the hierarchy?
	 */
3915 3916
	if (sgs->sum_nr_running)
		avg_load_per_task = sgs->sum_weighted_load / sgs->sum_nr_running;
3917

P
Peter Zijlstra 已提交
3918
	if ((max_cpu_load - min_cpu_load) >= avg_load_per_task && max_nr_running > 1)
3919 3920
		sgs->group_imb = 1;

3921
	sgs->group_capacity = DIV_ROUND_CLOSEST(group->sgp->power,
3922
						SCHED_POWER_SCALE);
3923 3924
	if (!sgs->group_capacity)
		sgs->group_capacity = fix_small_capacity(sd, group);
3925
	sgs->group_weight = group->group_weight;
3926 3927 3928

	if (sgs->group_capacity > sgs->sum_nr_running)
		sgs->group_has_capacity = 1;
3929 3930
}

3931 3932 3933 3934 3935
/**
 * update_sd_pick_busiest - return 1 on busiest group
 * @sd: sched_domain whose statistics are to be checked
 * @sds: sched_domain statistics
 * @sg: sched_group candidate to be checked for being the busiest
3936 3937
 * @sgs: sched_group statistics
 * @this_cpu: the current cpu
3938 3939 3940 3941 3942 3943 3944 3945 3946 3947 3948 3949 3950 3951 3952 3953 3954 3955 3956 3957 3958 3959 3960 3961 3962 3963 3964 3965 3966 3967 3968 3969 3970 3971 3972 3973
 *
 * Determine if @sg is a busier group than the previously selected
 * busiest group.
 */
static bool update_sd_pick_busiest(struct sched_domain *sd,
				   struct sd_lb_stats *sds,
				   struct sched_group *sg,
				   struct sg_lb_stats *sgs,
				   int this_cpu)
{
	if (sgs->avg_load <= sds->max_load)
		return false;

	if (sgs->sum_nr_running > sgs->group_capacity)
		return true;

	if (sgs->group_imb)
		return true;

	/*
	 * ASYM_PACKING needs to move all the work to the lowest
	 * numbered CPUs in the group, therefore mark all groups
	 * higher than ourself as busy.
	 */
	if ((sd->flags & SD_ASYM_PACKING) && sgs->sum_nr_running &&
	    this_cpu < group_first_cpu(sg)) {
		if (!sds->busiest)
			return true;

		if (group_first_cpu(sds->busiest) > group_first_cpu(sg))
			return true;
	}

	return false;
}

3974
/**
3975
 * update_sd_lb_stats - Update sched_domain's statistics for load balancing.
3976 3977 3978 3979 3980 3981 3982 3983
 * @sd: sched_domain whose statistics are to be updated.
 * @this_cpu: Cpu for which load balance is currently performed.
 * @idle: Idle status of this_cpu
 * @cpus: Set of cpus considered for load balancing.
 * @balance: Should we balance.
 * @sds: variable to hold the statistics for this sched_domain.
 */
static inline void update_sd_lb_stats(struct sched_domain *sd, int this_cpu,
3984 3985
			enum cpu_idle_type idle, const struct cpumask *cpus,
			int *balance, struct sd_lb_stats *sds)
3986 3987
{
	struct sched_domain *child = sd->child;
3988
	struct sched_group *sg = sd->groups;
3989 3990 3991 3992 3993 3994 3995 3996 3997 3998 3999 4000
	struct sg_lb_stats sgs;
	int load_idx, prefer_sibling = 0;

	if (child && child->flags & SD_PREFER_SIBLING)
		prefer_sibling = 1;

	init_sd_power_savings_stats(sd, sds, idle);
	load_idx = get_sd_load_idx(sd, idle);

	do {
		int local_group;

4001
		local_group = cpumask_test_cpu(this_cpu, sched_group_cpus(sg));
4002
		memset(&sgs, 0, sizeof(sgs));
4003
		update_sg_lb_stats(sd, sg, this_cpu, idle, load_idx,
4004 4005
				local_group, cpus, balance, &sgs);

P
Peter Zijlstra 已提交
4006
		if (local_group && !(*balance))
4007 4008 4009
			return;

		sds->total_load += sgs.group_load;
4010
		sds->total_pwr += sg->sgp->power;
4011 4012 4013

		/*
		 * In case the child domain prefers tasks go to siblings
4014
		 * first, lower the sg capacity to one so that we'll try
4015 4016 4017 4018 4019 4020
		 * and move all the excess tasks away. We lower the capacity
		 * of a group only if the local group has the capacity to fit
		 * these excess tasks, i.e. nr_running < group_capacity. The
		 * extra check prevents the case where you always pull from the
		 * heaviest group when it is already under-utilized (possible
		 * with a large weight task outweighs the tasks on the system).
4021
		 */
4022
		if (prefer_sibling && !local_group && sds->this_has_capacity)
4023 4024 4025 4026
			sgs.group_capacity = min(sgs.group_capacity, 1UL);

		if (local_group) {
			sds->this_load = sgs.avg_load;
4027
			sds->this = sg;
4028 4029
			sds->this_nr_running = sgs.sum_nr_running;
			sds->this_load_per_task = sgs.sum_weighted_load;
4030
			sds->this_has_capacity = sgs.group_has_capacity;
4031
			sds->this_idle_cpus = sgs.idle_cpus;
4032
		} else if (update_sd_pick_busiest(sd, sds, sg, &sgs, this_cpu)) {
4033
			sds->max_load = sgs.avg_load;
4034
			sds->busiest = sg;
4035
			sds->busiest_nr_running = sgs.sum_nr_running;
4036
			sds->busiest_idle_cpus = sgs.idle_cpus;
4037
			sds->busiest_group_capacity = sgs.group_capacity;
4038
			sds->busiest_load_per_task = sgs.sum_weighted_load;
4039
			sds->busiest_has_capacity = sgs.group_has_capacity;
4040
			sds->busiest_group_weight = sgs.group_weight;
4041 4042 4043
			sds->group_imb = sgs.group_imb;
		}

4044 4045 4046 4047 4048 4049 4050 4051 4052 4053 4054 4055 4056 4057 4058 4059 4060 4061 4062 4063 4064 4065
		update_sd_power_savings_stats(sg, sds, local_group, &sgs);
		sg = sg->next;
	} while (sg != sd->groups);
}

/**
 * check_asym_packing - Check to see if the group is packed into the
 *			sched doman.
 *
 * This is primarily intended to used at the sibling level.  Some
 * cores like POWER7 prefer to use lower numbered SMT threads.  In the
 * case of POWER7, it can move to lower SMT modes only when higher
 * threads are idle.  When in lower SMT modes, the threads will
 * perform better since they share less core resources.  Hence when we
 * have idle threads, we want them to be the higher ones.
 *
 * This packing function is run on idle threads.  It checks to see if
 * the busiest CPU in this domain (core in the P7 case) has a higher
 * CPU number than the packing function is being run on.  Here we are
 * assuming lower CPU number will be equivalent to lower a SMT thread
 * number.
 *
4066 4067 4068
 * Returns 1 when packing is required and a task should be moved to
 * this CPU.  The amount of the imbalance is returned in *imbalance.
 *
4069 4070 4071 4072 4073 4074 4075 4076 4077 4078 4079 4080 4081 4082 4083 4084 4085 4086 4087 4088 4089
 * @sd: The sched_domain whose packing is to be checked.
 * @sds: Statistics of the sched_domain which is to be packed
 * @this_cpu: The cpu at whose sched_domain we're performing load-balance.
 * @imbalance: returns amount of imbalanced due to packing.
 */
static int check_asym_packing(struct sched_domain *sd,
			      struct sd_lb_stats *sds,
			      int this_cpu, unsigned long *imbalance)
{
	int busiest_cpu;

	if (!(sd->flags & SD_ASYM_PACKING))
		return 0;

	if (!sds->busiest)
		return 0;

	busiest_cpu = group_first_cpu(sds->busiest);
	if (this_cpu > busiest_cpu)
		return 0;

4090
	*imbalance = DIV_ROUND_CLOSEST(sds->max_load * sds->busiest->sgp->power,
4091
				       SCHED_POWER_SCALE);
4092
	return 1;
4093 4094 4095 4096 4097 4098 4099 4100 4101 4102 4103 4104 4105 4106 4107
}

/**
 * fix_small_imbalance - Calculate the minor imbalance that exists
 *			amongst the groups of a sched_domain, during
 *			load balancing.
 * @sds: Statistics of the sched_domain whose imbalance is to be calculated.
 * @this_cpu: The cpu at whose sched_domain we're performing load-balance.
 * @imbalance: Variable to store the imbalance.
 */
static inline void fix_small_imbalance(struct sd_lb_stats *sds,
				int this_cpu, unsigned long *imbalance)
{
	unsigned long tmp, pwr_now = 0, pwr_move = 0;
	unsigned int imbn = 2;
4108
	unsigned long scaled_busy_load_per_task;
4109 4110 4111 4112 4113 4114 4115 4116 4117 4118

	if (sds->this_nr_running) {
		sds->this_load_per_task /= sds->this_nr_running;
		if (sds->busiest_load_per_task >
				sds->this_load_per_task)
			imbn = 1;
	} else
		sds->this_load_per_task =
			cpu_avg_load_per_task(this_cpu);

