sched_fair.c 105.2 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>
A
Arjan van de Ven 已提交
25

26
/*
27
 * Targeted preemption latency for CPU-bound tasks:
28
 * (default: 6ms * (1 + ilog(ncpus)), units: nanoseconds)
29
 *
30
 * NOTE: this latency value is not the same as the concept of
I
Ingo Molnar 已提交
31 32 33
 * 'timeslice length' - timeslices in CFS are of variable length
 * and have no persistent notion like in traditional, time-slice
 * based scheduling concepts.
34
 *
I
Ingo Molnar 已提交
35 36
 * (to see the precise effective timeslice length of your workload,
 *  run vmstat and monitor the context-switches (cs) field)
37
 */
38 39
unsigned int sysctl_sched_latency = 6000000ULL;
unsigned int normalized_sysctl_sched_latency = 6000000ULL;
40

41 42 43 44 45 46 47 48 49 50 51 52
/*
 * 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;

53
/*
54
 * Minimal preemption granularity for CPU-bound tasks:
55
 * (default: 0.75 msec * (1 + ilog(ncpus)), units: nanoseconds)
56
 */
57 58
unsigned int sysctl_sched_min_granularity = 750000ULL;
unsigned int normalized_sysctl_sched_min_granularity = 750000ULL;
59 60

/*
61 62
 * is kept at sysctl_sched_latency / sysctl_sched_min_granularity
 */
63
static unsigned int sched_nr_latency = 8;
64 65

/*
66
 * After fork, child runs first. If set to 0 (default) then
67
 * parent will (try to) run first.
68
 */
69
unsigned int sysctl_sched_child_runs_first __read_mostly;
70

71 72 73 74 75 76 77 78
/*
 * sys_sched_yield() compat mode
 *
 * This option switches the agressive yield implementation of the
 * old scheduler back on.
 */
unsigned int __read_mostly sysctl_sched_compat_yield;

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
static const struct sched_class fair_sched_class;

101 102 103 104
/**************************************************************
 * CFS operations on generic schedulable entities:
 */

105
#ifdef CONFIG_FAIR_GROUP_SCHED
106

107
/* cpu runqueue to which this cfs_rq is attached */
108 109
static inline struct rq *rq_of(struct cfs_rq *cfs_rq)
{
110
	return cfs_rq->rq;
111 112
}

113 114
/* An entity is a task if it doesn't "own" a runqueue */
#define entity_is_task(se)	(!se->my_q)
115

116 117 118 119 120 121 122 123
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
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
/* 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;
}

/* Given a group's cfs_rq on one cpu, return its corresponding cfs_rq on
 * another cpu ('this_cpu')
 */
static inline struct cfs_rq *cpu_cfs_rq(struct cfs_rq *cfs_rq, int this_cpu)
{
	return cfs_rq->tg->cfs_rq[this_cpu];
}

153 154 155
static inline void list_add_leaf_cfs_rq(struct cfs_rq *cfs_rq)
{
	if (!cfs_rq->on_list) {
156 157 158 159 160 161 162 163 164 165 166 167
		/*
		 * 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,
168
				&rq_of(cfs_rq)->leaf_cfs_rq_list);
169
		}
170 171 172 173 174 175 176 177 178 179 180 181 182

		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
183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201
/* 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;
}

202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244
/* 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);
	}
}

245 246 247 248 249 250
#else	/* !CONFIG_FAIR_GROUP_SCHED */

static inline struct task_struct *task_of(struct sched_entity *se)
{
	return container_of(se, struct task_struct, se);
}
251

252 253 254
static inline struct rq *rq_of(struct cfs_rq *cfs_rq)
{
	return container_of(cfs_rq, struct rq, cfs);
255 256 257 258
}

#define entity_is_task(se)	1

P
Peter Zijlstra 已提交
259 260
#define for_each_sched_entity(se) \
		for (; se; se = NULL)
261

P
Peter Zijlstra 已提交
262
static inline struct cfs_rq *task_cfs_rq(struct task_struct *p)
263
{
P
Peter Zijlstra 已提交
264
	return &task_rq(p)->cfs;
265 266
}

P
Peter Zijlstra 已提交
267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285
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;
}

static inline struct cfs_rq *cpu_cfs_rq(struct cfs_rq *cfs_rq, int this_cpu)
{
	return &cpu_rq(this_cpu)->cfs;
}

286 287 288 289 290 291 292 293
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
294 295 296 297 298 299 300 301 302 303 304 305 306 307
#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;
}

308 309 310 311 312
static inline void
find_matching_se(struct sched_entity **se, struct sched_entity **pse)
{
}

P
Peter Zijlstra 已提交
313 314
#endif	/* CONFIG_FAIR_GROUP_SCHED */

315 316 317 318 319

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

320
static inline u64 max_vruntime(u64 min_vruntime, u64 vruntime)
321
{
322 323
	s64 delta = (s64)(vruntime - min_vruntime);
	if (delta > 0)
324 325 326 327 328
		min_vruntime = vruntime;

	return min_vruntime;
}

329
static inline u64 min_vruntime(u64 min_vruntime, u64 vruntime)
P
Peter Zijlstra 已提交
330 331 332 333 334 335 336 337
{
	s64 delta = (s64)(vruntime - min_vruntime);
	if (delta < 0)
		min_vruntime = vruntime;

	return min_vruntime;
}

338 339 340 341 342 343
static inline int entity_before(struct sched_entity *a,
				struct sched_entity *b)
{
	return (s64)(a->vruntime - b->vruntime) < 0;
}

344
static inline s64 entity_key(struct cfs_rq *cfs_rq, struct sched_entity *se)
345
{
346
	return se->vruntime - cfs_rq->min_vruntime;
347 348
}

349 350 351 352 353 354 355 356 357 358 359 360
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
361
		if (!cfs_rq->curr)
362 363 364 365 366 367 368 369
			vruntime = se->vruntime;
		else
			vruntime = min_vruntime(vruntime, se->vruntime);
	}

	cfs_rq->min_vruntime = max_vruntime(cfs_rq->min_vruntime, vruntime);
}

370 371 372
/*
 * Enqueue an entity into the rb-tree:
 */
373
static void __enqueue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se)
374 375 376 377
{
	struct rb_node **link = &cfs_rq->tasks_timeline.rb_node;
	struct rb_node *parent = NULL;
	struct sched_entity *entry;
378
	s64 key = entity_key(cfs_rq, se);
379 380 381 382 383 384 385 386 387 388 389 390
	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.
		 */
391
		if (key < entity_key(cfs_rq, entry)) {
392 393 394 395 396 397 398 399 400 401 402
			link = &parent->rb_left;
		} else {
			link = &parent->rb_right;
			leftmost = 0;
		}
	}

	/*
	 * Maintain a cache of leftmost tree entries (it is frequently
	 * used):
	 */
403
	if (leftmost)
I
Ingo Molnar 已提交
404
		cfs_rq->rb_leftmost = &se->run_node;
405 406 407 408 409

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

410
static void __dequeue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se)
411
{
P
Peter Zijlstra 已提交
412 413 414 415 416 417
	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
418

419 420 421 422 423
	rb_erase(&se->run_node, &cfs_rq->tasks_timeline);
}

static struct sched_entity *__pick_next_entity(struct cfs_rq *cfs_rq)
{
424 425 426 427 428 429
	struct rb_node *left = cfs_rq->rb_leftmost;

	if (!left)
		return NULL;

	return rb_entry(left, struct sched_entity, run_node);
430 431
}

432
static struct sched_entity *__pick_last_entity(struct cfs_rq *cfs_rq)
433
{
434
	struct rb_node *last = rb_last(&cfs_rq->tasks_timeline);
435

436 437
	if (!last)
		return NULL;
438 439

	return rb_entry(last, struct sched_entity, run_node);
440 441
}

442 443 444 445
/**************************************************************
 * Scheduling class statistics methods:
 */

446
#ifdef CONFIG_SCHED_DEBUG
447
int sched_proc_update_handler(struct ctl_table *table, int write,
448
		void __user *buffer, size_t *lenp,
449 450
		loff_t *ppos)
{
451
	int ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
452
	int factor = get_update_sysctl_factor();
453 454 455 456 457 458 459

	if (ret || !write)
		return ret;

	sched_nr_latency = DIV_ROUND_UP(sysctl_sched_latency,
					sysctl_sched_min_granularity);

460 461 462 463 464 465 466
#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

467 468 469
	return 0;
}
#endif
470

471
/*
472
 * delta /= w
473 474 475 476
 */
static inline unsigned long
calc_delta_fair(unsigned long delta, struct sched_entity *se)
{
477 478
	if (unlikely(se->load.weight != NICE_0_LOAD))
		delta = calc_delta_mine(delta, NICE_0_LOAD, &se->load);
479 480 481 482

	return delta;
}

483 484 485 486 487 488 489 490
/*
 * 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
 */
491 492 493
static u64 __sched_period(unsigned long nr_running)
{
	u64 period = sysctl_sched_latency;
494
	unsigned long nr_latency = sched_nr_latency;
495 496

	if (unlikely(nr_running > nr_latency)) {
497
		period = sysctl_sched_min_granularity;
498 499 500 501 502 503
		period *= nr_running;
	}

	return period;
}

504 505 506 507
/*
 * We calculate the wall-time slice from the period by taking a part
 * proportional to the weight.
 *
508
 * s = p*P[w/rw]
509
 */
P
Peter Zijlstra 已提交
510
static u64 sched_slice(struct cfs_rq *cfs_rq, struct sched_entity *se)
511
{
M
Mike Galbraith 已提交
512
	u64 slice = __sched_period(cfs_rq->nr_running + !se->on_rq);
513

M
Mike Galbraith 已提交
514
	for_each_sched_entity(se) {
L
Lin Ming 已提交
515
		struct load_weight *load;
516
		struct load_weight lw;
L
Lin Ming 已提交
517 518 519

		cfs_rq = cfs_rq_of(se);
		load = &cfs_rq->load;
520

M
Mike Galbraith 已提交
521
		if (unlikely(!se->on_rq)) {
522
			lw = cfs_rq->load;
M
Mike Galbraith 已提交
523 524 525 526 527 528 529

			update_load_add(&lw, se->load.weight);
			load = &lw;
		}
		slice = calc_delta_mine(slice, se->load.weight, load);
	}
	return slice;
530 531
}

532
/*
533
 * We calculate the vruntime slice of a to be inserted task
534
 *
535
 * vs = s/w
536
 */
537
static u64 sched_vslice(struct cfs_rq *cfs_rq, struct sched_entity *se)
P
Peter Zijlstra 已提交
538
{
539
	return calc_delta_fair(sched_slice(cfs_rq, se), se);
540 541
}

542 543 544 545 546
/*
 * 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
547 548
__update_curr(struct cfs_rq *cfs_rq, struct sched_entity *curr,
	      unsigned long delta_exec)
549
{
550
	unsigned long delta_exec_weighted;
551

552 553
	schedstat_set(curr->statistics.exec_max,
		      max((u64)delta_exec, curr->statistics.exec_max));
554 555

	curr->sum_exec_runtime += delta_exec;
556
	schedstat_add(cfs_rq, exec_clock, delta_exec);
557
	delta_exec_weighted = calc_delta_fair(delta_exec, curr);
558

I
Ingo Molnar 已提交
559
	curr->vruntime += delta_exec_weighted;
560
	update_min_vruntime(cfs_rq);
561 562
}

