2 * Completely Fair Scheduling (CFS) Class (SCHED_NORMAL/SCHED_BATCH)
4 * Copyright (C) 2007 Red Hat, Inc., Ingo Molnar <mingo@redhat.com>
6 * Interactivity improvements by Mike Galbraith
7 * (C) 2007 Mike Galbraith <efault@gmx.de>
9 * Various enhancements by Dmitry Adamushko.
10 * (C) 2007 Dmitry Adamushko <dmitry.adamushko@gmail.com>
12 * Group scheduling enhancements by Srivatsa Vaddagiri
13 * Copyright IBM Corporation, 2007
14 * Author: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com>
16 * Scaled math optimizations by Thomas Gleixner
17 * Copyright (C) 2007, Thomas Gleixner <tglx@linutronix.de>
19 * Adaptive scheduling granularity, math enhancements by Peter Zijlstra
20 * Copyright (C) 2007 Red Hat, Inc., Peter Zijlstra <pzijlstr@redhat.com>
23 #include <linux/latencytop.h>
26 * Targeted preemption latency for CPU-bound tasks:
27 * (default: 20ms * (1 + ilog(ncpus)), units: nanoseconds)
29 * NOTE: this latency value is not the same as the concept of
30 * 'timeslice length' - timeslices in CFS are of variable length
31 * and have no persistent notion like in traditional, time-slice
32 * based scheduling concepts.
34 * (to see the precise effective timeslice length of your workload,
35 * run vmstat and monitor the context-switches (cs) field)
37 unsigned int sysctl_sched_latency
= 20000000ULL;
40 * Minimal preemption granularity for CPU-bound tasks:
41 * (default: 4 msec * (1 + ilog(ncpus)), units: nanoseconds)
43 unsigned int sysctl_sched_min_granularity
= 4000000ULL;
46 * is kept at sysctl_sched_latency / sysctl_sched_min_granularity
48 static unsigned int sched_nr_latency
= 5;
51 * After fork, child runs first. (default) If set to 0 then
52 * parent will (try to) run first.
54 const_debug
unsigned int sysctl_sched_child_runs_first
= 1;
57 * sys_sched_yield() compat mode
59 * This option switches the agressive yield implementation of the
60 * old scheduler back on.
62 unsigned int __read_mostly sysctl_sched_compat_yield
;
65 * SCHED_OTHER wake-up granularity.
66 * (default: 10 msec * (1 + ilog(ncpus)), units: nanoseconds)
68 * This option delays the preemption effects of decoupled workloads
69 * and reduces their over-scheduling. Synchronous workloads will still
70 * have immediate wakeup/sleep latencies.
72 unsigned int sysctl_sched_wakeup_granularity
= 10000000UL;
74 const_debug
unsigned int sysctl_sched_migration_cost
= 500000UL;
76 /**************************************************************
77 * CFS operations on generic schedulable entities:
80 static inline struct task_struct
*task_of(struct sched_entity
*se
)
82 return container_of(se
, struct task_struct
, se
);
85 #ifdef CONFIG_FAIR_GROUP_SCHED
87 /* cpu runqueue to which this cfs_rq is attached */
88 static inline struct rq
*rq_of(struct cfs_rq
*cfs_rq
)
93 /* An entity is a task if it doesn't "own" a runqueue */
94 #define entity_is_task(se) (!se->my_q)
96 /* Walk up scheduling entities hierarchy */
97 #define for_each_sched_entity(se) \
98 for (; se; se = se->parent)
100 static inline struct cfs_rq
*task_cfs_rq(struct task_struct
*p
)
105 /* runqueue on which this entity is (to be) queued */
106 static inline struct cfs_rq
*cfs_rq_of(struct sched_entity
*se
)
111 /* runqueue "owned" by this group */
112 static inline struct cfs_rq
*group_cfs_rq(struct sched_entity
*grp
)
117 /* Given a group's cfs_rq on one cpu, return its corresponding cfs_rq on
118 * another cpu ('this_cpu')
120 static inline struct cfs_rq
*cpu_cfs_rq(struct cfs_rq
*cfs_rq
, int this_cpu
)
122 return cfs_rq
->tg
->cfs_rq
[this_cpu
];
125 /* Iterate thr' all leaf cfs_rq's on a runqueue */
126 #define for_each_leaf_cfs_rq(rq, cfs_rq) \
127 list_for_each_entry_rcu(cfs_rq, &rq->leaf_cfs_rq_list, leaf_cfs_rq_list)
129 /* Do the two (enqueued) entities belong to the same group ? */
131 is_same_group(struct sched_entity
*se
, struct sched_entity
*pse
)
133 if (se
->cfs_rq
== pse
->cfs_rq
)
139 static inline struct sched_entity
*parent_entity(struct sched_entity
*se
)
144 #else /* CONFIG_FAIR_GROUP_SCHED */
146 static inline struct rq
*rq_of(struct cfs_rq
*cfs_rq
)
148 return container_of(cfs_rq
, struct rq
, cfs
);
151 #define entity_is_task(se) 1
153 #define for_each_sched_entity(se) \
154 for (; se; se = NULL)
156 static inline struct cfs_rq
*task_cfs_rq(struct task_struct
*p
)
158 return &task_rq(p
)->cfs
;
161 static inline struct cfs_rq
*cfs_rq_of(struct sched_entity
*se
)
163 struct task_struct
*p
= task_of(se
);
164 struct rq
*rq
= task_rq(p
);
169 /* runqueue "owned" by this group */
170 static inline struct cfs_rq
*group_cfs_rq(struct sched_entity
*grp
)
175 static inline struct cfs_rq
*cpu_cfs_rq(struct cfs_rq
*cfs_rq
, int this_cpu
)
177 return &cpu_rq(this_cpu
)->cfs
;