4119
	scaled_busy_load_per_task = sds->busiest_load_per_task
4120
					 * SCHED_POWER_SCALE;
4121
	scaled_busy_load_per_task /= sds->busiest->sgp->power;
4122 4123 4124

	if (sds->max_load - sds->this_load + scaled_busy_load_per_task >=
			(scaled_busy_load_per_task * imbn)) {
4125 4126 4127 4128 4129 4130 4131 4132 4133 4134
		*imbalance = sds->busiest_load_per_task;
		return;
	}

	/*
	 * OK, we don't have enough imbalance to justify moving tasks,
	 * however we may be able to increase total CPU power used by
	 * moving them.
	 */

4135
	pwr_now += sds->busiest->sgp->power *
4136
			min(sds->busiest_load_per_task, sds->max_load);
4137
	pwr_now += sds->this->sgp->power *
4138
			min(sds->this_load_per_task, sds->this_load);
4139
	pwr_now /= SCHED_POWER_SCALE;
4140 4141

	/* Amount of load we'd subtract */
4142
	tmp = (sds->busiest_load_per_task * SCHED_POWER_SCALE) /
4143
		sds->busiest->sgp->power;
4144
	if (sds->max_load > tmp)
4145
		pwr_move += sds->busiest->sgp->power *
4146 4147 4148
			min(sds->busiest_load_per_task, sds->max_load - tmp);

	/* Amount of load we'd add */
4149
	if (sds->max_load * sds->busiest->sgp->power <
4150
		sds->busiest_load_per_task * SCHED_POWER_SCALE)
4151 4152
		tmp = (sds->max_load * sds->busiest->sgp->power) /
			sds->this->sgp->power;
4153
	else
4154
		tmp = (sds->busiest_load_per_task * SCHED_POWER_SCALE) /
4155 4156
			sds->this->sgp->power;
	pwr_move += sds->this->sgp->power *
4157
			min(sds->this_load_per_task, sds->this_load + tmp);
4158
	pwr_move /= SCHED_POWER_SCALE;
4159 4160 4161 4162 4163 4164 4165 4166 4167 4168 4169 4170 4171 4172 4173 4174

	/* Move if we gain throughput */
	if (pwr_move > pwr_now)
		*imbalance = sds->busiest_load_per_task;
}

/**
 * calculate_imbalance - Calculate the amount of imbalance present within the
 *			 groups of a given sched_domain during load balance.
 * @sds: statistics of the sched_domain whose imbalance is to be calculated.
 * @this_cpu: Cpu for which currently load balance is being performed.
 * @imbalance: The variable to store the imbalance.
 */
static inline void calculate_imbalance(struct sd_lb_stats *sds, int this_cpu,
		unsigned long *imbalance)
{
4175 4176 4177 4178 4179 4180 4181 4182
	unsigned long max_pull, load_above_capacity = ~0UL;

	sds->busiest_load_per_task /= sds->busiest_nr_running;
	if (sds->group_imb) {
		sds->busiest_load_per_task =
			min(sds->busiest_load_per_task, sds->avg_load);
	}

4183 4184 4185 4186 4187 4188 4189 4190 4191 4192
	/*
	 * In the presence of smp nice balancing, certain scenarios can have
	 * max load less than avg load(as we skip the groups at or below
	 * its cpu_power, while calculating max_load..)
	 */
	if (sds->max_load < sds->avg_load) {
		*imbalance = 0;
		return fix_small_imbalance(sds, this_cpu, imbalance);
	}

4193 4194 4195 4196 4197 4198 4199
	if (!sds->group_imb) {
		/*
		 * Don't want to pull so many tasks that a group would go idle.
		 */
		load_above_capacity = (sds->busiest_nr_running -
						sds->busiest_group_capacity);

4200
		load_above_capacity *= (SCHED_LOAD_SCALE * SCHED_POWER_SCALE);
4201

4202
		load_above_capacity /= sds->busiest->sgp->power;
4203 4204 4205 4206 4207 4208 4209 4210 4211 4212 4213 4214 4215
	}

	/*
	 * We're trying to get all the cpus to the average_load, so we don't
	 * want to push ourselves above the average load, nor do we wish to
	 * reduce the max loaded cpu below the average load. At the same time,
	 * we also don't want to reduce the group load below the group capacity
	 * (so that we can implement power-savings policies etc). Thus we look
	 * for the minimum possible imbalance.
	 * Be careful of negative numbers as they'll appear as very large values
	 * with unsigned longs.
	 */
	max_pull = min(sds->max_load - sds->avg_load, load_above_capacity);
4216 4217

	/* How much load to actually move to equalise the imbalance */
4218 4219
	*imbalance = min(max_pull * sds->busiest->sgp->power,
		(sds->avg_load - sds->this_load) * sds->this->sgp->power)
4220
			/ SCHED_POWER_SCALE;
4221 4222 4223

	/*
	 * if *imbalance is less than the average load per runnable task
L
Lucas De Marchi 已提交
4224
	 * there is no guarantee that any tasks will be moved so we'll have
4225 4226 4227 4228 4229 4230 4231
	 * a think about bumping its value to force at least one task to be
	 * moved
	 */
	if (*imbalance < sds->busiest_load_per_task)
		return fix_small_imbalance(sds, this_cpu, imbalance);

}
4232

4233 4234 4235 4236 4237 4238 4239 4240 4241 4242 4243 4244 4245 4246 4247 4248 4249 4250 4251 4252 4253 4254 4255 4256 4257 4258 4259 4260 4261
/******* find_busiest_group() helpers end here *********************/

/**
 * find_busiest_group - Returns the busiest group within the sched_domain
 * if there is an imbalance. If there isn't an imbalance, and
 * the user has opted for power-savings, it returns a group whose
 * CPUs can be put to idle by rebalancing those tasks elsewhere, if
 * such a group exists.
 *
 * Also calculates the amount of weighted load which should be moved
 * to restore balance.
 *
 * @sd: The sched_domain whose busiest group is to be returned.
 * @this_cpu: The cpu for which load balancing is currently being performed.
 * @imbalance: Variable which stores amount of weighted load which should
 *		be moved to restore balance/put a group to idle.
 * @idle: The idle status of this_cpu.
 * @cpus: The set of CPUs under consideration for load-balancing.
 * @balance: Pointer to a variable indicating if this_cpu
 *	is the appropriate cpu to perform load balancing at this_level.
 *
 * Returns:	- the busiest group if imbalance exists.
 *		- If no imbalance and user has opted for power-savings balance,
 *		   return the least loaded group whose CPUs can be
 *		   put to idle by rebalancing its tasks onto our group.
 */
static struct sched_group *
find_busiest_group(struct sched_domain *sd, int this_cpu,
		   unsigned long *imbalance, enum cpu_idle_type idle,
4262
		   const struct cpumask *cpus, int *balance)
4263 4264 4265 4266 4267 4268 4269 4270 4271
{
	struct sd_lb_stats sds;

	memset(&sds, 0, sizeof(sds));

	/*
	 * Compute the various statistics relavent for load balancing at
	 * this level.
	 */
4272
	update_sd_lb_stats(sd, this_cpu, idle, cpus, balance, &sds);
4273

4274 4275 4276
	/*
	 * this_cpu is not the appropriate cpu to perform load balancing at
	 * this level.
4277
	 */
P
Peter Zijlstra 已提交
4278
	if (!(*balance))
4279 4280
		goto ret;

4281 4282 4283 4284
	if ((idle == CPU_IDLE || idle == CPU_NEWLY_IDLE) &&
	    check_asym_packing(sd, &sds, this_cpu, imbalance))
		return sds.busiest;

4285
	/* There is no busy sibling group to pull tasks from */
4286 4287 4288
	if (!sds.busiest || sds.busiest_nr_running == 0)
		goto out_balanced;

4289
	sds.avg_load = (SCHED_POWER_SCALE * sds.total_load) / sds.total_pwr;
4290

P
Peter Zijlstra 已提交
4291 4292 4293 4294 4295 4296 4297 4298
	/*
	 * If the busiest group is imbalanced the below checks don't
	 * work because they assumes all things are equal, which typically
	 * isn't true due to cpus_allowed constraints and the like.
	 */
	if (sds.group_imb)
		goto force_balance;

4299
	/* SD_BALANCE_NEWIDLE trumps SMP nice when underutilized */
4300 4301 4302 4303
	if (idle == CPU_NEWLY_IDLE && sds.this_has_capacity &&
			!sds.busiest_has_capacity)
		goto force_balance;