563
static void update_curr(struct cfs_rq *cfs_rq)
564
{
565
	struct sched_entity *curr = cfs_rq->curr;
566
	u64 now = rq_of(cfs_rq)->clock_task;
567 568 569 570 571 572 573 574 575 576
	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
577
	delta_exec = (unsigned long)(now - curr->exec_start);
P
Peter Zijlstra 已提交
578 579
	if (!delta_exec)
		return;
580

I
Ingo Molnar 已提交
581 582
	__update_curr(cfs_rq, curr, delta_exec);
	curr->exec_start = now;
583 584 585 586

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

587
		trace_sched_stat_runtime(curtask, delta_exec, curr->vruntime);
588
		cpuacct_charge(curtask, delta_exec);
589
		account_group_exec_runtime(curtask, delta_exec);
590
	}
591 592 593
}

static inline void
594
update_stats_wait_start(struct cfs_rq *cfs_rq, struct sched_entity *se)
595
{
596
	schedstat_set(se->statistics.wait_start, rq_of(cfs_rq)->clock);
597 598 599 600 601
}

/*
 * Task is being enqueued - update stats:
 */
602
static void update_stats_enqueue(struct cfs_rq *cfs_rq, struct sched_entity *se)
603 604 605 606 607
{
	/*
	 * Are we enqueueing a waiting task? (for current tasks
	 * a dequeue/enqueue event is a NOP)
	 */
608
	if (se != cfs_rq->curr)
609
		update_stats_wait_start(cfs_rq, se);
610 611 612
}

static void
613
update_stats_wait_end(struct cfs_rq *cfs_rq, struct sched_entity *se)
614
{
615 616 617 618 619
	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);
620 621 622
#ifdef CONFIG_SCHEDSTATS
	if (entity_is_task(se)) {
		trace_sched_stat_wait(task_of(se),
623
			rq_of(cfs_rq)->clock - se->statistics.wait_start);
624 625
	}
#endif
626
	schedstat_set(se->statistics.wait_start, 0);
627 628 629
}

static inline void
630
update_stats_dequeue(struct cfs_rq *cfs_rq, struct sched_entity *se)
631 632 633 634 635
{
	/*
	 * Mark the end of the wait period if dequeueing a
	 * waiting task:
	 */
636
	if (se != cfs_rq->curr)
637
		update_stats_wait_end(cfs_rq, se);
638 639 640 641 642 643
}

/*
 * We are picking a new current task - update its stats:
 */
static inline void
644
update_stats_curr_start(struct cfs_rq *cfs_rq, struct sched_entity *se)
645 646 647 648
{
	/*
	 * We are starting a new run period:
	 */
649
	se->exec_start = rq_of(cfs_rq)->clock_task;
650 651 652 653 654 655
}

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

656 657 658 659 660 661 662 663 664 665 666 667 668
#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

669 670 671 672
static void
account_entity_enqueue(struct cfs_rq *cfs_rq, struct sched_entity *se)
{
	update_load_add(&cfs_rq->load, se->load.weight);
673 674
	if (!parent_entity(se))
		inc_cpu_load(rq_of(cfs_rq), se->load.weight);
675
	if (entity_is_task(se)) {
676
		add_cfs_task_weight(cfs_rq, se->load.weight);
677 678
		list_add(&se->group_node, &cfs_rq->tasks);
	}
679 680 681 682 683 684 685
	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);
686 687
	if (!parent_entity(se))
		dec_cpu_load(rq_of(cfs_rq), se->load.weight);
688
	if (entity_is_task(se)) {
689
		add_cfs_task_weight(cfs_rq, -se->load.weight);
690 691
		list_del_init(&se->group_node);
	}
692 693 694
	cfs_rq->nr_running--;
}

P
Peter Zijlstra 已提交
695
#if defined CONFIG_SMP && defined CONFIG_FAIR_GROUP_SCHED
696
static void update_cfs_load(struct cfs_rq *cfs_rq)
P
Peter Zijlstra 已提交
697
{
698
	u64 period = sysctl_sched_shares_window;
P
Peter Zijlstra 已提交
699
	u64 now, delta;
700
	unsigned long load = cfs_rq->load.weight;
P
Peter Zijlstra 已提交
701 702 703 704 705 706 707

	if (!cfs_rq)
		return;

	now = rq_of(cfs_rq)->clock;
	delta = now - cfs_rq->load_stamp;

708 709 710 711 712 713 714
	/* 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;
	}

P
Peter Zijlstra 已提交
715 716
	cfs_rq->load_stamp = now;
	cfs_rq->load_period += delta;
717 718 719 720
	if (load) {
		cfs_rq->load_last = now;
		cfs_rq->load_avg += delta * load;
	}
P
Peter Zijlstra 已提交
721 722 723 724 725 726 727 728 729 730 731

	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;
	}
732

733 734
	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
735 736 737 738 739 740 741 742 743 744 745 746 747 748
}

static void reweight_entity(struct cfs_rq *cfs_rq, struct sched_entity *se,
			    unsigned long weight)
{
	if (se->on_rq)
		account_entity_dequeue(cfs_rq, se);

	update_load_set(&se->load, weight);

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

749
static void update_cfs_shares(struct cfs_rq *cfs_rq, long weight_delta)
P
Peter Zijlstra 已提交
750 751 752 753 754 755 756 757 758 759 760 761 762
{
	struct task_group *tg;
	struct sched_entity *se;
	long load_weight, load, shares;

	if (!cfs_rq)
		return;

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

763
	load = cfs_rq->load.weight + weight_delta;
P
Peter Zijlstra 已提交
764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780

	load_weight = atomic_read(&tg->load_weight);
	load_weight -= cfs_rq->load_contribution;
	load_weight += load;

	shares = (tg->shares * load);
	if (load_weight)
		shares /= load_weight;

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

	reweight_entity(cfs_rq_of(se), se, shares);
}
#else /* CONFIG_FAIR_GROUP_SCHED */
781
static inline void update_cfs_load(struct cfs_rq *cfs_rq)
P
Peter Zijlstra 已提交
782 783 784
{
}

785
static inline void update_cfs_shares(struct cfs_rq *cfs_rq, long weight_delta)
P
Peter Zijlstra 已提交
786 787 788 789
{
}
#endif /* CONFIG_FAIR_GROUP_SCHED */

790
static void enqueue_sleeper(struct cfs_rq *cfs_rq, struct sched_entity *se)
791 792
{
#ifdef CONFIG_SCHEDSTATS
793 794 795 796 797
	struct task_struct *tsk = NULL;

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

798 799
	if (se->statistics.sleep_start) {
		u64 delta = rq_of(cfs_rq)->clock - se->statistics.sleep_start;
800 801 802 803

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

804 805
		if (unlikely(delta > se->statistics.sleep_max))
			se->statistics.sleep_max = delta;
806

807 808
		se->statistics.sleep_start = 0;
		se->statistics.sum_sleep_runtime += delta;
A
Arjan van de Ven 已提交
809

810
		if (tsk) {
811
			account_scheduler_latency(tsk, delta >> 10, 1);
812 813
			trace_sched_stat_sleep(tsk, delta);
		}
814
	}
815 816
	if (se->statistics.block_start) {
		u64 delta = rq_of(cfs_rq)->clock - se->statistics.block_start;
817 818 819 820

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

821 822
		if (unlikely(delta > se->statistics.block_max))
			se->statistics.block_max = delta;
823

824 825
		se->statistics.block_start = 0;
		se->statistics.sum_sleep_runtime += delta;
I
Ingo Molnar 已提交
826

827
		if (tsk) {
828
			if (tsk->in_iowait) {
829 830
				se->statistics.iowait_sum += delta;
				se->statistics.iowait_count++;
831
				trace_sched_stat_iowait(tsk, delta);
832 833
			}

834 835 836 837 838 839 840 841 842 843 844
			/*
			 * 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
845
		}
846 847 848 849
	}
#endif
}

P
Peter Zijlstra 已提交
850 851 852 853 854 855 856 857 858 859 860 861 862
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
}

863 864 865
static void
place_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int initial)
{
866
	u64 vruntime = cfs_rq->min_vruntime;
P
Peter Zijlstra 已提交
867

868 869 870 871 872 873
	/*
	 * 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
874
	if (initial && sched_feat(START_DEBIT))
875
		vruntime += sched_vslice(cfs_rq, se);
876

877
	/* sleeps up to a single latency don't count. */
878
	if (!initial) {
879
		unsigned long thresh = sysctl_sched_latency;
880

881 882 883 884 885 886
		/*
		 * Halve their sleep time's effect, to allow
		 * for a gentler effect of sleepers:
		 */
		if (sched_feat(GENTLE_FAIR_SLEEPERS))
			thresh >>= 1;
887

888
		vruntime -= thresh;
889 890
	}

891 892 893
	/* ensure we never gain time by being placed backwards. */
	vruntime = max_vruntime(se->vruntime, vruntime);

P
Peter Zijlstra 已提交
894
	se->vruntime = vruntime;
895 896
}

897
static void
898
enqueue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int flags)
899
{
900 901 902 903
	/*
	 * Update the normalized vruntime before updating min_vruntime
	 * through callig update_curr().
	 */
904
	if (!(flags & ENQUEUE_WAKEUP) || (flags & ENQUEUE_WAKING))
905 906
		se->vruntime += cfs_rq->min_vruntime;

907
	/*
908
	 * Update run-time statistics of the 'current'.
909
	 */
910
	update_curr(cfs_rq);
911
	update_cfs_load(cfs_rq);
912
	update_cfs_shares(cfs_rq, se->load.weight);
P
Peter Zijlstra 已提交
913
	account_entity_enqueue(cfs_rq, se);
914

915
	if (flags & ENQUEUE_WAKEUP) {
916
		place_entity(cfs_rq, se, 0);
917
		enqueue_sleeper(cfs_rq, se);
I
Ingo Molnar 已提交
918
	}
919

920
	update_stats_enqueue(cfs_rq, se);
P
Peter Zijlstra 已提交
921
	check_spread(cfs_rq, se);
922 923
	if (se != cfs_rq->curr)
		__enqueue_entity(cfs_rq, se);
P
Peter Zijlstra 已提交
924
	se->on_rq = 1;
925 926 927

	if (cfs_rq->nr_running == 1)
		list_add_leaf_cfs_rq(cfs_rq);
928 929
}

P
Peter Zijlstra 已提交
930
static void __clear_buddies(struct cfs_rq *cfs_rq, struct sched_entity *se)
P
Peter Zijlstra 已提交
931
{
932
	if (!se || cfs_rq->last == se)
P
Peter Zijlstra 已提交
933 934
		cfs_rq->last = NULL;

935
	if (!se || cfs_rq->next == se)
P
Peter Zijlstra 已提交
936 937 938
		cfs_rq->next = NULL;
}

P
Peter Zijlstra 已提交
939 940 941 942 943 944
static void clear_buddies(struct cfs_rq *cfs_rq, struct sched_entity *se)
{
	for_each_sched_entity(se)
		__clear_buddies(cfs_rq_of(se), se);
}

945
static void
946
dequeue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int flags)
947
{
948 949 950 951 952
	/*
	 * Update run-time statistics of the 'current'.
	 */
	update_curr(cfs_rq);

953
	update_stats_dequeue(cfs_rq, se);
954
	if (flags & DEQUEUE_SLEEP) {
P
Peter Zijlstra 已提交
955
#ifdef CONFIG_SCHEDSTATS
956 957 958 959
		if (entity_is_task(se)) {
			struct task_struct *tsk = task_of(se);

			if (tsk->state & TASK_INTERRUPTIBLE)
960
				se->statistics.sleep_start = rq_of(cfs_rq)->clock;
961
			if (tsk->state & TASK_UNINTERRUPTIBLE)
962
				se->statistics.block_start = rq_of(cfs_rq)->clock;
963
		}
964
#endif
P
Peter Zijlstra 已提交
965 966
	}