180 #define for_each_leaf_cfs_rq(rq, cfs_rq) \
181 for (cfs_rq = &rq->cfs; cfs_rq; cfs_rq = NULL)
184 is_same_group(struct sched_entity
*se
, struct sched_entity
*pse
)
189 static inline struct sched_entity
*parent_entity(struct sched_entity
*se
)
194 #endif /* CONFIG_FAIR_GROUP_SCHED */
197 /**************************************************************
198 * Scheduling class tree data structure manipulation methods:
201 static inline u64
max_vruntime(u64 min_vruntime
, u64 vruntime
)
203 s64 delta
= (s64
)(vruntime
- min_vruntime
);
205 min_vruntime
= vruntime
;
210 static inline u64
min_vruntime(u64 min_vruntime
, u64 vruntime
)
212 s64 delta
= (s64
)(vruntime
- min_vruntime
);
214 min_vruntime
= vruntime
;
219 static inline s64
entity_key(struct cfs_rq
*cfs_rq
, struct sched_entity
*se
)
221 return se
->vruntime
- cfs_rq
->min_vruntime
;
225 * Enqueue an entity into the rb-tree:
227 static void __enqueue_entity(struct cfs_rq
*cfs_rq
, struct sched_entity
*se
)
229 struct rb_node
**link
= &cfs_rq
->tasks_timeline
.rb_node
;
230 struct rb_node
*parent
= NULL
;
231 struct sched_entity
*entry
;
232 s64 key
= entity_key(cfs_rq
, se
);
236 * Find the right place in the rbtree:
240 entry
= rb_entry(parent
, struct sched_entity
, run_node
);
242 * We dont care about collisions. Nodes with
243 * the same key stay together.
245 if (key
< entity_key(cfs_rq
, entry
)) {
246 link
= &parent
->rb_left
;
248 link
= &parent
->rb_right
;
254 * Maintain a cache of leftmost tree entries (it is frequently
258 cfs_rq
->rb_leftmost
= &se
->run_node
;
260 * maintain cfs_rq->min_vruntime to be a monotonic increasing
261 * value tracking the leftmost vruntime in the tree.
263 cfs_rq
->min_vruntime
=
264 max_vruntime(cfs_rq
->min_vruntime
, se
->vruntime
);
267 rb_link_node(&se
->run_node
, parent
, link
);
268 rb_insert_color(&se
->run_node
, &cfs_rq
->tasks_timeline
);
271 static void __dequeue_entity(struct cfs_rq
*cfs_rq
, struct sched_entity
*se
)
273 if (cfs_rq
->rb_leftmost
== &se
->run_node
) {
274 struct rb_node
*next_node
;
275 struct sched_entity
*next
;
277 next_node
= rb_next(&se
->run_node
);
278 cfs_rq
->rb_leftmost
= next_node
;
281 next
= rb_entry(next_node
,
282 struct sched_entity
, run_node
);
283 cfs_rq
->min_vruntime
=
284 max_vruntime(cfs_rq
->min_vruntime
,
289 if (cfs_rq
->next
== se
)
292 rb_erase(&se
->run_node
, &cfs_rq
->tasks_timeline
);
295 static inline struct rb_node
*first_fair(struct cfs_rq
*cfs_rq
)
297 return cfs_rq
->rb_leftmost
;
300 static struct sched_entity
*__pick_next_entity(struct cfs_rq
*cfs_rq
)
302 return rb_entry(first_fair(cfs_rq
), struct sched_entity
, run_node
);
305 static inline struct sched_entity
*__pick_last_entity(struct cfs_rq
*cfs_rq
)
307 struct rb_node
*last
= rb_last(&cfs_rq
->tasks_timeline
);
312 return rb_entry(last
, struct sched_entity
, run_node
);
315 /**************************************************************
316 * Scheduling class statistics methods:
319 #ifdef CONFIG_SCHED_DEBUG
320 int sched_nr_latency_handler(struct ctl_table
*table
, int write
,
321 struct file
*filp
, void __user
*buffer
, size_t *lenp
,
324 int ret
= proc_dointvec_minmax(table
, write
, filp
, buffer
, lenp
, ppos
);
329 sched_nr_latency
= DIV_ROUND_UP(sysctl_sched_latency
,
330 sysctl_sched_min_granularity
);
337 * The idea is to set a period in which each task runs once.
339 * When there are too many tasks (sysctl_sched_nr_latency) we have to stretch
340 * this period because otherwise the slices get too small.
342 * p = (nr <= nl) ? l : l*nr/nl
344 static u64
__sched_period(unsigned long nr_running
)
346 u64 period
= sysctl_sched_latency
;
347 unsigned long nr_latency
= sched_nr_latency
;
349 if (unlikely(nr_running
> nr_latency
)) {
350 period
= sysctl_sched_min_granularity
;
351 period
*= nr_running
;
358 * We calculate the wall-time slice from the period by taking a part
359 * proportional to the weight.
363 static u64
sched_slice(struct cfs_rq
*cfs_rq
, struct sched_entity
*se
)
365 u64 slice
= __sched_period(cfs_rq
->nr_running
);
367 for_each_sched_entity(se
) {
368 cfs_rq
= cfs_rq_of(se
);
370 slice
*= se
->load
.weight
;
371 do_div(slice
, cfs_rq
->load
.weight
);
379 * We calculate the vruntime slice of a to be inserted task
383 static u64
sched_vslice_add(struct cfs_rq
*cfs_rq
, struct sched_entity
*se
)
385 unsigned long nr_running
= cfs_rq
->nr_running
;
386 unsigned long weight
;
392 vslice
= __sched_period(nr_running
);
394 for_each_sched_entity(se
) {
395 cfs_rq
= cfs_rq_of(se
);
397 weight
= cfs_rq
->load
.weight
;
399 weight
+= se
->load
.weight
;
401 vslice
*= NICE_0_LOAD
;
402 do_div(vslice
, weight
);
409 * Update the current task's runtime statistics. Skip current tasks that
410 * are not in our scheduling class.