4304 4305 4306 4307
	/*
	 * If the local group is more busy than the selected busiest group
	 * don't try and pull any tasks.
	 */
4308 4309 4310
	if (sds.this_load >= sds.max_load)
		goto out_balanced;

4311 4312 4313 4314
	/*
	 * Don't pull any tasks if this group is already above the domain
	 * average load.
	 */
4315 4316 4317
	if (sds.this_load >= sds.avg_load)
		goto out_balanced;

4318
	if (idle == CPU_IDLE) {
4319 4320 4321 4322 4323 4324
		/*
		 * This cpu is idle. If the busiest group load doesn't
		 * have more tasks than the number of available cpu's and
		 * there is no imbalance between this and busiest group
		 * wrt to idle cpu's, it is balanced.
		 */
4325
		if ((sds.this_idle_cpus <= sds.busiest_idle_cpus + 1) &&
4326 4327
		    sds.busiest_nr_running <= sds.busiest_group_weight)
			goto out_balanced;
4328 4329 4330 4331 4332 4333 4334
	} else {
		/*
		 * In the CPU_NEWLY_IDLE, CPU_NOT_IDLE cases, use
		 * imbalance_pct to be conservative.
		 */
		if (100 * sds.max_load <= sd->imbalance_pct * sds.this_load)
			goto out_balanced;
4335
	}
4336

4337
force_balance:
4338 4339 4340 4341 4342 4343 4344 4345 4346 4347 4348 4349 4350 4351 4352 4353 4354 4355 4356 4357
	/* Looks like there is an imbalance. Compute it */
	calculate_imbalance(&sds, this_cpu, imbalance);
	return sds.busiest;

out_balanced:
	/*
	 * There is no obvious imbalance. But check if we can do some balancing
	 * to save power.
	 */
	if (check_power_save_busiest_group(&sds, this_cpu, imbalance))
		return sds.busiest;
ret:
	*imbalance = 0;
	return NULL;
}

/*
 * find_busiest_queue - find the busiest runqueue among the cpus in group.
 */
static struct rq *
4358 4359 4360
find_busiest_queue(struct sched_domain *sd, struct sched_group *group,
		   enum cpu_idle_type idle, unsigned long imbalance,
		   const struct cpumask *cpus)
4361 4362 4363 4364 4365 4366 4367
{
	struct rq *busiest = NULL, *rq;
	unsigned long max_load = 0;
	int i;

	for_each_cpu(i, sched_group_cpus(group)) {
		unsigned long power = power_of(i);
4368 4369
		unsigned long capacity = DIV_ROUND_CLOSEST(power,
							   SCHED_POWER_SCALE);
4370 4371
		unsigned long wl;

4372 4373 4374
		if (!capacity)
			capacity = fix_small_capacity(sd, group);

4375 4376 4377 4378
		if (!cpumask_test_cpu(i, cpus))
			continue;

		rq = cpu_rq(i);
4379
		wl = weighted_cpuload(i);
4380

4381 4382 4383 4384
		/*
		 * When comparing with imbalance, use weighted_cpuload()
		 * which is not scaled with the cpu power.
		 */
4385 4386 4387
		if (capacity && rq->nr_running == 1 && wl > imbalance)
			continue;

4388 4389 4390 4391 4392 4393
		/*
		 * For the load comparisons with the other cpu's, consider
		 * the weighted_cpuload() scaled with the cpu power, so that
		 * the load can be moved away from the cpu that is potentially
		 * running at a lower capacity.
		 */
4394
		wl = (wl * SCHED_POWER_SCALE) / power;
4395

4396 4397 4398 4399 4400 4401 4402 4403 4404 4405 4406 4407 4408 4409 4410 4411
		if (wl > max_load) {
			max_load = wl;
			busiest = rq;
		}
	}

	return busiest;
}

/*
 * Max backoff if we encounter pinned tasks. Pretty arbitrary value, but
 * so long as it is large enough.
 */
#define MAX_PINNED_INTERVAL	512

/* Working cpumask for load_balance and load_balance_newidle. */
4412
DEFINE_PER_CPU(cpumask_var_t, load_balance_tmpmask);
4413

4414
static int need_active_balance(struct sched_domain *sd, int idle,
4415
			       int busiest_cpu, int this_cpu)
4416 4417
{
	if (idle == CPU_NEWLY_IDLE) {
4418 4419 4420 4421 4422 4423 4424 4425 4426

		/*
		 * ASYM_PACKING needs to force migrate tasks from busy but
		 * higher numbered CPUs in order to pack all tasks in the
		 * lowest numbered CPUs.
		 */
		if ((sd->flags & SD_ASYM_PACKING) && busiest_cpu > this_cpu)
			return 1;

4427 4428 4429 4430 4431 4432 4433 4434 4435 4436 4437 4438 4439 4440 4441 4442 4443 4444 4445 4446 4447 4448 4449 4450 4451 4452
		/*
		 * The only task running in a non-idle cpu can be moved to this
		 * cpu in an attempt to completely freeup the other CPU
		 * package.
		 *
		 * The package power saving logic comes from
		 * find_busiest_group(). If there are no imbalance, then
		 * f_b_g() will return NULL. However when sched_mc={1,2} then
		 * f_b_g() will select a group from which a running task may be
		 * pulled to this cpu in order to make the other package idle.
		 * If there is no opportunity to make a package idle and if
		 * there are no imbalance, then f_b_g() will return NULL and no
		 * action will be taken in load_balance_newidle().
		 *
		 * Under normal task pull operation due to imbalance, there
		 * will be more than one task in the source run queue and
		 * move_tasks() will succeed.  ld_moved will be true and this
		 * active balance code will not be triggered.
		 */
		if (sched_mc_power_savings < POWERSAVINGS_BALANCE_WAKEUP)
			return 0;
	}

	return unlikely(sd->nr_balance_failed > sd->cache_nice_tries+2);
}

4453 4454
static int active_load_balance_cpu_stop(void *data);

4455 4456 4457 4458 4459 4460 4461 4462
/*
 * Check this_cpu to ensure it is balanced within domain. Attempt to move
 * tasks if there is an imbalance.
 */
static int load_balance(int this_cpu, struct rq *this_rq,
			struct sched_domain *sd, enum cpu_idle_type idle,
			int *balance)
{
4463
	int ld_moved, all_pinned = 0, active_balance = 0;
4464 4465 4466 4467 4468 4469 4470 4471 4472 4473 4474
	struct sched_group *group;
	unsigned long imbalance;
	struct rq *busiest;
	unsigned long flags;
	struct cpumask *cpus = __get_cpu_var(load_balance_tmpmask);

	cpumask_copy(cpus, cpu_active_mask);

	schedstat_inc(sd, lb_count[idle]);

redo:
4475
	group = find_busiest_group(sd, this_cpu, &imbalance, idle,
4476 4477 4478 4479 4480 4481 4482 4483 4484 4485
				   cpus, balance);

	if (*balance == 0)
		goto out_balanced;

	if (!group) {
		schedstat_inc(sd, lb_nobusyg[idle]);
		goto out_balanced;
	}

4486
	busiest = find_busiest_queue(sd, group, idle, imbalance, cpus);
4487 4488 4489 4490 4491 4492 4493 4494 4495 4496 4497 4498 4499 4500 4501 4502 4503
	if (!busiest) {
		schedstat_inc(sd, lb_nobusyq[idle]);
		goto out_balanced;
	}

	BUG_ON(busiest == this_rq);

	schedstat_add(sd, lb_imbalance[idle], imbalance);

	ld_moved = 0;
	if (busiest->nr_running > 1) {
		/*
		 * Attempt to move tasks. If find_busiest_group has found
		 * an imbalance but busiest->nr_running <= 1, the group is
		 * still unbalanced. ld_moved simply stays zero, so it is
		 * correctly treated as an imbalance.
		 */
K
Ken Chen 已提交
4504
		all_pinned = 1;
4505 4506 4507 4508 4509 4510 4511 4512 4513 4514 4515 4516 4517 4518 4519 4520 4521 4522 4523 4524 4525 4526 4527 4528
		local_irq_save(flags);
		double_rq_lock(this_rq, busiest);
		ld_moved = move_tasks(this_rq, this_cpu, busiest,
				      imbalance, sd, idle, &all_pinned);
		double_rq_unlock(this_rq, busiest);
		local_irq_restore(flags);

		/*
		 * some other cpu did the load balance for us.
		 */
		if (ld_moved && this_cpu != smp_processor_id())
			resched_cpu(this_cpu);

		/* All tasks on this runqueue were pinned by CPU affinity */
		if (unlikely(all_pinned)) {
			cpumask_clear_cpu(cpu_of(busiest), cpus);
			if (!cpumask_empty(cpus))
				goto redo;
			goto out_balanced;
		}
	}

	if (!ld_moved) {
		schedstat_inc(sd, lb_failed[idle]);
4529 4530 4531 4532 4533 4534 4535 4536
		/*
		 * Increment the failure counter only on periodic balance.
		 * We do not want newidle balance, which can be very
		 * frequent, pollute the failure counter causing
		 * excessive cache_hot migrations and active balances.
		 */
		if (idle != CPU_NEWLY_IDLE)
			sd->nr_balance_failed++;
4537