P
Peter Zijlstra 已提交
967
	clear_buddies(cfs_rq, se);
P
Peter Zijlstra 已提交
968

969
	if (se != cfs_rq->curr)
970
		__dequeue_entity(cfs_rq, se);
P
Peter Zijlstra 已提交
971
	se->on_rq = 0;
972
	update_cfs_load(cfs_rq);
973
	account_entity_dequeue(cfs_rq, se);
974
	update_min_vruntime(cfs_rq);
975
	update_cfs_shares(cfs_rq, 0);
976 977 978 979 980 981

	/*
	 * 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.
	 */
982
	if (!(flags & DEQUEUE_SLEEP))
983
		se->vruntime -= cfs_rq->min_vruntime;
984 985 986 987 988
}

/*
 * Preempt the current task with a newly woken task if needed:
 */
989
static void
I
Ingo Molnar 已提交
990
check_preempt_tick(struct cfs_rq *cfs_rq, struct sched_entity *curr)
991
{
992 993
	unsigned long ideal_runtime, delta_exec;

P
Peter Zijlstra 已提交
994
	ideal_runtime = sched_slice(cfs_rq, curr);
995
	delta_exec = curr->sum_exec_runtime - curr->prev_sum_exec_runtime;
996
	if (delta_exec > ideal_runtime) {
997
		resched_task(rq_of(cfs_rq)->curr);
998 999 1000 1001 1002
		/*
		 * The current task ran long enough, ensure it doesn't get
		 * re-elected due to buddy favours.
		 */
		clear_buddies(cfs_rq, curr);
1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022
		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 (!sched_feat(WAKEUP_PREEMPT))
		return;

	if (delta_exec < sysctl_sched_min_granularity)
		return;

	if (cfs_rq->nr_running > 1) {
		struct sched_entity *se = __pick_next_entity(cfs_rq);
		s64 delta = curr->vruntime - se->vruntime;

		if (delta > ideal_runtime)
			resched_task(rq_of(cfs_rq)->curr);
1023
	}
1024 1025
}

1026
static void
1027
set_next_entity(struct cfs_rq *cfs_rq, struct sched_entity *se)
1028
{
1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039
	/* '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);
	}

1040
	update_stats_curr_start(cfs_rq, se);
1041
	cfs_rq->curr = se;
I
Ingo Molnar 已提交
1042 1043 1044 1045 1046 1047
#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):
	 */
1048
	if (rq_of(cfs_rq)->load.weight >= 2*se->load.weight) {
1049
		se->statistics.slice_max = max(se->statistics.slice_max,
I
Ingo Molnar 已提交
1050 1051 1052
			se->sum_exec_runtime - se->prev_sum_exec_runtime);
	}
#endif
1053
	se->prev_sum_exec_runtime = se->sum_exec_runtime;
1054 1055
}

1056 1057 1058
static int
wakeup_preempt_entity(struct sched_entity *curr, struct sched_entity *se);

1059
static struct sched_entity *pick_next_entity(struct cfs_rq *cfs_rq)
1060
{
1061
	struct sched_entity *se = __pick_next_entity(cfs_rq);
1062
	struct sched_entity *left = se;
1063

1064 1065
	if (cfs_rq->next && wakeup_preempt_entity(cfs_rq->next, left) < 1)
		se = cfs_rq->next;
1066

1067 1068 1069 1070 1071 1072 1073
	/*
	 * 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;

	clear_buddies(cfs_rq, se);
P
Peter Zijlstra 已提交
1074 1075

	return se;
1076 1077
}

1078
static void put_prev_entity(struct cfs_rq *cfs_rq, struct sched_entity *prev)
1079 1080 1081 1082 1083 1084
{
	/*
	 * If still on the runqueue then deactivate_task()
	 * was not called and update_curr() has to be done:
	 */
	if (prev->on_rq)
1085
		update_curr(cfs_rq);
1086

P
Peter Zijlstra 已提交
1087
	check_spread(cfs_rq, prev);
1088
	if (prev->on_rq) {
1089
		update_stats_wait_start(cfs_rq, prev);
1090 1091 1092
		/* Put 'current' back into the tree. */
		__enqueue_entity(cfs_rq, prev);
	}
1093
	cfs_rq->curr = NULL;
1094 1095
}

P
Peter Zijlstra 已提交
1096 1097
static void
entity_tick(struct cfs_rq *cfs_rq, struct sched_entity *curr, int queued)
1098 1099
{
	/*
1100
	 * Update run-time statistics of the 'current'.
1101
	 */
1102
	update_curr(cfs_rq);
1103

P
Peter Zijlstra 已提交
1104 1105 1106 1107 1108
#ifdef CONFIG_SCHED_HRTICK
	/*
	 * queued ticks are scheduled to match the slice, so don't bother
	 * validating it and just reschedule.
	 */
1109 1110 1111 1112
	if (queued) {
		resched_task(rq_of(cfs_rq)->curr);
		return;
	}
P
Peter Zijlstra 已提交
1113 1114 1115 1116 1117 1118 1119 1120
	/*
	 * 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

1121
	if (cfs_rq->nr_running > 1 || !sched_feat(WAKEUP_PREEMPT))
I
Ingo Molnar 已提交
1122
		check_preempt_tick(cfs_rq, curr);
1123 1124 1125 1126 1127 1128
}

/**************************************************
 * CFS operations on tasks:
 */

P
Peter Zijlstra 已提交
1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151
#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.
		 */
1152
		if (rq->curr != p)
1153
			delta = max_t(s64, 10000LL, delta);
P
Peter Zijlstra 已提交
1154

1155
		hrtick_start(rq, delta);
P
Peter Zijlstra 已提交
1156 1157
	}
}
1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173

/*
 * 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);
}
1174
#else /* !CONFIG_SCHED_HRTICK */
P
Peter Zijlstra 已提交
1175 1176 1177 1178
static inline void
hrtick_start_fair(struct rq *rq, struct task_struct *p)
{
}
1179 1180 1181 1182

static inline void hrtick_update(struct rq *rq)
{
}
P
Peter Zijlstra 已提交
1183 1184
#endif

1185 1186 1187 1188 1189
/*
 * 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:
 */
1190
static void
1191
enqueue_task_fair(struct rq *rq, struct task_struct *p, int flags)
1192 1193
{
	struct cfs_rq *cfs_rq;
1194
	struct sched_entity *se = &p->se;
1195 1196

	for_each_sched_entity(se) {
1197
		if (se->on_rq)
1198 1199
			break;
		cfs_rq = cfs_rq_of(se);
1200 1201
		enqueue_entity(cfs_rq, se, flags);
		flags = ENQUEUE_WAKEUP;
1202
	}
P
Peter Zijlstra 已提交
1203

P
Peter Zijlstra 已提交
1204 1205 1206
	for_each_sched_entity(se) {
		struct cfs_rq *cfs_rq = cfs_rq_of(se);

1207
		update_cfs_load(cfs_rq);
1208
		update_cfs_shares(cfs_rq, 0);
P
Peter Zijlstra 已提交
1209 1210
	}

1211
	hrtick_update(rq);
1212 1213 1214 1215 1216 1217 1218
}

/*
 * The dequeue_task method is called before nr_running is
 * decreased. We remove the task from the rbtree and
 * update the fair scheduling stats:
 */
1219
static void dequeue_task_fair(struct rq *rq, struct task_struct *p, int flags)
1220 1221
{
	struct cfs_rq *cfs_rq;
1222
	struct sched_entity *se = &p->se;
1223 1224 1225

	for_each_sched_entity(se) {
		cfs_rq = cfs_rq_of(se);
1226
		dequeue_entity(cfs_rq, se, flags);
P
Peter Zijlstra 已提交
1227

1228
		/* Don't dequeue parent if it has other entities besides us */
1229
		if (cfs_rq->load.weight)
1230
			break;
1231
		flags |= DEQUEUE_SLEEP;
1232
	}
P
Peter Zijlstra 已提交
1233

P
Peter Zijlstra 已提交
1234 1235 1236
	for_each_sched_entity(se) {
		struct cfs_rq *cfs_rq = cfs_rq_of(se);

1237
		update_cfs_load(cfs_rq);
1238
		update_cfs_shares(cfs_rq, 0);
P
Peter Zijlstra 已提交
1239 1240
	}

1241
	hrtick_update(rq);
1242 1243 1244
}

/*
1245 1246 1247
 * sched_yield() support is very simple - we dequeue and enqueue.
 *
 * If compat_yield is turned on then we requeue to the end of the tree.
1248
 */
1249
static void yield_task_fair(struct rq *rq)
1250
{
1251 1252 1253
	struct task_struct *curr = rq->curr;
	struct cfs_rq *cfs_rq = task_cfs_rq(curr);
	struct sched_entity *rightmost, *se = &curr->se;
1254 1255

	/*
1256 1257 1258 1259 1260
	 * Are we the only task in the tree?
	 */
	if (unlikely(cfs_rq->nr_running == 1))
		return;

P
Peter Zijlstra 已提交
1261 1262
	clear_buddies(cfs_rq, se);

1263
	if (likely(!sysctl_sched_compat_yield) && curr->policy != SCHED_BATCH) {
1264
		update_rq_clock(rq);
1265
		/*
1266
		 * Update run-time statistics of the 'current'.
1267
		 */
D
Dmitry Adamushko 已提交
1268
		update_curr(cfs_rq);
1269 1270 1271 1272 1273

		return;
	}
	/*
	 * Find the rightmost entry in the rbtree:
1274
	 */
D
Dmitry Adamushko 已提交
1275
	rightmost = __pick_last_entity(cfs_rq);
1276 1277 1278
	/*
	 * Already in the rightmost position?
	 */
1279
	if (unlikely(!rightmost || entity_before(rightmost, se)))
1280 1281 1282 1283
		return;

	/*
	 * Minimally necessary key value to be last in the tree:
D
Dmitry Adamushko 已提交
1284 1285
	 * Upon rescheduling, sched_class::put_prev_task() will place
	 * 'current' within the tree based on its new key value.
1286
	 */
1287
	se->vruntime = rightmost->vruntime + 1;
1288 1289
}

1290
#ifdef CONFIG_SMP
1291

1292 1293 1294 1295 1296 1297 1298 1299
static void task_waking_fair(struct rq *rq, struct task_struct *p)
{
	struct sched_entity *se = &p->se;
	struct cfs_rq *cfs_rq = cfs_rq_of(se);

	se->vruntime -= cfs_rq->min_vruntime;
}

1300
#ifdef CONFIG_FAIR_GROUP_SCHED
1301 1302 1303 1304 1305 1306 1307
/*
 * 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.
 */
P
Peter Zijlstra 已提交
1308
static long effective_load(struct task_group *tg, int cpu, long wl, long wg)
1309
{
P
Peter Zijlstra 已提交
1310
	struct sched_entity *se = tg->se[cpu];
1311 1312 1313 1314

	if (!tg->parent)
		return wl;

P
Peter Zijlstra 已提交
1315
	for_each_sched_entity(se) {
1316
		long S, rw, s, a, b;
P
Peter Zijlstra 已提交
1317 1318

		S = se->my_q->tg->shares;
P
Peter Zijlstra 已提交
1319 1320
		s = se->load.weight;
		rw = se->my_q->load.weight;
1321