413 __update_curr(struct cfs_rq
*cfs_rq
, struct sched_entity
*curr
,
414 unsigned long delta_exec
)
416 unsigned long delta_exec_weighted
;
418 schedstat_set(curr
->exec_max
, max((u64
)delta_exec
, curr
->exec_max
));
420 curr
->sum_exec_runtime
+= delta_exec
;
421 schedstat_add(cfs_rq
, exec_clock
, delta_exec
);
422 delta_exec_weighted
= delta_exec
;
423 if (unlikely(curr
->load
.weight
!= NICE_0_LOAD
)) {
424 delta_exec_weighted
= calc_delta_fair(delta_exec_weighted
,
427 curr
->vruntime
+= delta_exec_weighted
;
430 static void update_curr(struct cfs_rq
*cfs_rq
)
432 struct sched_entity
*curr
= cfs_rq
->curr
;
433 u64 now
= rq_of(cfs_rq
)->clock
;
434 unsigned long delta_exec
;
440 * Get the amount of time the current task was running
441 * since the last time we changed load (this cannot
442 * overflow on 32 bits):
444 delta_exec
= (unsigned long)(now
- curr
->exec_start
);
446 __update_curr(cfs_rq
, curr
, delta_exec
);
447 curr
->exec_start
= now
;
449 if (entity_is_task(curr
)) {
450 struct task_struct
*curtask
= task_of(curr
);
452 cpuacct_charge(curtask
, delta_exec
);
457 update_stats_wait_start(struct cfs_rq
*cfs_rq
, struct sched_entity
*se
)
459 schedstat_set(se
->wait_start
, rq_of(cfs_rq
)->clock
);
463 * Task is being enqueued - update stats:
465 static void update_stats_enqueue(struct cfs_rq
*cfs_rq
, struct sched_entity
*se
)
468 * Are we enqueueing a waiting task? (for current tasks
469 * a dequeue/enqueue event is a NOP)
471 if (se
!= cfs_rq
->curr
)
472 update_stats_wait_start(cfs_rq
, se
);
476 update_stats_wait_end(struct cfs_rq
*cfs_rq
, struct sched_entity
*se
)
478 schedstat_set(se
->wait_max
, max(se
->wait_max
,
479 rq_of(cfs_rq
)->clock
- se
->wait_start
));
480 schedstat_set(se
->wait_count
, se
->wait_count
+ 1);
481 schedstat_set(se
->wait_sum
, se
->wait_sum
+
482 rq_of(cfs_rq
)->clock
- se
->wait_start
);
483 schedstat_set(se
->wait_start
, 0);
487 update_stats_dequeue(struct cfs_rq
*cfs_rq
, struct sched_entity
*se
)
490 * Mark the end of the wait period if dequeueing a
493 if (se
!= cfs_rq
->curr
)
494 update_stats_wait_end(cfs_rq
, se
);
498 * We are picking a new current task - update its stats:
501 update_stats_curr_start(struct cfs_rq
*cfs_rq
, struct sched_entity
*se
)
504 * We are starting a new run period:
506 se
->exec_start
= rq_of(cfs_rq
)->clock
;
509 /**************************************************
510 * Scheduling class queueing methods:
514 account_entity_enqueue(struct cfs_rq
*cfs_rq
, struct sched_entity
*se
)
516 update_load_add(&cfs_rq
->load
, se
->load
.weight
);
517 cfs_rq
->nr_running
++;
519 list_add(&se
->group_node
, &cfs_rq
->tasks
);
523 account_entity_dequeue(struct cfs_rq
*cfs_rq
, struct sched_entity
*se
)
525 update_load_sub(&cfs_rq
->load
, se
->load
.weight
);
526 cfs_rq
->nr_running
--;
528 list_del_init(&se
->group_node
);
531 static void enqueue_sleeper(struct cfs_rq
*cfs_rq
, struct sched_entity
*se
)
533 #ifdef CONFIG_SCHEDSTATS
534 if (se
->sleep_start
) {
535 u64 delta
= rq_of(cfs_rq
)->clock
- se
->sleep_start
;
536 struct task_struct
*tsk
= task_of(se
);
541 if (unlikely(delta
> se
->sleep_max
))
542 se
->sleep_max
= delta
;
545 se
->sum_sleep_runtime
+= delta
;
547 account_scheduler_latency(tsk
, delta
>> 10, 1);
549 if (se
->block_start
) {
550 u64 delta
= rq_of(cfs_rq
)->clock
- se
->block_start
;
551 struct task_struct
*tsk
= task_of(se
);
556 if (unlikely(delta
> se
->block_max
))
557 se
->block_max
= delta
;
560 se
->sum_sleep_runtime
+= delta
;
563 * Blocking time is in units of nanosecs, so shift by 20 to
564 * get a milliseconds-range estimation of the amount of
565 * time that the task spent sleeping:
567 if (unlikely(prof_on
== SLEEP_PROFILING
)) {
569 profile_hits(SLEEP_PROFILING
, (void *)get_wchan(tsk
),
572 account_scheduler_latency(tsk
, delta
>> 10, 0);
577 static void check_spread(struct cfs_rq
*cfs_rq
, struct sched_entity
*se
)
579 #ifdef CONFIG_SCHED_DEBUG
580 s64 d
= se
->vruntime
- cfs_rq
->min_vruntime
;
585 if (d
> 3*sysctl_sched_latency
)
586 schedstat_inc(cfs_rq
, nr_spread_over
);
591 place_entity(struct cfs_rq
*cfs_rq
, struct sched_entity
*se
, int initial
)
595 if (first_fair(cfs_rq
)) {
596 vruntime
= min_vruntime(cfs_rq
->min_vruntime
,
597 __pick_next_entity(cfs_rq
)->vruntime
);
599 vruntime
= cfs_rq
->min_vruntime
;
602 * The 'current' period is already promised to the current tasks,
603 * however the extra weight of the new task will slow them down a
604 * little, place the new task so that it fits in the slot that
605 * stays open at the end.