4538
		if (need_active_balance(sd, idle, cpu_of(busiest), this_cpu)) {
4539 4540
			raw_spin_lock_irqsave(&busiest->lock, flags);

4541 4542 4543
			/* don't kick the active_load_balance_cpu_stop,
			 * if the curr task on busiest cpu can't be
			 * moved to this_cpu
4544 4545
			 */
			if (!cpumask_test_cpu(this_cpu,
4546
					tsk_cpus_allowed(busiest->curr))) {
4547 4548 4549 4550 4551 4552
				raw_spin_unlock_irqrestore(&busiest->lock,
							    flags);
				all_pinned = 1;
				goto out_one_pinned;
			}

4553 4554 4555 4556 4557
			/*
			 * ->active_balance synchronizes accesses to
			 * ->active_balance_work.  Once set, it's cleared
			 * only after active load balance is finished.
			 */
4558 4559 4560 4561 4562 4563
			if (!busiest->active_balance) {
				busiest->active_balance = 1;
				busiest->push_cpu = this_cpu;
				active_balance = 1;
			}
			raw_spin_unlock_irqrestore(&busiest->lock, flags);
4564

4565
			if (active_balance)
4566 4567 4568
				stop_one_cpu_nowait(cpu_of(busiest),
					active_load_balance_cpu_stop, busiest,
					&busiest->active_balance_work);
4569 4570 4571 4572 4573 4574 4575 4576 4577 4578 4579 4580 4581 4582 4583 4584 4585 4586 4587 4588 4589 4590 4591 4592 4593 4594 4595 4596 4597 4598 4599 4600 4601 4602 4603 4604 4605

			/*
			 * We've kicked active balancing, reset the failure
			 * counter.
			 */
			sd->nr_balance_failed = sd->cache_nice_tries+1;
		}
	} else
		sd->nr_balance_failed = 0;

	if (likely(!active_balance)) {
		/* We were unbalanced, so reset the balancing interval */
		sd->balance_interval = sd->min_interval;
	} else {
		/*
		 * If we've begun active balancing, start to back off. This
		 * case may not be covered by the all_pinned logic if there
		 * is only 1 task on the busy runqueue (because we don't call
		 * move_tasks).
		 */
		if (sd->balance_interval < sd->max_interval)
			sd->balance_interval *= 2;
	}

	goto out;

out_balanced:
	schedstat_inc(sd, lb_balanced[idle]);

	sd->nr_balance_failed = 0;

out_one_pinned:
	/* tune up the balancing interval */
	if ((all_pinned && sd->balance_interval < MAX_PINNED_INTERVAL) ||
			(sd->balance_interval < sd->max_interval))
		sd->balance_interval *= 2;

4606
	ld_moved = 0;
4607 4608 4609 4610 4611 4612 4613 4614
out:
	return ld_moved;
}

/*
 * idle_balance is called by schedule() if this_cpu is about to become
 * idle. Attempts to pull tasks from other CPUs.
 */
4615
void idle_balance(int this_cpu, struct rq *this_rq)
4616 4617 4618 4619 4620 4621 4622 4623 4624 4625
{
	struct sched_domain *sd;
	int pulled_task = 0;
	unsigned long next_balance = jiffies + HZ;

	this_rq->idle_stamp = this_rq->clock;

	if (this_rq->avg_idle < sysctl_sched_migration_cost)
		return;

4626 4627 4628 4629 4630
	/*
	 * Drop the rq->lock, but keep IRQ/preempt disabled.
	 */
	raw_spin_unlock(&this_rq->lock);

P
Paul Turner 已提交
4631
	update_shares(this_cpu);
4632
	rcu_read_lock();
4633 4634
	for_each_domain(this_cpu, sd) {
		unsigned long interval;
4635
		int balance = 1;
4636 4637 4638 4639

		if (!(sd->flags & SD_LOAD_BALANCE))
			continue;

4640
		if (sd->flags & SD_BALANCE_NEWIDLE) {
4641
			/* If we've pulled tasks over stop searching: */
4642 4643 4644
			pulled_task = load_balance(this_cpu, this_rq,
						   sd, CPU_NEWLY_IDLE, &balance);
		}
4645 4646 4647 4648

		interval = msecs_to_jiffies(sd->balance_interval);
		if (time_after(next_balance, sd->last_balance + interval))
			next_balance = sd->last_balance + interval;
N
Nikhil Rao 已提交
4649 4650
		if (pulled_task) {
			this_rq->idle_stamp = 0;
4651
			break;
N
Nikhil Rao 已提交
4652
		}
4653
	}
4654
	rcu_read_unlock();
4655 4656 4657

	raw_spin_lock(&this_rq->lock);

4658 4659 4660 4661 4662 4663 4664 4665 4666 4667
	if (pulled_task || time_after(jiffies, this_rq->next_balance)) {
		/*
		 * We are going idle. next_balance may be set based on
		 * a busy processor. So reset next_balance.
		 */
		this_rq->next_balance = next_balance;
	}
}

/*
4668 4669 4670 4671
 * active_load_balance_cpu_stop is run by cpu stopper. It pushes
 * running tasks off the busiest CPU onto idle CPUs. It requires at
 * least 1 task to be running on each physical CPU where possible, and
 * avoids physical / logical imbalances.
4672
 */
4673
static int active_load_balance_cpu_stop(void *data)
4674
{
4675 4676
	struct rq *busiest_rq = data;
	int busiest_cpu = cpu_of(busiest_rq);
4677
	int target_cpu = busiest_rq->push_cpu;
4678
	struct rq *target_rq = cpu_rq(target_cpu);
4679
	struct sched_domain *sd;
4680 4681 4682 4683 4684 4685 4686

	raw_spin_lock_irq(&busiest_rq->lock);

	/* make sure the requested cpu hasn't gone down in the meantime */
	if (unlikely(busiest_cpu != smp_processor_id() ||
		     !busiest_rq->active_balance))
		goto out_unlock;
4687 4688 4689

	/* Is there any task to move? */
	if (busiest_rq->nr_running <= 1)
4690
		goto out_unlock;
4691 4692 4693 4694 4695 4696 4697 4698 4699 4700 4701 4702

	/*
	 * This condition is "impossible", if it occurs
	 * we need to fix it. Originally reported by
	 * Bjorn Helgaas on a 128-cpu setup.
	 */
	BUG_ON(busiest_rq == target_rq);

	/* move a task from busiest_rq to target_rq */
	double_lock_balance(busiest_rq, target_rq);

	/* Search for an sd spanning us and the target CPU. */
4703
	rcu_read_lock();
4704 4705 4706 4707 4708 4709 4710 4711 4712 4713 4714 4715 4716 4717 4718
	for_each_domain(target_cpu, sd) {
		if ((sd->flags & SD_LOAD_BALANCE) &&
		    cpumask_test_cpu(busiest_cpu, sched_domain_span(sd)))
				break;
	}

	if (likely(sd)) {
		schedstat_inc(sd, alb_count);

		if (move_one_task(target_rq, target_cpu, busiest_rq,
				  sd, CPU_IDLE))
			schedstat_inc(sd, alb_pushed);
		else
			schedstat_inc(sd, alb_failed);
	}
4719
	rcu_read_unlock();
4720
	double_unlock_balance(busiest_rq, target_rq);
4721 4722 4723 4724
out_unlock:
	busiest_rq->active_balance = 0;
	raw_spin_unlock_irq(&busiest_rq->lock);
	return 0;
4725 4726 4727
}