1322 1323
		a = S*(rw + wl);
		b = S*rw + s*wg;
P
Peter Zijlstra 已提交
1324

1325 1326 1327 1328 1329
		wl = s*(a-b);

		if (likely(b))
			wl /= b;

1330 1331 1332 1333 1334 1335 1336
		/*
		 * Assume the group is already running and will
		 * thus already be accounted for in the weight.
		 *
		 * That is, moving shares between CPUs, does not
		 * alter the group weight.
		 */
P
Peter Zijlstra 已提交
1337 1338
		wg = 0;
	}
1339

P
Peter Zijlstra 已提交
1340
	return wl;
1341
}
P
Peter Zijlstra 已提交
1342

1343
#else
P
Peter Zijlstra 已提交
1344

1345 1346
static inline unsigned long effective_load(struct task_group *tg, int cpu,
		unsigned long wl, unsigned long wg)
P
Peter Zijlstra 已提交
1347
{
1348
	return wl;
1349
}
P
Peter Zijlstra 已提交
1350

1351 1352
#endif

1353
static int wake_affine(struct sched_domain *sd, struct task_struct *p, int sync)
1354
{
1355 1356
	unsigned long this_load, load;
	int idx, this_cpu, prev_cpu;
1357
	unsigned long tl_per_task;
1358
	struct task_group *tg;
1359
	unsigned long weight;
1360
	int balanced;
1361

1362 1363 1364 1365 1366
	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);
1367

1368 1369 1370 1371 1372
	/*
	 * If sync wakeup then subtract the (maximum possible)
	 * effect of the currently running task from the load
	 * of the current CPU:
	 */
1373
	rcu_read_lock();
1374 1375 1376 1377
	if (sync) {
		tg = task_group(current);
		weight = current->se.load.weight;

1378
		this_load += effective_load(tg, this_cpu, -weight, -weight);
1379 1380
		load += effective_load(tg, prev_cpu, 0, -weight);
	}
1381

1382 1383
	tg = task_group(p);
	weight = p->se.load.weight;
1384

1385 1386
	/*
	 * In low-load situations, where prev_cpu is idle and this_cpu is idle
1387 1388 1389
	 * 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.
1390 1391 1392 1393
	 *
	 * Otherwise check if either cpus are near enough in load to allow this
	 * task to be woken on this_cpu.
	 */
1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408
	if (this_load) {
		unsigned long this_eff_load, prev_eff_load;

		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;
1409
	rcu_read_unlock();
1410

1411
	/*
I
Ingo Molnar 已提交
1412 1413 1414
	 * If the currently running task will sleep within
	 * a reasonable amount of time then attract this newly
	 * woken task:
1415
	 */
1416 1417
	if (sync && balanced)
		return 1;
1418

1419
	schedstat_inc(p, se.statistics.nr_wakeups_affine_attempts);
1420 1421
	tl_per_task = cpu_avg_load_per_task(this_cpu);

1422 1423 1424
	if (balanced ||
	    (this_load <= load &&
	     this_load + target_load(prev_cpu, idx) <= tl_per_task)) {
1425 1426 1427 1428 1429
		/*
		 * This domain has SD_WAKE_AFFINE and
		 * p is cache cold in this domain, and
		 * there is no bad imbalance.
		 */
1430
		schedstat_inc(sd, ttwu_move_affine);
1431
		schedstat_inc(p, se.statistics.nr_wakeups_affine);
1432 1433 1434 1435 1436 1437

		return 1;
	}
	return 0;
}

1438 1439 1440 1441 1442
/*
 * find_idlest_group finds and returns the least busy CPU group within the
 * domain.
 */
static struct sched_group *
P
Peter Zijlstra 已提交
1443
find_idlest_group(struct sched_domain *sd, struct task_struct *p,
1444
		  int this_cpu, int load_idx)
1445
{
1446
	struct sched_group *idlest = NULL, *group = sd->groups;
1447 1448
	unsigned long min_load = ULONG_MAX, this_load = 0;
	int imbalance = 100 + (sd->imbalance_pct-100)/2;
1449

1450 1451 1452 1453
	do {
		unsigned long load, avg_load;
		int local_group;
		int i;
1454

1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508
		/* Skip over this group if it has no CPUs allowed */
		if (!cpumask_intersects(sched_group_cpus(group),
					&p->cpus_allowed))
			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 */
		avg_load = (avg_load * SCHED_LOAD_SCALE) / group->cpu_power;

		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 */
	for_each_cpu_and(i, sched_group_cpus(group), &p->cpus_allowed) {
		load = weighted_cpuload(i);

		if (load < min_load || (load == min_load && i == this_cpu)) {
			min_load = load;
			idlest = i;
1509 1510 1511
		}
	}

1512 1513
	return idlest;
}
1514

1515 1516 1517
/*
 * Try and locate an idle CPU in the sched_domain.
 */
1518
static int select_idle_sibling(struct task_struct *p, int target)
1519 1520 1521
{
	int cpu = smp_processor_id();
	int prev_cpu = task_cpu(p);
1522
	struct sched_domain *sd;
1523 1524 1525
	int i;

	/*
1526 1527
	 * If the task is going to be woken-up on this cpu and if it is
	 * already idle, then it is the right target.
1528
	 */
1529 1530 1531 1532 1533 1534 1535 1536
	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))
1537
		return prev_cpu;
1538 1539

	/*
1540
	 * Otherwise, iterate the domains and find an elegible idle cpu.
1541
	 */
1542 1543
	for_each_domain(target, sd) {
		if (!(sd->flags & SD_SHARE_PKG_RESOURCES))
1544
			break;
1545 1546 1547 1548 1549 1550

		for_each_cpu_and(i, sched_domain_span(sd), &p->cpus_allowed) {
			if (idle_cpu(i)) {
				target = i;
				break;
			}
1551
		}
1552 1553 1554 1555 1556 1557 1558 1559

		/*
		 * Lets stop looking for an idle sibling when we reached
		 * the domain that spans the current cpu and prev_cpu.
		 */
		if (cpumask_test_cpu(cpu, sched_domain_span(sd)) &&
		    cpumask_test_cpu(prev_cpu, sched_domain_span(sd)))
			break;
1560 1561 1562 1563 1564
	}

	return target;
}

1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575
/*
 * 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.
 */
1576 1577
static int
select_task_rq_fair(struct rq *rq, struct task_struct *p, int sd_flag, int wake_flags)
1578
{
1579
	struct sched_domain *tmp, *affine_sd = NULL, *sd = NULL;
1580 1581 1582
	int cpu = smp_processor_id();
	int prev_cpu = task_cpu(p);
	int new_cpu = cpu;
1583
	int want_affine = 0;
1584
	int want_sd = 1;
1585
	int sync = wake_flags & WF_SYNC;
1586

1587
	if (sd_flag & SD_BALANCE_WAKE) {
1588
		if (cpumask_test_cpu(cpu, &p->cpus_allowed))
1589 1590 1591
			want_affine = 1;
		new_cpu = prev_cpu;
	}
1592 1593

	for_each_domain(cpu, tmp) {
1594 1595 1596
		if (!(tmp->flags & SD_LOAD_BALANCE))
			continue;

1597
		/*
1598 1599
		 * If power savings logic is enabled for a domain, see if we
		 * are not overloaded, if so, don't balance wider.
1600
		 */
P
Peter Zijlstra 已提交
1601
		if (tmp->flags & (SD_POWERSAVINGS_BALANCE|SD_PREFER_LOCAL)) {
1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613
			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;
			}

			capacity = DIV_ROUND_CLOSEST(power, SCHED_LOAD_SCALE);

P
Peter Zijlstra 已提交
1614 1615 1616 1617
			if (tmp->flags & SD_POWERSAVINGS_BALANCE)
				nr_running /= 2;

			if (nr_running < capacity)
1618
				want_sd = 0;
1619
		}
1620

1621
		/*
1622 1623
		 * If both cpu and prev_cpu are part of this domain,
		 * cpu is a valid SD_WAKE_AFFINE target.
1624
		 */
1625 1626 1627 1628
		if (want_affine && (tmp->flags & SD_WAKE_AFFINE) &&
		    cpumask_test_cpu(prev_cpu, sched_domain_span(tmp))) {
			affine_sd = tmp;
			want_affine = 0;
1629 1630
		}

1631 1632 1633
		if (!want_sd && !want_affine)
			break;

1634
		if (!(tmp->flags & sd_flag))
1635 1636
			continue;

1637 1638 1639 1640
		if (want_sd)
			sd = tmp;
	}

1641
	if (affine_sd) {
1642 1643 1644 1645
		if (cpu == prev_cpu || wake_affine(affine_sd, p, sync))
			return select_idle_sibling(p, cpu);
		else
			return select_idle_sibling(p, prev_cpu);
1646
	}
1647

1648
	while (sd) {
1649
		int load_idx = sd->forkexec_idx;
1650
		struct sched_group *group;
1651
		int weight;
1652

1653
		if (!(sd->flags & sd_flag)) {
1654 1655 1656
			sd = sd->child;
			continue;
		}
1657

1658 1659
		if (sd_flag & SD_BALANCE_WAKE)
			load_idx = sd->wake_idx;
1660

1661
		group = find_idlest_group(sd, p, cpu, load_idx);
1662 1663 1664 1665
		if (!group) {
			sd = sd->child;
			continue;
		}
I
Ingo Molnar 已提交
1666

1667
		new_cpu = find_idlest_cpu(group, p, cpu);
1668 1669 1670 1671
		if (new_cpu == -1 || new_cpu == cpu) {
			/* Now try balancing at a lower domain level of cpu */
			sd = sd->child;
			continue;
1672
		}
1673 1674 1675

		/* Now try balancing at a lower domain level of new_cpu */
		cpu = new_cpu;
1676
		weight = sd->span_weight;
1677 1678
		sd = NULL;
		for_each_domain(cpu, tmp) {
1679
			if (weight <= tmp->span_weight)
1680
				break;
1681
			if (tmp->flags & sd_flag)
1682 1683 1684
				sd = tmp;
		}
		/* while loop will break here if sd == NULL */
1685 1686
	}

1687
	return new_cpu;
1688 1689 1690
}
#endif /* CONFIG_SMP */

P
Peter Zijlstra 已提交
1691 1692
static unsigned long
wakeup_gran(struct sched_entity *curr, struct sched_entity *se)
1693 1694 1695 1696
{
	unsigned long gran = sysctl_sched_wakeup_granularity;

	/*
P
Peter Zijlstra 已提交
1697 1698
	 * Since its curr running now, convert the gran from real-time
	 * to virtual-time in his units.
M
Mike Galbraith 已提交
1699 1700 1701 1702 1703 1704 1705 1706 1707
	 *
	 * 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.
1708
	 */
M
Mike Galbraith 已提交
1709 1710
	if (unlikely(se->load.weight != NICE_0_LOAD))
		gran = calc_delta_fair(gran, se);
1711 1712 1713 1714

	return gran;
}

1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736
/*
 * 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
1737
	gran = wakeup_gran(curr, se);
1738 1739 1740 1741 1742 1743
	if (vdiff > gran)
		return 1;

	return 0;
}

1744 1745
static void set_last_buddy(struct sched_entity *se)
{
1746 1747 1748 1749
	if (likely(task_of(se)->policy != SCHED_IDLE)) {
		for_each_sched_entity(se)
			cfs_rq_of(se)->last = se;
	}
1750 1751 1752 1753
}

static void set_next_buddy(struct sched_entity *se)
{
1754 1755 1756 1757
	if (likely(task_of(se)->policy != SCHED_IDLE)) {
		for_each_sched_entity(se)
			cfs_rq_of(se)->next = se;
	}
1758 1759
}