607 if (initial
&& sched_feat(START_DEBIT
))
608 vruntime
+= sched_vslice_add(cfs_rq
, se
);
611 /* sleeps upto a single latency don't count. */
612 if (sched_feat(NEW_FAIR_SLEEPERS
))
613 vruntime
-= sysctl_sched_latency
;
615 /* ensure we never gain time by being placed backwards. */
616 vruntime
= max_vruntime(se
->vruntime
, vruntime
);
619 se
->vruntime
= vruntime
;
623 enqueue_entity(struct cfs_rq
*cfs_rq
, struct sched_entity
*se
, int wakeup
)
626 * Update run-time statistics of the 'current'.
629 account_entity_enqueue(cfs_rq
, se
);
632 place_entity(cfs_rq
, se
, 0);
633 enqueue_sleeper(cfs_rq
, se
);
636 update_stats_enqueue(cfs_rq
, se
);
637 check_spread(cfs_rq
, se
);
638 if (se
!= cfs_rq
->curr
)
639 __enqueue_entity(cfs_rq
, se
);
642 static void update_avg(u64
*avg
, u64 sample
)
644 s64 diff
= sample
- *avg
;
648 static void update_avg_stats(struct cfs_rq
*cfs_rq
, struct sched_entity
*se
)
650 if (!se
->last_wakeup
)
653 update_avg(&se
->avg_overlap
, se
->sum_exec_runtime
- se
->last_wakeup
);
658 dequeue_entity(struct cfs_rq
*cfs_rq
, struct sched_entity
*se
, int sleep
)
661 * Update run-time statistics of the 'current'.
665 update_stats_dequeue(cfs_rq
, se
);
667 update_avg_stats(cfs_rq
, se
);
668 #ifdef CONFIG_SCHEDSTATS
669 if (entity_is_task(se
)) {
670 struct task_struct
*tsk
= task_of(se
);
672 if (tsk
->state
& TASK_INTERRUPTIBLE
)
673 se
->sleep_start
= rq_of(cfs_rq
)->clock
;
674 if (tsk
->state
& TASK_UNINTERRUPTIBLE
)
675 se
->block_start
= rq_of(cfs_rq
)->clock
;
680 if (se
!= cfs_rq
->curr
)
681 __dequeue_entity(cfs_rq
, se
);
682 account_entity_dequeue(cfs_rq
, se
);
686 * Preempt the current task with a newly woken task if needed:
689 check_preempt_tick(struct cfs_rq
*cfs_rq
, struct sched_entity
*curr
)
691 unsigned long ideal_runtime
, delta_exec
;
693 ideal_runtime
= sched_slice(cfs_rq
, curr
);
694 delta_exec
= curr
->sum_exec_runtime
- curr
->prev_sum_exec_runtime
;
695 if (delta_exec
> ideal_runtime
)
696 resched_task(rq_of(cfs_rq
)->curr
);
700 set_next_entity(struct cfs_rq
*cfs_rq
, struct sched_entity
*se
)
702 /* 'current' is not kept within the tree. */
705 * Any task has to be enqueued before it get to execute on
706 * a CPU. So account for the time it spent waiting on the
709 update_stats_wait_end(cfs_rq
, se
);
710 __dequeue_entity(cfs_rq
, se
);
713 update_stats_curr_start(cfs_rq
, se
);
715 #ifdef CONFIG_SCHEDSTATS
717 * Track our maximum slice length, if the CPU's load is at
718 * least twice that of our own weight (i.e. dont track it
719 * when there are only lesser-weight tasks around):
721 if (rq_of(cfs_rq
)->load
.weight
>= 2*se
->load
.weight
) {
722 se
->slice_max
= max(se
->slice_max
,
723 se
->sum_exec_runtime
- se
->prev_sum_exec_runtime
);
726 se
->prev_sum_exec_runtime
= se
->sum_exec_runtime
;
730 wakeup_preempt_entity(struct sched_entity
*curr
, struct sched_entity
*se
);
732 static struct sched_entity
*
733 pick_next(struct cfs_rq
*cfs_rq
, struct sched_entity
*se
)
738 if (wakeup_preempt_entity(cfs_rq
->next
, se
) != 0)
744 static struct sched_entity
*pick_next_entity(struct cfs_rq
*cfs_rq
)
746 struct sched_entity
*se
= NULL
;
748 if (first_fair(cfs_rq
)) {
749 se
= __pick_next_entity(cfs_rq
);
750 se
= pick_next(cfs_rq
, se
);
751 set_next_entity(cfs_rq
, se
);
757 static void put_prev_entity(struct cfs_rq
*cfs_rq
, struct sched_entity
*prev
)
760 * If still on the runqueue then deactivate_task()
761 * was not called and update_curr() has to be done:
766 check_spread(cfs_rq
, prev
);
768 update_stats_wait_start(cfs_rq
, prev
);
769 /* Put 'current' back into the tree. */
770 __enqueue_entity(cfs_rq
, prev
);
776 entity_tick(struct cfs_rq
*cfs_rq
, struct sched_entity
*curr
, int queued
)
779 * Update run-time statistics of the 'current'.