#ifdef CONFIG_NO_HZ
4728 4729 4730 4731 4732 4733 4734 4735 4736 4737
/*
 * idle load balancing details
 * - One of the idle CPUs nominates itself as idle load_balancer, while
 *   entering idle.
 * - This idle load balancer CPU will also go into tickless mode when
 *   it is idle, just like all other idle CPUs
 * - When one of the busy CPUs notice that there may be an idle rebalancing
 *   needed, they will kick the idle load balancer, which then does idle
 *   load balancing for all the idle CPUs.
 */
4738 4739
static struct {
	atomic_t load_balancer;
4740 4741 4742 4743 4744 4745
	atomic_t first_pick_cpu;
	atomic_t second_pick_cpu;
	cpumask_var_t idle_cpus_mask;
	cpumask_var_t grp_idle_mask;
	unsigned long next_balance;     /* in jiffy units */
} nohz ____cacheline_aligned;
4746 4747 4748 4749 4750 4751 4752 4753 4754 4755 4756 4757 4758 4759 4760 4761 4762 4763 4764 4765 4766

int get_nohz_load_balancer(void)
{
	return atomic_read(&nohz.load_balancer);
}

#if defined(CONFIG_SCHED_MC) || defined(CONFIG_SCHED_SMT)
/**
 * lowest_flag_domain - Return lowest sched_domain containing flag.
 * @cpu:	The cpu whose lowest level of sched domain is to
 *		be returned.
 * @flag:	The flag to check for the lowest sched_domain
 *		for the given cpu.
 *
 * Returns the lowest sched_domain of a cpu which contains the given flag.
 */
static inline struct sched_domain *lowest_flag_domain(int cpu, int flag)
{
	struct sched_domain *sd;

	for_each_domain(cpu, sd)
4767
		if (sd->flags & flag)
4768 4769 4770 4771 4772 4773 4774 4775 4776 4777 4778 4779 4780 4781 4782 4783 4784 4785 4786 4787 4788 4789 4790 4791 4792 4793 4794 4795 4796 4797 4798
			break;

	return sd;
}

/**
 * for_each_flag_domain - Iterates over sched_domains containing the flag.
 * @cpu:	The cpu whose domains we're iterating over.
 * @sd:		variable holding the value of the power_savings_sd
 *		for cpu.
 * @flag:	The flag to filter the sched_domains to be iterated.
 *
 * Iterates over all the scheduler domains for a given cpu that has the 'flag'
 * set, starting from the lowest sched_domain to the highest.
 */
#define for_each_flag_domain(cpu, sd, flag) \
	for (sd = lowest_flag_domain(cpu, flag); \
		(sd && (sd->flags & flag)); sd = sd->parent)

/**
 * is_semi_idle_group - Checks if the given sched_group is semi-idle.
 * @ilb_group:	group to be checked for semi-idleness
 *
 * Returns:	1 if the group is semi-idle. 0 otherwise.
 *
 * We define a sched_group to be semi idle if it has atleast one idle-CPU
 * and atleast one non-idle CPU. This helper function checks if the given
 * sched_group is semi-idle or not.
 */
static inline int is_semi_idle_group(struct sched_group *ilb_group)
{
4799
	cpumask_and(nohz.grp_idle_mask, nohz.idle_cpus_mask,
4800 4801 4802 4803 4804 4805
					sched_group_cpus(ilb_group));

	/*
	 * A sched_group is semi-idle when it has atleast one busy cpu
	 * and atleast one idle cpu.
	 */
4806
	if (cpumask_empty(nohz.grp_idle_mask))
4807 4808
		return 0;

4809
	if (cpumask_equal(nohz.grp_idle_mask, sched_group_cpus(ilb_group)))
4810 4811 4812 4813 4814 4815 4816 4817 4818 4819 4820 4821 4822 4823 4824 4825 4826 4827 4828 4829
		return 0;

	return 1;
}
/**
 * find_new_ilb - Finds the optimum idle load balancer for nomination.
 * @cpu:	The cpu which is nominating a new idle_load_balancer.
 *
 * Returns:	Returns the id of the idle load balancer if it exists,
 *		Else, returns >= nr_cpu_ids.
 *
 * This algorithm picks the idle load balancer such that it belongs to a
 * semi-idle powersavings sched_domain. The idea is to try and avoid
 * completely idle packages/cores just for the purpose of idle load balancing
 * when there are other idle cpu's which are better suited for that job.
 */
static int find_new_ilb(int cpu)
{
	struct sched_domain *sd;
	struct sched_group *ilb_group;
4830
	int ilb = nr_cpu_ids;
4831 4832 4833 4834 4835 4836 4837 4838 4839 4840 4841 4842

	/*
	 * Have idle load balancer selection from semi-idle packages only
	 * when power-aware load balancing is enabled
	 */
	if (!(sched_smt_power_savings || sched_mc_power_savings))
		goto out_done;

	/*
	 * Optimize for the case when we have no idle CPUs or only one
	 * idle CPU. Don't walk the sched_domain hierarchy in such cases
	 */
4843
	if (cpumask_weight(nohz.idle_cpus_mask) < 2)
4844 4845
		goto out_done;

4846
	rcu_read_lock();
4847 4848 4849 4850
	for_each_flag_domain(cpu, sd, SD_POWERSAVINGS_BALANCE) {
		ilb_group = sd->groups;

		do {
4851 4852 4853 4854
			if (is_semi_idle_group(ilb_group)) {
				ilb = cpumask_first(nohz.grp_idle_mask);
				goto unlock;
			}
4855 4856 4857 4858 4859

			ilb_group = ilb_group->next;

		} while (ilb_group != sd->groups);
	}
4860 4861
unlock:
	rcu_read_unlock();
4862 4863

out_done:
4864
	return ilb;
4865 4866 4867 4868
}
#else /*  (CONFIG_SCHED_MC || CONFIG_SCHED_SMT) */
static inline int find_new_ilb(int call_cpu)
{
4869
	return nr_cpu_ids;
4870 4871 4872
}
#endif

4873 4874 4875 4876 4877 4878 4879 4880 4881 4882 4883 4884 4885 4886 4887 4888 4889 4890 4891
/*
 * Kick a CPU to do the nohz balancing, if it is time for it. We pick the
 * nohz_load_balancer CPU (if there is one) otherwise fallback to any idle
 * CPU (if there is one).
 */
static void nohz_balancer_kick(int cpu)
{
	int ilb_cpu;

	nohz.next_balance++;

	ilb_cpu = get_nohz_load_balancer();

	if (ilb_cpu >= nr_cpu_ids) {
		ilb_cpu = cpumask_first(nohz.idle_cpus_mask);
		if (ilb_cpu >= nr_cpu_ids)
			return;
	}

4892 4893 4894 4895 4896 4897 4898 4899 4900
	if (test_and_set_bit(NOHZ_BALANCE_KICK, nohz_flags(cpu)))
		return;
	/*
	 * Use smp_send_reschedule() instead of resched_cpu().
	 * This way we generate a sched IPI on the target cpu which
	 * is idle. And the softirq performing nohz idle load balance
	 * will be run before returning from the IPI.
	 */
	smp_send_reschedule(ilb_cpu);
4901 4902 4903
	return;
}

4904 4905 4906
/*
 * This routine will try to nominate the ilb (idle load balancing)
 * owner among the cpus whose ticks are stopped. ilb owner will do the idle
4907
 * load balancing on behalf of all those cpus.
4908
 *
4909 4910 4911
 * When the ilb owner becomes busy, we will not have new ilb owner until some
 * idle CPU wakes up and goes back to idle or some busy CPU tries to kick
 * idle load balancing by kicking one of the idle CPUs.
4912
 *
4913 4914 4915
 * Ticks are stopped for the ilb owner as well, with busy CPU kicking this
 * ilb owner CPU in future (when there is a need for idle load balancing on
 * behalf of all idle CPUs).
4916
 */
4917
void select_nohz_load_balancer(int stop_tick)
4918 4919 4920 4921 4922 4923
{
	int cpu = smp_processor_id();

	if (stop_tick) {
		if (!cpu_active(cpu)) {
			if (atomic_read(&nohz.load_balancer) != cpu)
4924
				return;
4925 4926 4927 4928 4929

			/*
			 * If we are going offline and still the leader,
			 * give up!
			 */
4930 4931
			if (atomic_cmpxchg(&nohz.load_balancer, cpu,
					   nr_cpu_ids) != cpu)
4932 4933
				BUG();

4934
			return;
4935 4936
		}

4937
		cpumask_set_cpu(cpu, nohz.idle_cpus_mask);
4938

4939 4940 4941 4942
		if (atomic_read(&nohz.first_pick_cpu) == cpu)
			atomic_cmpxchg(&nohz.first_pick_cpu, cpu, nr_cpu_ids);
		if (atomic_read(&nohz.second_pick_cpu) == cpu)
			atomic_cmpxchg(&nohz.second_pick_cpu, cpu, nr_cpu_ids);
4943

4944
		if (atomic_read(&nohz.load_balancer) >= nr_cpu_ids) {
4945 4946
			int new_ilb;

4947 4948 4949 4950 4951
			/* make me the ilb owner */
			if (atomic_cmpxchg(&nohz.load_balancer, nr_cpu_ids,
					   cpu) != nr_cpu_ids)
				return;

4952 4953 4954 4955 4956 4957
			/*
			 * Check to see if there is a more power-efficient
			 * ilb.
			 */
			new_ilb = find_new_ilb(cpu);
			if (new_ilb < nr_cpu_ids && new_ilb != cpu) {
4958
				atomic_set(&nohz.load_balancer, nr_cpu_ids);
4959
				resched_cpu(new_ilb);
4960
				return;
4961
			}
4962
			return;
4963
		}
4964 4965

		set_bit(NOHZ_TICK_STOPPED, nohz_flags(cpu));
4966
	} else {
4967 4968
		if (!cpumask_test_cpu(cpu, nohz.idle_cpus_mask))
			return;
4969