1760 1761 1762
/*
 * Preempt the current task with a newly woken task if needed:
 */
1763
static void check_preempt_wakeup(struct rq *rq, struct task_struct *p, int wake_flags)
1764 1765
{
	struct task_struct *curr = rq->curr;
1766
	struct sched_entity *se = &curr->se, *pse = &p->se;
1767
	struct cfs_rq *cfs_rq = task_cfs_rq(curr);
1768
	int scale = cfs_rq->nr_running >= sched_nr_latency;
1769

1770 1771
	if (unlikely(rt_prio(p->prio)))
		goto preempt;
1772

P
Peter Zijlstra 已提交
1773 1774 1775
	if (unlikely(p->sched_class != &fair_sched_class))
		return;

I
Ingo Molnar 已提交
1776 1777 1778
	if (unlikely(se == pse))
		return;

1779
	if (sched_feat(NEXT_BUDDY) && scale && !(wake_flags & WF_FORK))
M
Mike Galbraith 已提交
1780
		set_next_buddy(pse);
P
Peter Zijlstra 已提交
1781

1782 1783 1784 1785 1786 1787 1788
	/*
	 * We can come here with TIF_NEED_RESCHED already set from new task
	 * wake up path.
	 */
	if (test_tsk_need_resched(curr))
		return;

1789
	/*
1790
	 * Batch and idle tasks do not preempt (their preemption is driven by
1791 1792
	 * the tick):
	 */
1793
	if (unlikely(p->policy != SCHED_NORMAL))
1794
		return;
1795

1796
	/* Idle tasks are by definition preempted by everybody. */
1797 1798
	if (unlikely(curr->policy == SCHED_IDLE))
		goto preempt;
1799

1800 1801 1802
	if (!sched_feat(WAKEUP_PREEMPT))
		return;

1803
	update_curr(cfs_rq);
1804
	find_matching_se(&se, &pse);
1805
	BUG_ON(!pse);
1806 1807
	if (wakeup_preempt_entity(se, pse) == 1)
		goto preempt;
1808

1809
	return;
1810

1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826
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);
1827 1828
}

1829
static struct task_struct *pick_next_task_fair(struct rq *rq)
1830
{
P
Peter Zijlstra 已提交
1831
	struct task_struct *p;
1832 1833 1834
	struct cfs_rq *cfs_rq = &rq->cfs;
	struct sched_entity *se;

1835
	if (!cfs_rq->nr_running)
1836 1837 1838
		return NULL;

	do {
1839
		se = pick_next_entity(cfs_rq);
1840
		set_next_entity(cfs_rq, se);
1841 1842 1843
		cfs_rq = group_cfs_rq(se);
	} while (cfs_rq);

P
Peter Zijlstra 已提交
1844 1845 1846 1847
	p = task_of(se);
	hrtick_start_fair(rq, p);

	return p;
1848 1849 1850 1851 1852
}

/*
 * Account for a descheduled task:
 */
1853
static void put_prev_task_fair(struct rq *rq, struct task_struct *prev)
1854 1855 1856 1857 1858 1859
{
	struct sched_entity *se = &prev->se;
	struct cfs_rq *cfs_rq;

	for_each_sched_entity(se) {
		cfs_rq = cfs_rq_of(se);
1860
		put_prev_entity(cfs_rq, se);
1861 1862 1863
	}
}

1864
#ifdef CONFIG_SMP
1865 1866 1867 1868
/**************************************************
 * Fair scheduling class load-balancing methods:
 */

1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879
/*
 * 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);
1880 1881 1882 1883

	/* re-arm NEWIDLE balancing when moving tasks */
	src_rq->avg_idle = this_rq->avg_idle = 2*sysctl_sched_migration_cost;
	this_rq->idle_stamp = 0;
1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901
}

/*
 * 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.
	 */
	if (!cpumask_test_cpu(this_cpu, &p->cpus_allowed)) {
1902
		schedstat_inc(p, se.statistics.nr_failed_migrations_affine);
1903 1904 1905 1906 1907
		return 0;
	}
	*all_pinned = 0;

	if (task_running(rq, p)) {
1908
		schedstat_inc(p, se.statistics.nr_failed_migrations_running);
1909 1910 1911 1912 1913 1914 1915 1916 1917
		return 0;
	}

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

1918
	tsk_cache_hot = task_hot(p, rq->clock_task, sd);
1919 1920 1921 1922 1923
	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]);
1924
			schedstat_inc(p, se.statistics.nr_forced_migrations);
1925 1926 1927 1928 1929 1930
		}
#endif
		return 1;
	}

	if (tsk_cache_hot) {
1931
		schedstat_inc(p, se.statistics.nr_failed_migrations_hot);
1932 1933 1934 1935 1936
		return 0;
	}
	return 1;
}

1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972
/*
 * 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) {

			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;
}

1973 1974 1975 1976
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,
1977
	      int *this_best_prio, struct cfs_rq *busiest_cfs_rq)
1978 1979 1980
{
	int loops = 0, pulled = 0, pinned = 0;
	long rem_load_move = max_load_move;
1981
	struct task_struct *p, *n;
1982 1983 1984 1985 1986 1987

	if (max_load_move == 0)
		goto out;

	pinned = 1;

1988 1989 1990
	list_for_each_entry_safe(p, n, &busiest_cfs_rq->tasks, se.group_node) {
		if (loops++ > sysctl_sched_nr_migrate)
			break;
1991

1992 1993 1994
		if ((p->se.load.weight >> 1) > rem_load_move ||
		    !can_migrate_task(p, busiest, this_cpu, sd, idle, &pinned))
			continue;
1995

1996 1997 1998
		pull_task(busiest, p, this_rq, this_cpu);
		pulled++;
		rem_load_move -= p->se.load.weight;
1999 2000

#ifdef CONFIG_PREEMPT
2001 2002 2003 2004 2005 2006 2007
		/*
		 * 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;
2008 2009
#endif

2010 2011 2012 2013 2014 2015 2016
		/*
		 * We only want to steal up to the prescribed amount of
		 * weighted load.
		 */
		if (rem_load_move <= 0)
			break;

2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033
		if (p->prio < *this_best_prio)
			*this_best_prio = p->prio;
	}
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);

	if (all_pinned)
		*all_pinned = pinned;

	return max_load_move - rem_load_move;
}

P
Peter Zijlstra 已提交
2034
#ifdef CONFIG_FAIR_GROUP_SCHED
2035 2036 2037
/*
 * update tg->load_weight by folding this cpu's load_avg
 */
2038
static int update_shares_cpu(struct task_group *tg, int cpu)
2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053
{
	struct cfs_rq *cfs_rq;
	unsigned long flags;
	struct rq *rq;
	long load_avg;

	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);
2054
	update_cfs_load(cfs_rq);
2055 2056 2057 2058 2059 2060 2061 2062 2063 2064

	load_avg = div64_u64(cfs_rq->load_avg, cfs_rq->load_period+1);
	load_avg -= cfs_rq->load_contribution;
	atomic_add(load_avg, &tg->load_weight);
	cfs_rq->load_contribution += load_avg;

	/*
	 * We need to update shares after updating tg->load_weight in
	 * order to adjust the weight of groups with long running tasks.
	 */
2065
	update_cfs_shares(cfs_rq, 0);
2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077

	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();
2078 2079
	for_each_leaf_cfs_rq(rq, cfs_rq)
		update_shares_cpu(cfs_rq->tg, cpu);
2080 2081 2082
	rcu_read_unlock();
}

P
Peter Zijlstra 已提交
2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129
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,
		  int *all_pinned, int *this_best_prio)
{
	long rem_load_move = max_load_move;
	int busiest_cpu = cpu_of(busiest);
	struct task_group *tg;

	rcu_read_lock();
	update_h_load(busiest_cpu);

	list_for_each_entry_rcu(tg, &task_groups, list) {
		struct cfs_rq *busiest_cfs_rq = tg->cfs_rq[busiest_cpu];
		unsigned long busiest_h_load = busiest_cfs_rq->h_load;
		unsigned long busiest_weight = busiest_cfs_rq->load.weight;
		u64 rem_load, moved_load;

		/*
		 * empty group
		 */
		if (!busiest_cfs_rq->task_weight)
			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,
				rem_load, sd, idle, all_pinned, this_best_prio,
				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
2130 2131 2132 2133
static inline void update_shares(int cpu)
{
}

P
Peter Zijlstra 已提交
2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145
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,
		  int *all_pinned, int *this_best_prio)
{
	return balance_tasks(this_rq, this_cpu, busiest,
			max_load_move, sd, idle, all_pinned,
			this_best_prio, &busiest->cfs);
}
#endif

2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157
/*
 * 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)
{
2158
	unsigned long total_load_moved = 0, load_moved;
2159 2160 2161
	int this_best_prio = this_rq->curr->prio;

	do {
2162
		load_moved = load_balance_fair(this_rq, this_cpu, busiest,
2163 2164
				max_load_move - total_load_moved,
				sd, idle, all_pinned, &this_best_prio);
2165 2166

		total_load_moved += load_moved;
2167 2168 2169 2170 2171 2172 2173 2174 2175

#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;
2176 2177 2178 2179

		if (raw_spin_is_contended(&this_rq->lock) ||
				raw_spin_is_contended(&busiest->lock))
			break;
2180
#endif
2181
	} while (load_moved && max_load_move > total_load_moved);
2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201

	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;
2202
	unsigned long this_has_capacity;
2203 2204 2205 2206 2207

	/* Statistics of the busiest group */
	unsigned long max_load;
	unsigned long busiest_load_per_task;
	unsigned long busiest_nr_running;
2208
	unsigned long busiest_group_capacity;
2209
	unsigned long busiest_has_capacity;
2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231

	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;
	int group_imb; /* Is there an imbalance in the group ? */
2232
	int group_has_capacity; /* Is there extra capacity in the group? */
2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 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 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 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 2421 2422 2423
};

/**
 * group_first_cpu - Returns the first cpu in the cpumask of a sched_group.
 * @group: The group whose first cpu is to be returned.
 */
static inline unsigned int group_first_cpu(struct sched_group *group)
{
	return cpumask_first(sched_group_cpus(group));
}

/**
 * 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)
{
	return SCHED_LOAD_SCALE;
}

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)
{
2424
	unsigned long weight = sd->span_weight;
2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442
	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);
2443 2444 2445 2446 2447 2448 2449

	if (unlikely(total < rq->rt_avg)) {
		/* Ensures that power won't end up being negative */
		available = 0;
	} else {
		available = total - rq->rt_avg;
	}
2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460

	if (unlikely((s64)total < SCHED_LOAD_SCALE))
		total = SCHED_LOAD_SCALE;

	total >>= SCHED_LOAD_SHIFT;

	return div_u64(available, total);
}

static void update_cpu_power(struct sched_domain *sd, int cpu)
{
2461
	unsigned long weight = sd->span_weight;
2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473
	unsigned long power = SCHED_LOAD_SCALE;
	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);

		power >>= SCHED_LOAD_SHIFT;
	}

2474 2475 2476 2477 2478 2479 2480 2481 2482
	sdg->cpu_power_orig = power;

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

	power >>= SCHED_LOAD_SHIFT;

2483 2484 2485 2486 2487 2488
	power *= scale_rt_power(cpu);
	power >>= SCHED_LOAD_SHIFT;

	if (!power)
		power = 1;