783 #ifdef CONFIG_SCHED_HRTICK
785 * queued ticks are scheduled to match the slice, so don't bother
786 * validating it and just reschedule.
789 resched_task(rq_of(cfs_rq
)->curr
);
793 * don't let the period tick interfere with the hrtick preemption
795 if (!sched_feat(DOUBLE_TICK
) &&
796 hrtimer_active(&rq_of(cfs_rq
)->hrtick_timer
))
800 if (cfs_rq
->nr_running
> 1 || !sched_feat(WAKEUP_PREEMPT
))
801 check_preempt_tick(cfs_rq
, curr
);
804 /**************************************************
805 * CFS operations on tasks:
808 #ifdef CONFIG_SCHED_HRTICK
809 static void hrtick_start_fair(struct rq
*rq
, struct task_struct
*p
)
811 int requeue
= rq
->curr
== p
;
812 struct sched_entity
*se
= &p
->se
;
813 struct cfs_rq
*cfs_rq
= cfs_rq_of(se
);
815 WARN_ON(task_rq(p
) != rq
);
817 if (hrtick_enabled(rq
) && cfs_rq
->nr_running
> 1) {
818 u64 slice
= sched_slice(cfs_rq
, se
);
819 u64 ran
= se
->sum_exec_runtime
- se
->prev_sum_exec_runtime
;
820 s64 delta
= slice
- ran
;
829 * Don't schedule slices shorter than 10000ns, that just
830 * doesn't make sense. Rely on vruntime for fairness.
833 delta
= max(10000LL, delta
);
835 hrtick_start(rq
, delta
, requeue
);
840 hrtick_start_fair(struct rq
*rq
, struct task_struct
*p
)
846 * The enqueue_task method is called before nr_running is
847 * increased. Here we update the fair scheduling stats and
848 * then put the task into the rbtree:
850 static void enqueue_task_fair(struct rq
*rq
, struct task_struct
*p
, int wakeup
)
852 struct cfs_rq
*cfs_rq
;
853 struct sched_entity
*se
= &p
->se
;
855 for_each_sched_entity(se
) {
858 cfs_rq
= cfs_rq_of(se
);
859 enqueue_entity(cfs_rq
, se
, wakeup
);
863 hrtick_start_fair(rq
, rq
->curr
);
867 * The dequeue_task method is called before nr_running is
868 * decreased. We remove the task from the rbtree and
869 * update the fair scheduling stats:
871 static void dequeue_task_fair(struct rq
*rq
, struct task_struct
*p
, int sleep
)
873 struct cfs_rq
*cfs_rq
;
874 struct sched_entity
*se
= &p
->se
;
876 for_each_sched_entity(se
) {
877 cfs_rq
= cfs_rq_of(se
);
878 dequeue_entity(cfs_rq
, se
, sleep
);
879 /* Don't dequeue parent if it has other entities besides us */
880 if (cfs_rq
->load
.weight
)
885 hrtick_start_fair(rq
, rq
->curr
);
889 * sched_yield() support is very simple - we dequeue and enqueue.
891 * If compat_yield is turned on then we requeue to the end of the tree.
893 static void yield_task_fair(struct rq
*rq
)
895 struct task_struct
*curr
= rq
->curr
;
896 struct cfs_rq
*cfs_rq
= task_cfs_rq(curr
);
897 struct sched_entity
*rightmost
, *se
= &curr
->se
;
900 * Are we the only task in the tree?
902 if (unlikely(cfs_rq
->nr_running
== 1))
905 if (likely(!sysctl_sched_compat_yield
) && curr
->policy
!= SCHED_BATCH
) {
908 * Update run-time statistics of the 'current'.
915 * Find the rightmost entry in the rbtree:
917 rightmost
= __pick_last_entity(cfs_rq
);
919 * Already in the rightmost position?
921 if (unlikely(!rightmost
|| rightmost
->vruntime
< se
->vruntime
))
925 * Minimally necessary key value to be last in the tree:
926 * Upon rescheduling, sched_class::put_prev_task() will place
927 * 'current' within the tree based on its new key value.
929 se
->vruntime
= rightmost
->vruntime
+ 1;
933 * wake_idle() will wake a task on an idle cpu if task->cpu is
934 * not idle and an idle cpu is available. The span of cpus to
935 * search starts with cpus closest then further out as needed,
936 * so we always favor a closer, idle cpu.
938 * Returns the CPU we should wake onto.
940 #if defined(ARCH_HAS_SCHED_WAKE_IDLE)
941 static int wake_idle(int cpu
, struct task_struct
*p
)
944 struct sched_domain
*sd
;
948 * If it is idle, then it is the best cpu to run this task.
950 * This cpu is also the best, if it has more than one task already.
951 * Siblings must be also busy(in most cases) as they didn't already
952 * pickup the extra load from this cpu and hence we need not check
953 * sibling runqueue info. This will avoid the checks and cache miss
954 * penalities associated with that.