4970
		cpumask_clear_cpu(cpu, nohz.idle_cpus_mask);
4971 4972

		if (atomic_read(&nohz.load_balancer) == cpu)
4973 4974
			if (atomic_cmpxchg(&nohz.load_balancer, cpu,
					   nr_cpu_ids) != cpu)
4975 4976
				BUG();
	}
4977
	return;
4978 4979 4980 4981 4982
}
#endif

static DEFINE_SPINLOCK(balancing);

4983 4984 4985 4986 4987 4988
static unsigned long __read_mostly max_load_balance_interval = HZ/10;

/*
 * Scale the max load_balance interval with the number of CPUs in the system.
 * This trades load-balance latency on larger machines for less cross talk.
 */
4989
void update_max_interval(void)
4990 4991 4992 4993
{
	max_load_balance_interval = HZ*num_online_cpus()/10;
}

4994 4995 4996 4997 4998 4999 5000 5001 5002 5003 5004 5005 5006 5007 5008 5009 5010
/*
 * It checks each scheduling domain to see if it is due to be balanced,
 * and initiates a balancing operation if so.
 *
 * Balancing parameters are set up in arch_init_sched_domains.
 */
static void rebalance_domains(int cpu, enum cpu_idle_type idle)
{
	int balance = 1;
	struct rq *rq = cpu_rq(cpu);
	unsigned long interval;
	struct sched_domain *sd;
	/* Earliest time when we have to do rebalance again */
	unsigned long next_balance = jiffies + 60*HZ;
	int update_next_balance = 0;
	int need_serialize;

P
Peter Zijlstra 已提交
5011 5012
	update_shares(cpu);

5013
	rcu_read_lock();
5014 5015 5016 5017 5018 5019 5020 5021 5022 5023
	for_each_domain(cpu, sd) {
		if (!(sd->flags & SD_LOAD_BALANCE))
			continue;

		interval = sd->balance_interval;
		if (idle != CPU_IDLE)
			interval *= sd->busy_factor;

		/* scale ms to jiffies */
		interval = msecs_to_jiffies(interval);
5024
		interval = clamp(interval, 1UL, max_load_balance_interval);
5025 5026 5027 5028 5029 5030 5031 5032 5033 5034 5035 5036

		need_serialize = sd->flags & SD_SERIALIZE;

		if (need_serialize) {
			if (!spin_trylock(&balancing))
				goto out;
		}

		if (time_after_eq(jiffies, sd->last_balance + interval)) {
			if (load_balance(cpu, rq, sd, idle, &balance)) {
				/*
				 * We've pulled tasks over so either we're no
5037
				 * longer idle.
5038 5039 5040 5041 5042 5043 5044 5045 5046 5047 5048 5049 5050 5051 5052 5053 5054 5055 5056 5057 5058
				 */
				idle = CPU_NOT_IDLE;
			}
			sd->last_balance = jiffies;
		}
		if (need_serialize)
			spin_unlock(&balancing);
out:
		if (time_after(next_balance, sd->last_balance + interval)) {
			next_balance = sd->last_balance + interval;
			update_next_balance = 1;
		}

		/*
		 * Stop the load balance at this level. There is another
		 * CPU in our sched group which is doing load balancing more
		 * actively.
		 */
		if (!balance)
			break;
	}
5059
	rcu_read_unlock();
5060 5061 5062 5063 5064 5065 5066 5067 5068 5069

	/*
	 * next_balance will be updated only when there is a need.
	 * When the cpu is attached to null domain for ex, it will not be
	 * updated.
	 */
	if (likely(update_next_balance))
		rq->next_balance = next_balance;
}

5070
#ifdef CONFIG_NO_HZ
5071
/*
5072
 * In CONFIG_NO_HZ case, the idle balance kickee will do the
5073 5074
 * rebalancing for all the cpus for whom scheduler ticks are stopped.
 */
5075 5076 5077 5078 5079 5080
static void nohz_idle_balance(int this_cpu, enum cpu_idle_type idle)
{
	struct rq *this_rq = cpu_rq(this_cpu);
	struct rq *rq;
	int balance_cpu;

5081 5082 5083
	if (idle != CPU_IDLE ||
	    !test_bit(NOHZ_BALANCE_KICK, nohz_flags(this_cpu)))
		goto end;
5084 5085 5086 5087 5088 5089 5090 5091 5092 5093

	for_each_cpu(balance_cpu, nohz.idle_cpus_mask) {
		if (balance_cpu == this_cpu)
			continue;

		/*
		 * If this cpu gets work to do, stop the load balancing
		 * work being done for other cpus. Next load
		 * balancing owner will pick it up.
		 */
5094
		if (need_resched())
5095 5096 5097
			break;

		raw_spin_lock_irq(&this_rq->lock);
5098
		update_rq_clock(this_rq);
5099 5100 5101 5102 5103 5104 5105 5106 5107 5108
		update_cpu_load(this_rq);
		raw_spin_unlock_irq(&this_rq->lock);

		rebalance_domains(balance_cpu, CPU_IDLE);

		rq = cpu_rq(balance_cpu);
		if (time_after(this_rq->next_balance, rq->next_balance))
			this_rq->next_balance = rq->next_balance;
	}
	nohz.next_balance = this_rq->next_balance;
5109 5110
end:
	clear_bit(NOHZ_BALANCE_KICK, nohz_flags(this_cpu));
5111 5112 5113 5114 5115 5116 5117 5118 5119 5120 5121 5122 5123 5124 5125 5126 5127 5128 5129 5130
}

/*
 * Current heuristic for kicking the idle load balancer
 * - first_pick_cpu is the one of the busy CPUs. It will kick
 *   idle load balancer when it has more than one process active. This
 *   eliminates the need for idle load balancing altogether when we have
 *   only one running process in the system (common case).
 * - If there are more than one busy CPU, idle load balancer may have
 *   to run for active_load_balance to happen (i.e., two busy CPUs are
 *   SMT or core siblings and can run better if they move to different
 *   physical CPUs). So, second_pick_cpu is the second of the busy CPUs
 *   which will kick idle load balancer as soon as it has any load.
 */
static inline int nohz_kick_needed(struct rq *rq, int cpu)
{
	unsigned long now = jiffies;
	int ret;
	int first_pick_cpu, second_pick_cpu;

5131
	if (unlikely(idle_cpu(cpu)))
5132 5133
		return 0;

5134 5135 5136 5137 5138 5139 5140 5141
       /*
	* We may be recently in ticked or tickless idle mode. At the first
	* busy tick after returning from idle, we will update the busy stats.
	*/
	if (unlikely(test_bit(NOHZ_TICK_STOPPED, nohz_flags(cpu))))
		clear_bit(NOHZ_TICK_STOPPED, nohz_flags(cpu));

	if (time_before(now, nohz.next_balance))
5142 5143 5144 5145 5146 5147 5148 5149 5150 5151 5152 5153 5154 5155 5156 5157 5158 5159 5160 5161 5162 5163 5164 5165 5166 5167 5168 5169 5170 5171 5172
		return 0;

	first_pick_cpu = atomic_read(&nohz.first_pick_cpu);
	second_pick_cpu = atomic_read(&nohz.second_pick_cpu);

	if (first_pick_cpu < nr_cpu_ids && first_pick_cpu != cpu &&
	    second_pick_cpu < nr_cpu_ids && second_pick_cpu != cpu)
		return 0;

	ret = atomic_cmpxchg(&nohz.first_pick_cpu, nr_cpu_ids, cpu);
	if (ret == nr_cpu_ids || ret == cpu) {
		atomic_cmpxchg(&nohz.second_pick_cpu, cpu, nr_cpu_ids);
		if (rq->nr_running > 1)
			return 1;
	} else {
		ret = atomic_cmpxchg(&nohz.second_pick_cpu, nr_cpu_ids, cpu);
		if (ret == nr_cpu_ids || ret == cpu) {
			if (rq->nr_running)
				return 1;
		}
	}
	return 0;
}
#else
static void nohz_idle_balance(int this_cpu, enum cpu_idle_type idle) { }
#endif

/*
 * run_rebalance_domains is triggered when needed from the scheduler tick.
 * Also triggered for nohz idle balancing (with nohz_balancing_kick set).
 */
5173 5174 5175 5176
static void run_rebalance_domains(struct softirq_action *h)
{
	int this_cpu = smp_processor_id();
	struct rq *this_rq = cpu_rq(this_cpu);
5177
	enum cpu_idle_type idle = this_rq->idle_balance ?
5178 5179 5180 5181 5182
						CPU_IDLE : CPU_NOT_IDLE;

	rebalance_domains(this_cpu, idle);

	/*
5183
	 * If this cpu has a pending nohz_balance_kick, then do the
5184 5185 5186
	 * balancing on behalf of the other idle cpus whose ticks are
	 * stopped.
	 */
5187
	nohz_idle_balance(this_cpu, idle);
5188 5189 5190 5191
}

static inline int on_null_domain(int cpu)
{
5192
	return !rcu_dereference_sched(cpu_rq(cpu)->sd);
5193 5194 5195 5196 5197
}