2489
	cpu_rq(cpu)->cpu_power = power;
2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514
	sdg->cpu_power = power;
}

static void update_group_power(struct sched_domain *sd, int cpu)
{
	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 {
		power += group->cpu_power;
		group = group->next;
	} while (group != child->groups);

	sdg->cpu_power = power;
}

2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533
/*
 * 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)
{
	/*
	 * Only siblings can have significantly less than SCHED_LOAD_SCALE
	 */
	if (sd->level != SD_LV_SIBLING)
		return 0;

	/*
	 * If ~90% of the cpu_power is still there, we're good.
	 */
M
Michael Neuling 已提交
2534
	if (group->cpu_power * 32 > group->cpu_power_orig * 29)
2535 2536 2537 2538 2539
		return 1;

	return 0;
}

2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558
/**
 * 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.
 * @sd_idle: Idle status of the sched_domain containing group.
 * @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,
			enum cpu_idle_type idle, int load_idx, int *sd_idle,
			int local_group, const struct cpumask *cpus,
			int *balance, struct sg_lb_stats *sgs)
{
2559
	unsigned long load, max_cpu_load, min_cpu_load, max_nr_running;
2560 2561
	int i;
	unsigned int balance_cpu = -1, first_idle_cpu = 0;
2562
	unsigned long avg_load_per_task = 0;
2563

2564
	if (local_group)
2565 2566 2567 2568 2569
		balance_cpu = group_first_cpu(group);

	/* Tally up the load of all CPUs in the group */
	max_cpu_load = 0;
	min_cpu_load = ~0UL;
2570
	max_nr_running = 0;
2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583 2584 2585 2586 2587

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

		if (*sd_idle && rq->nr_running)
			*sd_idle = 0;

		/* 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);
2588
			if (load > max_cpu_load) {
2589
				max_cpu_load = load;
2590 2591
				max_nr_running = rq->nr_running;
			}
2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607
			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);

	}

	/*
	 * 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.
	 */
2608 2609 2610 2611 2612 2613
	if (idle != CPU_NEWLY_IDLE && local_group) {
		if (balance_cpu != this_cpu) {
			*balance = 0;
			return;
		}
		update_group_power(sd, this_cpu);
2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627
	}

	/* Adjust by relative CPU power of the group */
	sgs->avg_load = (sgs->group_load * SCHED_LOAD_SCALE) / group->cpu_power;

	/*
	 * Consider the group unbalanced when the imbalance is larger
	 * than the average weight of two tasks.
	 *
	 * 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?
	 */
2628 2629
	if (sgs->sum_nr_running)
		avg_load_per_task = sgs->sum_weighted_load / sgs->sum_nr_running;
2630

2631
	if ((max_cpu_load - min_cpu_load) > 2*avg_load_per_task && max_nr_running > 1)
2632 2633
		sgs->group_imb = 1;

2634
	sgs->group_capacity = DIV_ROUND_CLOSEST(group->cpu_power, SCHED_LOAD_SCALE);
2635 2636
	if (!sgs->group_capacity)
		sgs->group_capacity = fix_small_capacity(sd, group);
2637 2638 2639

	if (sgs->group_capacity > sgs->sum_nr_running)
		sgs->group_has_capacity = 1;
2640 2641
}

2642 2643 2644 2645 2646
/**
 * 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
2647 2648
 * @sgs: sched_group statistics
 * @this_cpu: the current cpu
2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684
 *
 * 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;
}

2685 2686 2687 2688 2689
/**
 * update_sd_lb_stats - Update sched_group's statistics for load balancing.
 * @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
2690
 * @sd_idle: Idle status of the sched_domain containing sg.
2691 2692 2693 2694 2695 2696 2697 2698 2699 2700
 * @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,
			enum cpu_idle_type idle, int *sd_idle,
			const struct cpumask *cpus, int *balance,
			struct sd_lb_stats *sds)
{
	struct sched_domain *child = sd->child;
2701
	struct sched_group *sg = sd->groups;
2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713
	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;

2714
		local_group = cpumask_test_cpu(this_cpu, sched_group_cpus(sg));
2715
		memset(&sgs, 0, sizeof(sgs));
2716
		update_sg_lb_stats(sd, sg, this_cpu, idle, load_idx, sd_idle,
2717 2718
				local_group, cpus, balance, &sgs);

P
Peter Zijlstra 已提交
2719
		if (local_group && !(*balance))
2720 2721 2722
			return;

		sds->total_load += sgs.group_load;
2723
		sds->total_pwr += sg->cpu_power;
2724 2725 2726

		/*
		 * In case the child domain prefers tasks go to siblings
2727
		 * first, lower the sg capacity to one so that we'll try
2728 2729 2730 2731 2732 2733
		 * 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).
2734
		 */
2735
		if (prefer_sibling && !local_group && sds->this_has_capacity)
2736 2737 2738 2739
			sgs.group_capacity = min(sgs.group_capacity, 1UL);

		if (local_group) {
			sds->this_load = sgs.avg_load;
2740
			sds->this = sg;
2741 2742
			sds->this_nr_running = sgs.sum_nr_running;
			sds->this_load_per_task = sgs.sum_weighted_load;
2743
			sds->this_has_capacity = sgs.group_has_capacity;
2744
		} else if (update_sd_pick_busiest(sd, sds, sg, &sgs, this_cpu)) {
2745
			sds->max_load = sgs.avg_load;
2746
			sds->busiest = sg;
2747
			sds->busiest_nr_running = sgs.sum_nr_running;
2748
			sds->busiest_group_capacity = sgs.group_capacity;
2749
			sds->busiest_load_per_task = sgs.sum_weighted_load;
2750
			sds->busiest_has_capacity = sgs.group_has_capacity;
2751 2752 2753
			sds->group_imb = sgs.group_imb;
		}

2754 2755 2756 2757 2758
		update_sd_power_savings_stats(sg, sds, local_group, &sgs);
		sg = sg->next;
	} while (sg != sd->groups);
}

M
Michael Neuling 已提交
2759
int __weak arch_sd_sibling_asym_packing(void)
2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780
{
       return 0*SD_ASYM_PACKING;
}

/**
 * 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.
 *
2781 2782 2783
 * Returns 1 when packing is required and a task should be moved to
 * this CPU.  The amount of the imbalance is returned in *imbalance.
 *
2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807
 * @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;

	*imbalance = DIV_ROUND_CLOSEST(sds->max_load * sds->busiest->cpu_power,
				       SCHED_LOAD_SCALE);
	return 1;
2808 2809 2810 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822
}

/**
 * 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;
2823
	unsigned long scaled_busy_load_per_task;
2824 2825 2826 2827 2828 2829 2830 2831 2832 2833

	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);

2834 2835 2836 2837 2838 2839
	scaled_busy_load_per_task = sds->busiest_load_per_task
						 * SCHED_LOAD_SCALE;
	scaled_busy_load_per_task /= sds->busiest->cpu_power;

	if (sds->max_load - sds->this_load + scaled_busy_load_per_task >=
			(scaled_busy_load_per_task * imbn)) {
2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889
		*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.
	 */

	pwr_now += sds->busiest->cpu_power *
			min(sds->busiest_load_per_task, sds->max_load);
	pwr_now += sds->this->cpu_power *
			min(sds->this_load_per_task, sds->this_load);
	pwr_now /= SCHED_LOAD_SCALE;

	/* Amount of load we'd subtract */
	tmp = (sds->busiest_load_per_task * SCHED_LOAD_SCALE) /
		sds->busiest->cpu_power;
	if (sds->max_load > tmp)
		pwr_move += sds->busiest->cpu_power *
			min(sds->busiest_load_per_task, sds->max_load - tmp);

	/* Amount of load we'd add */
	if (sds->max_load * sds->busiest->cpu_power <
		sds->busiest_load_per_task * SCHED_LOAD_SCALE)
		tmp = (sds->max_load * sds->busiest->cpu_power) /
			sds->this->cpu_power;
	else
		tmp = (sds->busiest_load_per_task * SCHED_LOAD_SCALE) /
			sds->this->cpu_power;
	pwr_move += sds->this->cpu_power *
			min(sds->this_load_per_task, sds->this_load + tmp);
	pwr_move /= SCHED_LOAD_SCALE;

	/* 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)
{
2890 2891 2892 2893 2894 2895 2896 2897
	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);
	}

2898 2899 2900 2901 2902 2903 2904 2905 2906 2907
	/*
	 * 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);
	}

2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930
	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);

		load_above_capacity *= (SCHED_LOAD_SCALE * SCHED_LOAD_SCALE);

		load_above_capacity /= sds->busiest->cpu_power;
	}

	/*
	 * 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);
2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946

	/* How much load to actually move to equalise the imbalance */
	*imbalance = min(max_pull * sds->busiest->cpu_power,
		(sds->avg_load - sds->this_load) * sds->this->cpu_power)
			/ SCHED_LOAD_SCALE;

	/*
	 * if *imbalance is less than the average load per runnable task
	 * there is no gaurantee that any tasks will be moved so we'll have
	 * 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);

}
2947

2948 2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959 2960 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998
/******* 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.
 * @sd_idle: The idleness of sd
 * @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,
		   int *sd_idle, const struct cpumask *cpus, int *balance)
{
	struct sd_lb_stats sds;

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

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

	/* Cases where imbalance does not exist from POV of this_cpu */
	/* 1) this_cpu is not the appropriate cpu to perform load balancing
	 *    at this level.
	 * 2) There is no busy sibling group to pull from.
	 * 3) This group is the busiest group.
	 * 4) This group is more busy than the avg busieness at this
	 *    sched_domain.
	 * 5) The imbalance is within the specified limit.
2999 3000 3001 3002 3003
	 *
	 * Note: when doing newidle balance, if the local group has excess
	 * capacity (i.e. nr_running < group_capacity) and the busiest group
	 * does not have any capacity, we force a load balance to pull tasks
	 * to the local group. In this case, we skip past checks 3, 4 and 5.
3004
	 */
P
Peter Zijlstra 已提交
3005
	if (!(*balance))
3006 3007
		goto ret;

3008 3009 3010 3011
	if ((idle == CPU_IDLE || idle == CPU_NEWLY_IDLE) &&
	    check_asym_packing(sd, &sds, this_cpu, imbalance))
		return sds.busiest;

3012 3013 3014
	if (!sds.busiest || sds.busiest_nr_running == 0)
		goto out_balanced;

3015 3016 3017 3018 3019
	/*  SD_BALANCE_NEWIDLE trumps SMP nice when underutilized */
	if (idle == CPU_NEWLY_IDLE && sds.this_has_capacity &&
			!sds.busiest_has_capacity)
		goto force_balance;

3020 3021 3022 3023 3024 3025 3026 3027 3028 3029 3030
	if (sds.this_load >= sds.max_load)
		goto out_balanced;

	sds.avg_load = (SCHED_LOAD_SCALE * sds.total_load) / sds.total_pwr;

	if (sds.this_load >= sds.avg_load)
		goto out_balanced;

	if (100 * sds.max_load <= sd->imbalance_pct * sds.this_load)
		goto out_balanced;

3031
force_balance:
3032 3033 3034 3035 3036 3037 3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049 3050 3051
	/* 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 *
3052 3053 3054
find_busiest_queue(struct sched_domain *sd, struct sched_group *group,
		   enum cpu_idle_type idle, unsigned long imbalance,
		   const struct cpumask *cpus)
3055 3056 3057 3058 3059 3060 3061 3062 3063 3064
{
	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);
		unsigned long capacity = DIV_ROUND_CLOSEST(power, SCHED_LOAD_SCALE);
		unsigned long wl;