956 if (idle_cpu(cpu
) || cpu_rq(cpu
)->cfs
.nr_running
> 1)
959 for_each_domain(cpu
, sd
) {
960 if ((sd
->flags
& SD_WAKE_IDLE
)
961 || ((sd
->flags
& SD_WAKE_IDLE_FAR
)
962 && !task_hot(p
, task_rq(p
)->clock
, sd
))) {
963 cpus_and(tmp
, sd
->span
, p
->cpus_allowed
);
964 for_each_cpu_mask(i
, tmp
) {
966 if (i
!= task_cpu(p
)) {
980 static inline int wake_idle(int cpu
, struct task_struct
*p
)
988 static const struct sched_class fair_sched_class
;
991 wake_affine(struct rq
*rq
, struct sched_domain
*this_sd
, struct rq
*this_rq
,
992 struct task_struct
*p
, int prev_cpu
, int this_cpu
, int sync
,
993 int idx
, unsigned long load
, unsigned long this_load
,
994 unsigned int imbalance
)
996 struct task_struct
*curr
= this_rq
->curr
;
997 unsigned long tl
= this_load
;
998 unsigned long tl_per_task
;
1001 if (!(this_sd
->flags
& SD_WAKE_AFFINE
) || !sched_feat(AFFINE_WAKEUPS
))
1005 * If sync wakeup then subtract the (maximum possible)
1006 * effect of the currently running task from the load
1007 * of the current CPU:
1010 tl
-= current
->se
.load
.weight
;
1012 balanced
= 100*(tl
+ p
->se
.load
.weight
) <= imbalance
*load
;
1015 * If the currently running task will sleep within
1016 * a reasonable amount of time then attract this newly
1019 if (sync
&& balanced
&& curr
->sched_class
== &fair_sched_class
) {
1020 if (curr
->se
.avg_overlap
< sysctl_sched_migration_cost
&&
1021 p
->se
.avg_overlap
< sysctl_sched_migration_cost
)
1025 schedstat_inc(p
, se
.nr_wakeups_affine_attempts
);
1026 tl_per_task
= cpu_avg_load_per_task(this_cpu
);
1028 if ((tl
<= load
&& tl
+ target_load(prev_cpu
, idx
) <= tl_per_task
) ||
1031 * This domain has SD_WAKE_AFFINE and
1032 * p is cache cold in this domain, and
1033 * there is no bad imbalance.
1035 schedstat_inc(this_sd
, ttwu_move_affine
);
1036 schedstat_inc(p
, se
.nr_wakeups_affine
);
1043 static int select_task_rq_fair(struct task_struct
*p
, int sync
)
1045 struct sched_domain
*sd
, *this_sd
= NULL
;
1046 int prev_cpu
, this_cpu
, new_cpu
;
1047 unsigned long load
, this_load
;
1048 struct rq
*rq
, *this_rq
;
1049 unsigned int imbalance
;
1052 prev_cpu
= task_cpu(p
);
1054 this_cpu
= smp_processor_id();
1055 this_rq
= cpu_rq(this_cpu
);
1059 * 'this_sd' is the first domain that both
1060 * this_cpu and prev_cpu are present in:
1062 for_each_domain(this_cpu
, sd
) {
1063 if (cpu_isset(prev_cpu
, sd
->span
)) {
1069 if (unlikely(!cpu_isset(this_cpu
, p
->cpus_allowed
)))
1073 * Check for affine wakeup and passive balancing possibilities.
1078 idx
= this_sd
->wake_idx
;
1080 imbalance
= 100 + (this_sd
->imbalance_pct
- 100) / 2;
1082 load
= source_load(prev_cpu
, idx
);
1083 this_load
= target_load(this_cpu
, idx
);
1085 if (wake_affine(rq
, this_sd
, this_rq
, p
, prev_cpu
, this_cpu
, sync
, idx
,
1086 load
, this_load
, imbalance
))
1089 if (prev_cpu
== this_cpu
)
1093 * Start passive balancing when half the imbalance_pct
1096 if (this_sd
->flags
& SD_WAKE_BALANCE
) {
1097 if (imbalance
*this_load
<= 100*load
) {
1098 schedstat_inc(this_sd
, ttwu_move_balance
);
1099 schedstat_inc(p
, se
.nr_wakeups_passive
);
1105 return wake_idle(new_cpu
, p
);
1107 #endif /* CONFIG_SMP */
1109 static unsigned long wakeup_gran(struct sched_entity
*se
)
1111 unsigned long gran
= sysctl_sched_wakeup_granularity
;
1114 * More easily preempt - nice tasks, while not making
1115 * it harder for + nice tasks.
1117 if (unlikely(se
->load
.weight
> NICE_0_LOAD
))
1118 gran
= calc_delta_fair(gran
, &se
->load
);
1124 * Should 'se' preempt 'curr'.