/*
 * Trigger the SCHED_SOFTIRQ if it is time to do periodic load balancing.
 */
5198
void trigger_load_balance(struct rq *rq, int cpu)
5199 5200 5201 5202 5203
{
	/* Don't need to rebalance while attached to NULL domain */
	if (time_after_eq(jiffies, rq->next_balance) &&
	    likely(!on_null_domain(cpu)))
		raise_softirq(SCHED_SOFTIRQ);
5204
#ifdef CONFIG_NO_HZ
5205
	if (nohz_kick_needed(rq, cpu) && likely(!on_null_domain(cpu)))
5206 5207
		nohz_balancer_kick(cpu);
#endif
5208 5209
}

5210 5211 5212 5213 5214 5215 5216 5217 5218 5219
static void rq_online_fair(struct rq *rq)
{
	update_sysctl();
}

static void rq_offline_fair(struct rq *rq)
{
	update_sysctl();
}

5220
#endif /* CONFIG_SMP */
5221

5222 5223 5224
/*
 * scheduler tick hitting a task of our scheduling class:
 */
P
Peter Zijlstra 已提交
5225
static void task_tick_fair(struct rq *rq, struct task_struct *curr, int queued)
5226 5227 5228 5229 5230 5231
{
	struct cfs_rq *cfs_rq;
	struct sched_entity *se = &curr->se;

	for_each_sched_entity(se) {
		cfs_rq = cfs_rq_of(se);
P
Peter Zijlstra 已提交
5232
		entity_tick(cfs_rq, se, queued);
5233 5234 5235 5236
	}
}

/*
P
Peter Zijlstra 已提交
5237 5238 5239
 * called on fork with the child task as argument from the parent's context
 *  - child not yet on the tasklist
 *  - preemption disabled
5240
 */
P
Peter Zijlstra 已提交
5241
static void task_fork_fair(struct task_struct *p)
5242
{
P
Peter Zijlstra 已提交
5243
	struct cfs_rq *cfs_rq = task_cfs_rq(current);
5244
	struct sched_entity *se = &p->se, *curr = cfs_rq->curr;
5245
	int this_cpu = smp_processor_id();
P
Peter Zijlstra 已提交
5246 5247 5248
	struct rq *rq = this_rq();
	unsigned long flags;

5249
	raw_spin_lock_irqsave(&rq->lock, flags);
5250

5251 5252
	update_rq_clock(rq);

5253 5254
	if (unlikely(task_cpu(p) != this_cpu)) {
		rcu_read_lock();
P
Peter Zijlstra 已提交
5255
		__set_task_cpu(p, this_cpu);
5256 5257
		rcu_read_unlock();
	}
5258

5259
	update_curr(cfs_rq);
P
Peter Zijlstra 已提交
5260

5261 5262
	if (curr)
		se->vruntime = curr->vruntime;
5263
	place_entity(cfs_rq, se, 1);
5264

P
Peter Zijlstra 已提交
5265
	if (sysctl_sched_child_runs_first && curr && entity_before(curr, se)) {
D
Dmitry Adamushko 已提交
5266
		/*
5267 5268 5269
		 * Upon rescheduling, sched_class::put_prev_task() will place
		 * 'current' within the tree based on its new key value.
		 */
5270
		swap(curr->vruntime, se->vruntime);
5271
		resched_task(rq->curr);
5272
	}
5273

5274 5275
	se->vruntime -= cfs_rq->min_vruntime;

5276
	raw_spin_unlock_irqrestore(&rq->lock, flags);
5277 5278
}

5279 5280 5281 5282
/*
 * Priority of the task has changed. Check to see if we preempt
 * the current task.
 */
P
Peter Zijlstra 已提交
5283 5284
static void
prio_changed_fair(struct rq *rq, struct task_struct *p, int oldprio)
5285
{
P
Peter Zijlstra 已提交
5286 5287 5288
	if (!p->se.on_rq)
		return;

5289 5290 5291 5292 5293
	/*
	 * Reschedule if we are currently running on this runqueue and
	 * our priority decreased, or if we are not currently running on
	 * this runqueue and our priority is higher than the current's
	 */
P
Peter Zijlstra 已提交
5294
	if (rq->curr == p) {
5295 5296 5297
		if (p->prio > oldprio)
			resched_task(rq->curr);
	} else
5298
		check_preempt_curr(rq, p, 0);
5299 5300
}

P
Peter Zijlstra 已提交
5301 5302 5303 5304 5305 5306 5307 5308 5309 5310 5311 5312 5313 5314 5315 5316 5317 5318 5319 5320 5321 5322 5323 5324
static void switched_from_fair(struct rq *rq, struct task_struct *p)
{
	struct sched_entity *se = &p->se;
	struct cfs_rq *cfs_rq = cfs_rq_of(se);

	/*
	 * Ensure the task's vruntime is normalized, so that when its
	 * switched back to the fair class the enqueue_entity(.flags=0) will
	 * do the right thing.
	 *
	 * If it was on_rq, then the dequeue_entity(.flags=0) will already
	 * have normalized the vruntime, if it was !on_rq, then only when
	 * the task is sleeping will it still have non-normalized vruntime.
	 */
	if (!se->on_rq && p->state != TASK_RUNNING) {
		/*
		 * Fix up our vruntime so that the current sleep doesn't
		 * cause 'unlimited' sleep bonus.
		 */
		place_entity(cfs_rq, se, 0);
		se->vruntime -= cfs_rq->min_vruntime;
	}
}

5325 5326 5327
/*
 * We switched to the sched_fair class.
 */
P
Peter Zijlstra 已提交
5328
static void switched_to_fair(struct rq *rq, struct task_struct *p)
5329
{
P
Peter Zijlstra 已提交
5330 5331 5332
	if (!p->se.on_rq)
		return;

5333 5334 5335 5336 5337
	/*
	 * We were most likely switched from sched_rt, so
	 * kick off the schedule if running, otherwise just see
	 * if we can still preempt the current task.
	 */
P
Peter Zijlstra 已提交
5338
	if (rq->curr == p)
5339 5340
		resched_task(rq->curr);
	else
5341
		check_preempt_curr(rq, p, 0);
5342 5343
}

5344 5345 5346 5347 5348 5349 5350 5351 5352
/* Account for a task changing its policy or group.
 *
 * This routine is mostly called to set cfs_rq->curr field when a task
 * migrates between groups/classes.
 */
static void set_curr_task_fair(struct rq *rq)
{
	struct sched_entity *se = &rq->curr->se;

5353 5354 5355 5356 5357 5358 5359
	for_each_sched_entity(se) {
		struct cfs_rq *cfs_rq = cfs_rq_of(se);

		set_next_entity(cfs_rq, se);
		/* ensure bandwidth has been allocated on our new cfs_rq */
		account_cfs_rq_runtime(cfs_rq, 0);
	}
5360 5361
}

5362 5363 5364 5365 5366 5367 5368 5369 5370 5371
void init_cfs_rq(struct cfs_rq *cfs_rq)
{
	cfs_rq->tasks_timeline = RB_ROOT;
	INIT_LIST_HEAD(&cfs_rq->tasks);
	cfs_rq->min_vruntime = (u64)(-(1LL << 20));
#ifndef CONFIG_64BIT
	cfs_rq->min_vruntime_copy = cfs_rq->min_vruntime;
#endif
}

P
Peter Zijlstra 已提交
5372
#ifdef CONFIG_FAIR_GROUP_SCHED
5373
static void task_move_group_fair(struct task_struct *p, int on_rq)
P
Peter Zijlstra 已提交
5374
{
5375 5376 5377 5378 5379 5380 5381 5382 5383 5384 5385 5386 5387 5388 5389 5390
	/*
	 * If the task was not on the rq at the time of this cgroup movement
	 * it must have been asleep, sleeping tasks keep their ->vruntime
	 * absolute on their old rq until wakeup (needed for the fair sleeper
	 * bonus in place_entity()).
	 *
	 * If it was on the rq, we've just 'preempted' it, which does convert
	 * ->vruntime to a relative base.
	 *
	 * Make sure both cases convert their relative position when migrating
	 * to another cgroup's rq. This does somewhat interfere with the
	 * fair sleeper stuff for the first placement, but who cares.
	 */
	if (!on_rq)
		p->se.vruntime -= cfs_rq_of(&p->se)->min_vruntime;
	set_task_rq(p, task_cpu(p));
5391
	if (!on_rq)
5392
		p->se.vruntime += cfs_rq_of(&p->se)->min_vruntime;
P
Peter Zijlstra 已提交
5393
}
5394 5395 5396 5397 5398 5399 5400 5401 5402 5403 5404 5405 5406 5407 5408 5409 5410 5411 5412 5413 5414 5415 5416 5417 5418 5419 5420 5421 5422 5423 5424 5425 5426 5427 5428 5429 5430 5431 5432 5433 5434 5435 5436 5437 5438 5439 5440 5441 5442 5443 5444 5445 5446 5447 5448 5449 5450 5451 5452 5453 5454 5455 5456 5457 5458 5459 5460 5461 5462 5463 5464 5465 5466 5467 5468 5469 5470 5471 5472 5473 5474 5475 5476 5477 5478 5479