3065 3066 3067
		if (!capacity)
			capacity = fix_small_capacity(sd, group);

3068 3069 3070 3071
		if (!cpumask_test_cpu(i, cpus))
			continue;

		rq = cpu_rq(i);
3072
		wl = weighted_cpuload(i);
3073

3074 3075 3076 3077
		/*
		 * When comparing with imbalance, use weighted_cpuload()
		 * which is not scaled with the cpu power.
		 */
3078 3079 3080
		if (capacity && rq->nr_running == 1 && wl > imbalance)
			continue;

3081 3082 3083 3084 3085 3086 3087 3088
		/*
		 * 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.
		 */
		wl = (wl * SCHED_LOAD_SCALE) / power;

3089 3090 3091 3092 3093 3094 3095 3096 3097 3098 3099 3100 3101 3102 3103 3104 3105 3106
		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. */
static DEFINE_PER_CPU(cpumask_var_t, load_balance_tmpmask);

3107 3108
static int need_active_balance(struct sched_domain *sd, int sd_idle, int idle,
			       int busiest_cpu, int this_cpu)
3109 3110
{
	if (idle == CPU_NEWLY_IDLE) {
3111 3112 3113 3114 3115 3116 3117 3118 3119

		/*
		 * 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;

3120 3121 3122 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
		/*
		 * 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 (!sd_idle && sd->flags & SD_SHARE_CPUPOWER &&
		    !test_sd_parent(sd, SD_POWERSAVINGS_BALANCE))
			return 0;

		if (sched_mc_power_savings < POWERSAVINGS_BALANCE_WAKEUP)
			return 0;
	}

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

3150 3151
static int active_load_balance_cpu_stop(void *data);

3152 3153 3154 3155 3156 3157 3158 3159 3160 3161 3162 3163 3164 3165 3166 3167 3168 3169 3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180 3181 3182 3183 3184 3185 3186 3187 3188 3189 3190 3191 3192
/*
 * 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)
{
	int ld_moved, all_pinned = 0, active_balance = 0, sd_idle = 0;
	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);

	/*
	 * When power savings policy is enabled for the parent domain, idle
	 * sibling can pick up load irrespective of busy siblings. In this case,
	 * let the state of idle sibling percolate up as CPU_IDLE, instead of
	 * portraying it as CPU_NOT_IDLE.
	 */
	if (idle != CPU_NOT_IDLE && sd->flags & SD_SHARE_CPUPOWER &&
	    !test_sd_parent(sd, SD_POWERSAVINGS_BALANCE))
		sd_idle = 1;

	schedstat_inc(sd, lb_count[idle]);

redo:
	group = find_busiest_group(sd, this_cpu, &imbalance, idle, &sd_idle,
				   cpus, balance);

	if (*balance == 0)
		goto out_balanced;

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

3193
	busiest = find_busiest_queue(sd, group, idle, imbalance, cpus);
3194 3195 3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207 3208 3209 3210 3211 3212 3213 3214 3215 3216 3217 3218 3219 3220 3221 3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 3232 3233 3234
	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.
		 */
		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]);
3235 3236 3237 3238 3239 3240 3241 3242
		/*
		 * 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++;
3243

3244 3245
		if (need_active_balance(sd, sd_idle, idle, cpu_of(busiest),
					this_cpu)) {
3246 3247
			raw_spin_lock_irqsave(&busiest->lock, flags);

3248 3249 3250
			/* don't kick the active_load_balance_cpu_stop,
			 * if the curr task on busiest cpu can't be
			 * moved to this_cpu
3251 3252 3253 3254 3255 3256 3257 3258 3259
			 */
			if (!cpumask_test_cpu(this_cpu,
					      &busiest->curr->cpus_allowed)) {
				raw_spin_unlock_irqrestore(&busiest->lock,
							    flags);
				all_pinned = 1;
				goto out_one_pinned;
			}

3260 3261 3262 3263 3264
			/*
			 * ->active_balance synchronizes accesses to
			 * ->active_balance_work.  Once set, it's cleared
			 * only after active load balance is finished.
			 */
3265 3266 3267 3268 3269 3270
			if (!busiest->active_balance) {
				busiest->active_balance = 1;
				busiest->push_cpu = this_cpu;
				active_balance = 1;
			}
			raw_spin_unlock_irqrestore(&busiest->lock, flags);
3271

3272
			if (active_balance)
3273 3274 3275
				stop_one_cpu_nowait(cpu_of(busiest),
					active_load_balance_cpu_stop, busiest,
					&busiest->active_balance_work);
3276 3277 3278 3279 3280 3281 3282 3283 3284 3285 3286 3287 3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298 3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340

			/*
			 * 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;
	}

	if (!ld_moved && !sd_idle && sd->flags & SD_SHARE_CPUPOWER &&
	    !test_sd_parent(sd, SD_POWERSAVINGS_BALANCE))
		ld_moved = -1;

	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;

	if (!sd_idle && sd->flags & SD_SHARE_CPUPOWER &&
	    !test_sd_parent(sd, SD_POWERSAVINGS_BALANCE))
		ld_moved = -1;
	else
		ld_moved = 0;
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.
 */
static void idle_balance(int this_cpu, struct rq *this_rq)
{
	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;

3341 3342 3343 3344 3345
	/*
	 * Drop the rq->lock, but keep IRQ/preempt disabled.
	 */
	raw_spin_unlock(&this_rq->lock);

P
Paul Turner 已提交
3346
	update_shares(this_cpu);
3347 3348
	for_each_domain(this_cpu, sd) {
		unsigned long interval;
3349
		int balance = 1;
3350 3351 3352 3353

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

3354
		if (sd->flags & SD_BALANCE_NEWIDLE) {
3355
			/* If we've pulled tasks over stop searching: */
3356 3357 3358
			pulled_task = load_balance(this_cpu, this_rq,
						   sd, CPU_NEWLY_IDLE, &balance);
		}
3359 3360 3361 3362

		interval = msecs_to_jiffies(sd->balance_interval);
		if (time_after(next_balance, sd->last_balance + interval))
			next_balance = sd->last_balance + interval;
3363
		if (pulled_task)
3364 3365
			break;
	}
3366 3367 3368

	raw_spin_lock(&this_rq->lock);

3369 3370 3371 3372 3373 3374 3375 3376 3377 3378
	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;
	}
}

/*
3379 3380 3381 3382
 * 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.
3383
 */
3384
static int active_load_balance_cpu_stop(void *data)
3385
{
3386 3387
	struct rq *busiest_rq = data;
	int busiest_cpu = cpu_of(busiest_rq);
3388
	int target_cpu = busiest_rq->push_cpu;
3389
	struct rq *target_rq = cpu_rq(target_cpu);
3390
	struct sched_domain *sd;
3391 3392 3393 3394 3395 3396 3397

	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;
3398 3399 3400

	/* Is there any task to move? */
	if (busiest_rq->nr_running <= 1)
3401
		goto out_unlock;
3402 3403 3404 3405 3406 3407 3408 3409 3410 3411 3412 3413 3414 3415 3416 3417 3418 3419 3420 3421 3422 3423 3424 3425 3426 3427 3428 3429

	/*
	 * 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. */
	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);
	}
	double_unlock_balance(busiest_rq, target_rq);
3430 3431 3432 3433
out_unlock:
	busiest_rq->active_balance = 0;
	raw_spin_unlock_irq(&busiest_rq->lock);
	return 0;
3434 3435 3436
}

#ifdef CONFIG_NO_HZ
3437 3438 3439 3440 3441 3442 3443 3444 3445 3446 3447 3448 3449 3450 3451 3452 3453 3454 3455 3456 3457 3458 3459 3460 3461 3462

static DEFINE_PER_CPU(struct call_single_data, remote_sched_softirq_cb);

static void trigger_sched_softirq(void *data)
{
	raise_softirq_irqoff(SCHED_SOFTIRQ);
}

static inline void init_sched_softirq_csd(struct call_single_data *csd)
{
	csd->func = trigger_sched_softirq;
	csd->info = NULL;
	csd->flags = 0;
	csd->priv = 0;
}

/*
 * 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.
 */
3463 3464
static struct {
	atomic_t load_balancer;
3465 3466 3467 3468 3469 3470
	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;
3471 3472 3473 3474 3475 3476 3477 3478 3479 3480 3481 3482 3483 3484 3485 3486 3487 3488 3489 3490 3491 3492 3493 3494 3495 3496 3497 3498 3499 3500 3501 3502 3503 3504 3505 3506 3507 3508 3509 3510 3511 3512 3513 3514 3515 3516 3517 3518 3519 3520 3521 3522 3523

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)
		if (sd && (sd->flags & flag))
			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)
{
3524
	cpumask_and(nohz.grp_idle_mask, nohz.idle_cpus_mask,
3525 3526 3527 3528 3529 3530
					sched_group_cpus(ilb_group));

	/*
	 * A sched_group is semi-idle when it has atleast one busy cpu
	 * and atleast one idle cpu.
	 */
3531
	if (cpumask_empty(nohz.grp_idle_mask))
3532 3533
		return 0;

3534
	if (cpumask_equal(nohz.grp_idle_mask, sched_group_cpus(ilb_group)))
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
		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;

	/*
	 * 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
	 */
3567
	if (cpumask_weight(nohz.idle_cpus_mask) < 2)
3568 3569 3570 3571 3572 3573 3574
		goto out_done;

	for_each_flag_domain(cpu, sd, SD_POWERSAVINGS_BALANCE) {
		ilb_group = sd->groups;

		do {
			if (is_semi_idle_group(ilb_group))
3575
				return cpumask_first(nohz.grp_idle_mask);
3576 3577 3578 3579 3580 3581 3582

			ilb_group = ilb_group->next;

		} while (ilb_group != sd->groups);
	}

out_done:
3583
	return nr_cpu_ids;
3584 3585 3586 3587
}
#else /*  (CONFIG_SCHED_MC || CONFIG_SCHED_SMT) */
static inline int find_new_ilb(int call_cpu)
{
3588
	return nr_cpu_ids;
3589 3590 3591
}
#endif

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
/*
 * 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;
	}

	if (!cpu_rq(ilb_cpu)->nohz_balance_kick) {
		struct call_single_data *cp;

		cpu_rq(ilb_cpu)->nohz_balance_kick = 1;
		cp = &per_cpu(remote_sched_softirq_cb, cpu);
		__smp_call_function_single(ilb_cpu, cp, 0);
	}
	return;
}

3621 3622 3623
/*
 * 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
3624
 * load balancing on behalf of all those cpus.
3625
 *
3626 3627 3628
 * 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.
3629
 *
3630 3631 3632
 * 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).
3633
 */
3634
void select_nohz_load_balancer(int stop_tick)
3635 3636 3637 3638 3639 3640
{
	int cpu = smp_processor_id();

	if (stop_tick) {
		if (!cpu_active(cpu)) {
			if (atomic_read(&nohz.load_balancer) != cpu)
3641
				return;
3642 3643 3644 3645 3646

			/*
			 * If we are going offline and still the leader,
			 * give up!
			 */
3647 3648
			if (atomic_cmpxchg(&nohz.load_balancer, cpu,
					   nr_cpu_ids) != cpu)
3649 3650
				BUG();

3651
			return;
3652 3653
		}

3654
		cpumask_set_cpu(cpu, nohz.idle_cpus_mask);
3655

3656 3657 3658 3659
		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);
3660

3661
		if (atomic_read(&nohz.load_balancer) >= nr_cpu_ids) {
3662 3663
			int new_ilb;