1138 wakeup_preempt_entity(struct sched_entity
*curr
, struct sched_entity
*se
)
1140 s64 gran
, vdiff
= curr
->vruntime
- se
->vruntime
;
1145 gran
= wakeup_gran(curr
);
1152 /* return depth at which a sched entity is present in the hierarchy */
1153 static inline int depth_se(struct sched_entity
*se
)
1157 for_each_sched_entity(se
)
1164 * Preempt the current task with a newly woken task if needed:
1166 static void check_preempt_wakeup(struct rq
*rq
, struct task_struct
*p
)
1168 struct task_struct
*curr
= rq
->curr
;
1169 struct cfs_rq
*cfs_rq
= task_cfs_rq(curr
);
1170 struct sched_entity
*se
= &curr
->se
, *pse
= &p
->se
;
1171 int se_depth
, pse_depth
;
1173 if (unlikely(rt_prio(p
->prio
))) {
1174 update_rq_clock(rq
);
1175 update_curr(cfs_rq
);
1180 se
->last_wakeup
= se
->sum_exec_runtime
;
1181 if (unlikely(se
== pse
))
1184 cfs_rq_of(pse
)->next
= pse
;
1187 * Batch tasks do not preempt (their preemption is driven by
1190 if (unlikely(p
->policy
== SCHED_BATCH
))
1193 if (!sched_feat(WAKEUP_PREEMPT
))
1197 * preemption test can be made between sibling entities who are in the
1198 * same cfs_rq i.e who have a common parent. Walk up the hierarchy of
1199 * both tasks until we find their ancestors who are siblings of common
1203 /* First walk up until both entities are at same depth */
1204 se_depth
= depth_se(se
);
1205 pse_depth
= depth_se(pse
);
1207 while (se_depth
> pse_depth
) {
1209 se
= parent_entity(se
);
1212 while (pse_depth
> se_depth
) {
1214 pse
= parent_entity(pse
);
1217 while (!is_same_group(se
, pse
)) {
1218 se
= parent_entity(se
);
1219 pse
= parent_entity(pse
);
1222 if (wakeup_preempt_entity(se
, pse
) == 1)
1226 static struct task_struct
*pick_next_task_fair(struct rq
*rq
)
1228 struct task_struct
*p
;
1229 struct cfs_rq
*cfs_rq
= &rq
->cfs
;
1230 struct sched_entity
*se
;
1232 if (unlikely(!cfs_rq
->nr_running
))
1236 se
= pick_next_entity(cfs_rq
);
1237 cfs_rq
= group_cfs_rq(se
);
1241 hrtick_start_fair(rq
, p
);
1247 * Account for a descheduled task:
1249 static void put_prev_task_fair(struct rq
*rq
, struct task_struct
*prev
)
1251 struct sched_entity
*se
= &prev
->se
;
1252 struct cfs_rq
*cfs_rq
;
1254 for_each_sched_entity(se
) {
1255 cfs_rq
= cfs_rq_of(se
);
1256 put_prev_entity(cfs_rq
, se
);
1261 /**************************************************
1262 * Fair scheduling class load-balancing methods:
1266 * Load-balancing iterator. Note: while the runqueue stays locked
1267 * during the whole iteration, the current task might be
1268 * dequeued so the iterator has to be dequeue-safe. Here we
1269 * achieve that by always pre-iterating before returning
1272 static struct task_struct
*
1273 __load_balance_iterator(struct cfs_rq
*cfs_rq
, struct list_head
*next
)
1275 struct task_struct
*p
= NULL
;
1276 struct sched_entity
*se
;
1278 if (next
== &cfs_rq
->tasks
)
1281 /* Skip over entities that are not tasks */
1283 se
= list_entry(next
, struct sched_entity
, group_node
);
1285 } while (next
!= &cfs_rq
->tasks
&& !entity_is_task(se
));
1287 if (next
== &cfs_rq
->tasks
)
1290 cfs_rq
->balance_iterator
= next
;
1292 if (entity_is_task(se
))
1298 static struct task_struct
*load_balance_start_fair(void *arg
)
1300 struct cfs_rq
*cfs_rq
= arg
;
1302 return __load_balance_iterator(cfs_rq
, cfs_rq
->tasks
.next
);
1305 static struct task_struct
*load_balance_next_fair(void *arg
)
1307 struct cfs_rq
*cfs_rq
= arg
;
1309 return __load_balance_iterator(cfs_rq
, cfs_rq
->balance_iterator
);
1312 #ifdef CONFIG_FAIR_GROUP_SCHED
1313 static int cfs_rq_best_prio(struct cfs_rq
*cfs_rq
)
1315 struct sched_entity
*curr
;
1316 struct task_struct
*p
;
1318 if (!cfs_rq
->nr_running
|| !first_fair(cfs_rq
))
1321 curr
= cfs_rq
->curr
;
1323 curr
= __pick_next_entity(cfs_rq
);
1331 static unsigned long
1332 load_balance_fair(struct rq
*this_rq
, int this_cpu
, struct rq
*busiest
,
1333 unsigned long max_load_move
,
1334 struct sched_domain
*sd
, enum cpu_idle_type idle
,
1335 int *all_pinned
, int *this_best_prio
)
1337 struct cfs_rq
*busy_cfs_rq
;
1338 long rem_load_move
= max_load_move
;
1339 struct rq_iterator cfs_rq_iterator
;
1341 cfs_rq_iterator
.start
= load_balance_start_fair
;
1342 cfs_rq_iterator
.next
= load_balance_next_fair
;
1344 for_each_leaf_cfs_rq(busiest
, busy_cfs_rq
) {
1345 #ifdef CONFIG_FAIR_GROUP_SCHED
1346 struct cfs_rq
*this_cfs_rq
;
1348 unsigned long maxload
;
1350 this_cfs_rq
= cpu_cfs_rq(busy_cfs_rq
, this_cpu
);
1352 imbalance
= busy_cfs_rq
->load
.weight
- this_cfs_rq
->load
.weight
;
1353 /* Don't pull if this_cfs_rq has more load than busy_cfs_rq */
1357 /* Don't pull more than imbalance/2 */
1359 maxload
= min(rem_load_move
, imbalance
);
1361 *this_best_prio
= cfs_rq_best_prio(this_cfs_rq
);
1363 # define maxload rem_load_move
1366 * pass busy_cfs_rq argument into
1367 * load_balance_[start|next]_fair iterators
1369 cfs_rq_iterator
.arg
= busy_cfs_rq
;
1370 rem_load_move
-= balance_tasks(this_rq
, this_cpu
, busiest
,
1371 maxload
, sd
, idle
, all_pinned
,
1375 if (rem_load_move
<= 0)
1379 return max_load_move
- rem_load_move
;
1383 move_one_task_fair(struct rq
*this_rq
, int this_cpu
, struct rq
*busiest
,
1384 struct sched_domain
*sd
, enum cpu_idle_type idle
)
1386 struct cfs_rq
*busy_cfs_rq
;
1387 struct rq_iterator cfs_rq_iterator
;
1389 cfs_rq_iterator
.start
= load_balance_start_fair
;
1390 cfs_rq_iterator
.next
= load_balance_next_fair
;
1392 for_each_leaf_cfs_rq(busiest
, busy_cfs_rq
) {
1394 * pass busy_cfs_rq argument into
1395 * load_balance_[start|next]_fair iterators
1397 cfs_rq_iterator
.arg
= busy_cfs_rq
;
1398 if (iter_move_one_task(this_rq
, this_cpu
, busiest
, sd
, idle
,
1408 * scheduler tick hitting a task of our scheduling class:
1410 static void task_tick_fair(struct rq
*rq
, struct task_struct
*curr
, int queued
)
1412 struct cfs_rq
*cfs_rq
;
1413 struct sched_entity
*se
= &curr
->se
;
1415 for_each_sched_entity(se
) {
1416 cfs_rq
= cfs_rq_of(se
);
1417 entity_tick(cfs_rq
, se
, queued
);
1421 #define swap(a, b) do { typeof(a) tmp = (a); (a) = (b); (b) = tmp; } while (0)
1424 * Share the fairness runtime between parent and child, thus the
1425 * total amount of pressure for CPU stays equal - new tasks
1426 * get a chance to run but frequent forkers are not allowed to
1427 * monopolize the CPU. Note: the parent runqueue is locked,
1428 * the child is not running yet.