void free_fair_sched_group(struct task_group *tg)
{
	int i;

	destroy_cfs_bandwidth(tg_cfs_bandwidth(tg));

	for_each_possible_cpu(i) {
		if (tg->cfs_rq)
			kfree(tg->cfs_rq[i]);
		if (tg->se)
			kfree(tg->se[i]);
	}

	kfree(tg->cfs_rq);
	kfree(tg->se);
}

int alloc_fair_sched_group(struct task_group *tg, struct task_group *parent)
{
	struct cfs_rq *cfs_rq;
	struct sched_entity *se;
	int i;

	tg->cfs_rq = kzalloc(sizeof(cfs_rq) * nr_cpu_ids, GFP_KERNEL);
	if (!tg->cfs_rq)
		goto err;
	tg->se = kzalloc(sizeof(se) * nr_cpu_ids, GFP_KERNEL);
	if (!tg->se)
		goto err;

	tg->shares = NICE_0_LOAD;

	init_cfs_bandwidth(tg_cfs_bandwidth(tg));

	for_each_possible_cpu(i) {
		cfs_rq = kzalloc_node(sizeof(struct cfs_rq),
				      GFP_KERNEL, cpu_to_node(i));
		if (!cfs_rq)
			goto err;

		se = kzalloc_node(sizeof(struct sched_entity),
				  GFP_KERNEL, cpu_to_node(i));
		if (!se)
			goto err_free_rq;

		init_cfs_rq(cfs_rq);
		init_tg_cfs_entry(tg, cfs_rq, se, i, parent->se[i]);
	}

	return 1;

err_free_rq:
	kfree(cfs_rq);
err:
	return 0;
}

void unregister_fair_sched_group(struct task_group *tg, int cpu)
{
	struct rq *rq = cpu_rq(cpu);
	unsigned long flags;

	/*
	* Only empty task groups can be destroyed; so we can speculatively
	* check on_list without danger of it being re-added.
	*/
	if (!tg->cfs_rq[cpu]->on_list)
		return;

	raw_spin_lock_irqsave(&rq->lock, flags);
	list_del_leaf_cfs_rq(tg->cfs_rq[cpu]);
	raw_spin_unlock_irqrestore(&rq->lock, flags);
}

void init_tg_cfs_entry(struct task_group *tg, struct cfs_rq *cfs_rq,
			struct sched_entity *se, int cpu,
			struct sched_entity *parent)
{
	struct rq *rq = cpu_rq(cpu);

	cfs_rq->tg = tg;
	cfs_rq->rq = rq;
#ifdef CONFIG_SMP
	/* allow initial update_cfs_load() to truncate */
	cfs_rq->load_stamp = 1;
P
Peter Zijlstra 已提交
5480
#endif
5481 5482 5483 5484 5485 5486 5487 5488 5489 5490 5491 5492 5493 5494 5495 5496 5497 5498 5499 5500 5501 5502 5503 5504 5505 5506 5507 5508 5509 5510 5511 5512 5513 5514 5515 5516 5517 5518 5519 5520 5521 5522 5523 5524 5525 5526 5527 5528 5529 5530 5531 5532 5533 5534 5535 5536 5537 5538 5539 5540 5541 5542 5543 5544 5545 5546 5547 5548
	init_cfs_rq_runtime(cfs_rq);

	tg->cfs_rq[cpu] = cfs_rq;
	tg->se[cpu] = se;

	/* se could be NULL for root_task_group */
	if (!se)
		return;

	if (!parent)
		se->cfs_rq = &rq->cfs;
	else
		se->cfs_rq = parent->my_q;

	se->my_q = cfs_rq;
	update_load_set(&se->load, 0);
	se->parent = parent;
}

static DEFINE_MUTEX(shares_mutex);

int sched_group_set_shares(struct task_group *tg, unsigned long shares)
{
	int i;
	unsigned long flags;

	/*
	 * We can't change the weight of the root cgroup.
	 */
	if (!tg->se[0])
		return -EINVAL;

	shares = clamp(shares, scale_load(MIN_SHARES), scale_load(MAX_SHARES));

	mutex_lock(&shares_mutex);
	if (tg->shares == shares)
		goto done;

	tg->shares = shares;
	for_each_possible_cpu(i) {
		struct rq *rq = cpu_rq(i);
		struct sched_entity *se;

		se = tg->se[i];
		/* Propagate contribution to hierarchy */
		raw_spin_lock_irqsave(&rq->lock, flags);
		for_each_sched_entity(se)
			update_cfs_shares(group_cfs_rq(se));
		raw_spin_unlock_irqrestore(&rq->lock, flags);
	}

done:
	mutex_unlock(&shares_mutex);
	return 0;
}
#else /* CONFIG_FAIR_GROUP_SCHED */

void free_fair_sched_group(struct task_group *tg) { }

int alloc_fair_sched_group(struct task_group *tg, struct task_group *parent)
{
	return 1;
}

void unregister_fair_sched_group(struct task_group *tg, int cpu) { }

#endif /* CONFIG_FAIR_GROUP_SCHED */

P
Peter Zijlstra 已提交
5549

5550
static unsigned int get_rr_interval_fair(struct rq *rq, struct task_struct *task)
5551 5552 5553 5554 5555 5556 5557 5558 5559 5560 5561 5562 5563 5564
{
	struct sched_entity *se = &task->se;
	unsigned int rr_interval = 0;

	/*
	 * Time slice is 0 for SCHED_OTHER tasks that are on an otherwise
	 * idle runqueue:
	 */
	if (rq->cfs.load.weight)
		rr_interval = NS_TO_JIFFIES(sched_slice(&rq->cfs, se));

	return rr_interval;
}

5565 5566 5567
/*
 * All the scheduling class methods:
 */
5568
const struct sched_class fair_sched_class = {
5569
	.next			= &idle_sched_class,
5570 5571 5572
	.enqueue_task		= enqueue_task_fair,
	.dequeue_task		= dequeue_task_fair,
	.yield_task		= yield_task_fair,
5573
	.yield_to_task		= yield_to_task_fair,
5574

I
Ingo Molnar 已提交
5575
	.check_preempt_curr	= check_preempt_wakeup,
5576 5577 5578 5579

	.pick_next_task		= pick_next_task_fair,
	.put_prev_task		= put_prev_task_fair,

5580
#ifdef CONFIG_SMP
L
Li Zefan 已提交
5581 5582
	.select_task_rq		= select_task_rq_fair,

5583 5584
	.rq_online		= rq_online_fair,
	.rq_offline		= rq_offline_fair,
5585 5586

	.task_waking		= task_waking_fair,
5587
#endif
5588

5589
	.set_curr_task          = set_curr_task_fair,
5590
	.task_tick		= task_tick_fair,
P
Peter Zijlstra 已提交
5591
	.task_fork		= task_fork_fair,
5592 5593

	.prio_changed		= prio_changed_fair,
P
Peter Zijlstra 已提交
5594
	.switched_from		= switched_from_fair,
5595
	.switched_to		= switched_to_fair,
P
Peter Zijlstra 已提交
5596

5597 5598
	.get_rr_interval	= get_rr_interval_fair,

P
Peter Zijlstra 已提交
5599
#ifdef CONFIG_FAIR_GROUP_SCHED
5600
	.task_move_group	= task_move_group_fair,
P
Peter Zijlstra 已提交
5601
#endif
5602 5603 5604
};

#ifdef CONFIG_SCHED_DEBUG
5605
void print_cfs_stats(struct seq_file *m, int cpu)
5606 5607 5608
{
	struct cfs_rq *cfs_rq;

5609
	rcu_read_lock();
5610
	for_each_leaf_cfs_rq(cpu_rq(cpu), cfs_rq)
5611
		print_cfs_rq(m, cpu, cfs_rq);
5612
	rcu_read_unlock();
5613 5614
}
#endif
5615 5616 5617 5618 5619 5620 5621 5622 5623 5624 5625 5626 5627 5628 5629 5630

__init void init_sched_fair_class(void)
{
#ifdef CONFIG_SMP
	open_softirq(SCHED_SOFTIRQ, run_rebalance_domains);

#ifdef CONFIG_NO_HZ
	zalloc_cpumask_var(&nohz.idle_cpus_mask, GFP_NOWAIT);
	alloc_cpumask_var(&nohz.grp_idle_mask, GFP_NOWAIT);
	atomic_set(&nohz.load_balancer, nr_cpu_ids);
	atomic_set(&nohz.first_pick_cpu, nr_cpu_ids);
	atomic_set(&nohz.second_pick_cpu, nr_cpu_ids);
#endif
#endif /* SMP */

}