3664 3665 3666 3667 3668
			/* make me the ilb owner */
			if (atomic_cmpxchg(&nohz.load_balancer, nr_cpu_ids,
					   cpu) != nr_cpu_ids)
				return;

3669 3670 3671 3672 3673 3674
			/*
			 * 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) {
3675
				atomic_set(&nohz.load_balancer, nr_cpu_ids);
3676
				resched_cpu(new_ilb);
3677
				return;
3678
			}
3679
			return;
3680 3681
		}
	} else {
3682 3683
		if (!cpumask_test_cpu(cpu, nohz.idle_cpus_mask))
			return;
3684

3685
		cpumask_clear_cpu(cpu, nohz.idle_cpus_mask);
3686 3687

		if (atomic_read(&nohz.load_balancer) == cpu)
3688 3689
			if (atomic_cmpxchg(&nohz.load_balancer, cpu,
					   nr_cpu_ids) != cpu)
3690 3691
				BUG();
	}
3692
	return;
3693 3694 3695 3696 3697 3698 3699 3700 3701 3702 3703 3704 3705 3706 3707 3708 3709 3710 3711 3712 3713 3714
}
#endif

static DEFINE_SPINLOCK(balancing);

/*
 * 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
3715 3716
	update_shares(cpu);

3717 3718 3719 3720 3721 3722 3723 3724 3725 3726 3727 3728 3729 3730 3731 3732 3733 3734 3735 3736 3737 3738 3739 3740 3741 3742 3743 3744 3745 3746 3747 3748 3749 3750 3751 3752 3753 3754 3755 3756 3757 3758 3759 3760 3761 3762 3763 3764 3765 3766 3767 3768 3769 3770 3771 3772 3773 3774 3775
	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);
		if (unlikely(!interval))
			interval = 1;
		if (interval > HZ*NR_CPUS/10)
			interval = HZ*NR_CPUS/10;

		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
				 * longer idle, or one of our SMT siblings is
				 * not idle.
				 */
				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;
	}

	/*
	 * 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;
}

3776
#ifdef CONFIG_NO_HZ
3777
/*
3778
 * In CONFIG_NO_HZ case, the idle balance kickee will do the
3779 3780
 * rebalancing for all the cpus for whom scheduler ticks are stopped.
 */
3781 3782 3783 3784 3785 3786 3787 3788 3789 3790 3791 3792 3793 3794 3795 3796 3797 3798 3799 3800 3801 3802 3803 3804
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;

	if (idle != CPU_IDLE || !this_rq->nohz_balance_kick)
		return;

	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.
		 */
		if (need_resched()) {
			this_rq->nohz_balance_kick = 0;
			break;
		}

		raw_spin_lock_irq(&this_rq->lock);
3805
		update_rq_clock(this_rq);
3806 3807 3808 3809 3810 3811 3812 3813 3814 3815 3816 3817 3818 3819 3820 3821 3822 3823 3824 3825 3826 3827 3828 3829 3830 3831 3832 3833 3834 3835 3836 3837 3838 3839
		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;
	this_rq->nohz_balance_kick = 0;
}

/*
 * 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;

	if (time_before(now, nohz.next_balance))
		return 0;

S
Suresh Siddha 已提交
3840
	if (rq->idle_at_tick)
3841 3842 3843 3844 3845 3846 3847 3848 3849 3850 3851 3852 3853 3854 3855 3856 3857 3858 3859 3860 3861 3862 3863 3864 3865 3866 3867 3868 3869 3870 3871
		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).
 */
3872 3873 3874 3875 3876 3877 3878 3879 3880 3881
static void run_rebalance_domains(struct softirq_action *h)
{
	int this_cpu = smp_processor_id();
	struct rq *this_rq = cpu_rq(this_cpu);
	enum cpu_idle_type idle = this_rq->idle_at_tick ?
						CPU_IDLE : CPU_NOT_IDLE;

	rebalance_domains(this_cpu, idle);

	/*
3882
	 * If this cpu has a pending nohz_balance_kick, then do the
3883 3884 3885
	 * balancing on behalf of the other idle cpus whose ticks are
	 * stopped.
	 */
3886
	nohz_idle_balance(this_cpu, idle);
3887 3888 3889 3890
}

static inline int on_null_domain(int cpu)
{
3891
	return !rcu_dereference_sched(cpu_rq(cpu)->sd);
3892 3893 3894 3895 3896 3897 3898 3899 3900 3901 3902
}

/*
 * Trigger the SCHED_SOFTIRQ if it is time to do periodic load balancing.
 */
static inline void trigger_load_balance(struct rq *rq, int cpu)
{
	/* 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);
3903 3904 3905 3906
#ifdef CONFIG_NO_HZ
	else if (nohz_kick_needed(rq, cpu) && likely(!on_null_domain(cpu)))
		nohz_balancer_kick(cpu);
#endif
3907 3908
}

3909 3910 3911 3912 3913 3914 3915 3916 3917 3918
static void rq_online_fair(struct rq *rq)
{
	update_sysctl();
}

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

3919 3920 3921 3922 3923 3924 3925 3926 3927
#else	/* CONFIG_SMP */

/*
 * on UP we do not need to balance between CPUs:
 */
static inline void idle_balance(int cpu, struct rq *rq)
{
}

3928
#endif /* CONFIG_SMP */
3929

3930 3931 3932
/*
 * scheduler tick hitting a task of our scheduling class:
 */
P
Peter Zijlstra 已提交
3933
static void task_tick_fair(struct rq *rq, struct task_struct *curr, int queued)
3934 3935 3936 3937 3938 3939
{
	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
3940
		entity_tick(cfs_rq, se, queued);
3941 3942 3943 3944
	}
}

/*
P
Peter Zijlstra 已提交
3945 3946 3947
 * called on fork with the child task as argument from the parent's context
 *  - child not yet on the tasklist
 *  - preemption disabled
3948
 */
P
Peter Zijlstra 已提交
3949
static void task_fork_fair(struct task_struct *p)
3950
{
P
Peter Zijlstra 已提交
3951
	struct cfs_rq *cfs_rq = task_cfs_rq(current);
3952
	struct sched_entity *se = &p->se, *curr = cfs_rq->curr;
3953
	int this_cpu = smp_processor_id();
P
Peter Zijlstra 已提交
3954 3955 3956
	struct rq *rq = this_rq();
	unsigned long flags;

3957
	raw_spin_lock_irqsave(&rq->lock, flags);
3958

3959 3960
	update_rq_clock(rq);

3961 3962
	if (unlikely(task_cpu(p) != this_cpu)) {
		rcu_read_lock();
P
Peter Zijlstra 已提交
3963
		__set_task_cpu(p, this_cpu);
3964 3965
		rcu_read_unlock();
	}
3966

3967
	update_curr(cfs_rq);
P
Peter Zijlstra 已提交
3968

3969 3970
	if (curr)
		se->vruntime = curr->vruntime;
3971
	place_entity(cfs_rq, se, 1);
3972

P
Peter Zijlstra 已提交
3973
	if (sysctl_sched_child_runs_first && curr && entity_before(curr, se)) {
D
Dmitry Adamushko 已提交
3974
		/*
3975 3976 3977
		 * Upon rescheduling, sched_class::put_prev_task() will place
		 * 'current' within the tree based on its new key value.
		 */
3978
		swap(curr->vruntime, se->vruntime);
3979
		resched_task(rq->curr);
3980
	}
3981

3982 3983
	se->vruntime -= cfs_rq->min_vruntime;

3984
	raw_spin_unlock_irqrestore(&rq->lock, flags);
3985 3986
}

3987 3988 3989 3990 3991 3992 3993 3994 3995 3996 3997 3998 3999 4000 4001 4002
/*
 * Priority of the task has changed. Check to see if we preempt
 * the current task.
 */
static void prio_changed_fair(struct rq *rq, struct task_struct *p,
			      int oldprio, int running)
{
	/*
	 * 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
	 */
	if (running) {
		if (p->prio > oldprio)
			resched_task(rq->curr);
	} else
4003
		check_preempt_curr(rq, p, 0);
4004 4005 4006 4007 4008 4009 4010 4011 4012 4013 4014 4015 4016 4017 4018 4019
}

/*
 * We switched to the sched_fair class.
 */
static void switched_to_fair(struct rq *rq, struct task_struct *p,
			     int running)
{
	/*
	 * 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.
	 */
	if (running)
		resched_task(rq->curr);
	else
4020
		check_preempt_curr(rq, p, 0);
4021 4022
}

4023 4024 4025 4026 4027 4028 4029 4030 4031 4032 4033 4034 4035
/* 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;

	for_each_sched_entity(se)
		set_next_entity(cfs_rq_of(se), se);
}

P
Peter Zijlstra 已提交
4036
#ifdef CONFIG_FAIR_GROUP_SCHED
4037
static void task_move_group_fair(struct task_struct *p, int on_rq)
P
Peter Zijlstra 已提交
4038
{
4039 4040 4041 4042 4043 4044 4045 4046 4047 4048 4049 4050 4051 4052 4053 4054
	/*
	 * 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));
4055
	if (!on_rq)
4056
		p->se.vruntime += cfs_rq_of(&p->se)->min_vruntime;
P
Peter Zijlstra 已提交
4057 4058 4059
}
#endif

4060
static unsigned int get_rr_interval_fair(struct rq *rq, struct task_struct *task)
4061 4062 4063 4064 4065 4066 4067 4068 4069 4070 4071 4072 4073 4074
{
	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;
}

4075 4076 4077
/*
 * All the scheduling class methods:
 */
4078 4079
static const struct sched_class fair_sched_class = {
	.next			= &idle_sched_class,
4080 4081 4082 4083
	.enqueue_task		= enqueue_task_fair,
	.dequeue_task		= dequeue_task_fair,
	.yield_task		= yield_task_fair,

I
Ingo Molnar 已提交
4084
	.check_preempt_curr	= check_preempt_wakeup,
4085 4086 4087 4088

	.pick_next_task		= pick_next_task_fair,
	.put_prev_task		= put_prev_task_fair,

4089
#ifdef CONFIG_SMP
L
Li Zefan 已提交
4090 4091
	.select_task_rq		= select_task_rq_fair,

4092 4093
	.rq_online		= rq_online_fair,
	.rq_offline		= rq_offline_fair,
4094 4095

	.task_waking		= task_waking_fair,
4096
#endif
4097

4098
	.set_curr_task          = set_curr_task_fair,
4099
	.task_tick		= task_tick_fair,
P
Peter Zijlstra 已提交
4100
	.task_fork		= task_fork_fair,
4101 4102 4103

	.prio_changed		= prio_changed_fair,
	.switched_to		= switched_to_fair,
P
Peter Zijlstra 已提交
4104

4105 4106
	.get_rr_interval	= get_rr_interval_fair,

P
Peter Zijlstra 已提交
4107
#ifdef CONFIG_FAIR_GROUP_SCHED
4108
	.task_move_group	= task_move_group_fair,
P
Peter Zijlstra 已提交
4109
#endif
4110 4111 4112
};

#ifdef CONFIG_SCHED_DEBUG
4113
static void print_cfs_stats(struct seq_file *m, int cpu)
4114 4115 4116
{
	struct cfs_rq *cfs_rq;

4117
	rcu_read_lock();
4118
	for_each_leaf_cfs_rq(cpu_rq(cpu), cfs_rq)
4119
		print_cfs_rq(m, cpu, cfs_rq);
4120
	rcu_read_unlock();
4121 4122
}
#endif