1430 static void task_new_fair(struct rq
*rq
, struct task_struct
*p
)
1432 struct cfs_rq
*cfs_rq
= task_cfs_rq(p
);
1433 struct sched_entity
*se
= &p
->se
, *curr
= cfs_rq
->curr
;
1434 int this_cpu
= smp_processor_id();
1436 sched_info_queued(p
);
1438 update_curr(cfs_rq
);
1439 place_entity(cfs_rq
, se
, 1);
1441 /* 'curr' will be NULL if the child belongs to a different group */
1442 if (sysctl_sched_child_runs_first
&& this_cpu
== task_cpu(p
) &&
1443 curr
&& curr
->vruntime
< se
->vruntime
) {
1445 * Upon rescheduling, sched_class::put_prev_task() will place
1446 * 'current' within the tree based on its new key value.
1448 swap(curr
->vruntime
, se
->vruntime
);
1451 enqueue_task_fair(rq
, p
, 0);
1452 resched_task(rq
->curr
);
1456 * Priority of the task has changed. Check to see if we preempt
1459 static void prio_changed_fair(struct rq
*rq
, struct task_struct
*p
,
1460 int oldprio
, int running
)
1463 * Reschedule if we are currently running on this runqueue and
1464 * our priority decreased, or if we are not currently running on
1465 * this runqueue and our priority is higher than the current's
1468 if (p
->prio
> oldprio
)
1469 resched_task(rq
->curr
);
1471 check_preempt_curr(rq
, p
);
1475 * We switched to the sched_fair class.
1477 static void switched_to_fair(struct rq
*rq
, struct task_struct
*p
,
1481 * We were most likely switched from sched_rt, so
1482 * kick off the schedule if running, otherwise just see
1483 * if we can still preempt the current task.
1486 resched_task(rq
->curr
);
1488 check_preempt_curr(rq
, p
);
1491 /* Account for a task changing its policy or group.
1493 * This routine is mostly called to set cfs_rq->curr field when a task
1494 * migrates between groups/classes.
1496 static void set_curr_task_fair(struct rq
*rq
)
1498 struct sched_entity
*se
= &rq
->curr
->se
;
1500 for_each_sched_entity(se
)
1501 set_next_entity(cfs_rq_of(se
), se
);
1504 #ifdef CONFIG_FAIR_GROUP_SCHED
1505 static void moved_group_fair(struct task_struct
*p
)
1507 struct cfs_rq
*cfs_rq
= task_cfs_rq(p
);
1509 update_curr(cfs_rq
);
1510 place_entity(cfs_rq
, &p
->se
, 1);
1515 * All the scheduling class methods:
1517 static const struct sched_class fair_sched_class
= {
1518 .next
= &idle_sched_class
,
1519 .enqueue_task
= enqueue_task_fair
,
1520 .dequeue_task
= dequeue_task_fair
,
1521 .yield_task
= yield_task_fair
,
1523 .select_task_rq
= select_task_rq_fair
,
1524 #endif /* CONFIG_SMP */
1526 .check_preempt_curr
= check_preempt_wakeup
,
1528 .pick_next_task
= pick_next_task_fair
,
1529 .put_prev_task
= put_prev_task_fair
,
1532 .load_balance
= load_balance_fair
,
1533 .move_one_task
= move_one_task_fair
,
1536 .set_curr_task
= set_curr_task_fair
,
1537 .task_tick
= task_tick_fair
,
1538 .task_new
= task_new_fair
,
1540 .prio_changed
= prio_changed_fair
,
1541 .switched_to
= switched_to_fair
,
1543 #ifdef CONFIG_FAIR_GROUP_SCHED
1544 .moved_group
= moved_group_fair
,
1548 #ifdef CONFIG_SCHED_DEBUG
1549 static void print_cfs_stats(struct seq_file
*m
, int cpu
)
1551 struct cfs_rq
*cfs_rq
;
1554 for_each_leaf_cfs_rq(cpu_rq(cpu
), cfs_rq
)
1555 print_cfs_rq(m
, cpu
, cfs_rq
);