sched: Provide p->on_rq
[linux-2.6.git] / kernel / sched_rt.c
1 /*
2  * Real-Time Scheduling Class (mapped to the SCHED_FIFO and SCHED_RR
3  * policies)
4  */
5
6 #ifdef CONFIG_RT_GROUP_SCHED
7
8 #define rt_entity_is_task(rt_se) (!(rt_se)->my_q)
9
10 static inline struct task_struct *rt_task_of(struct sched_rt_entity *rt_se)
11 {
12 #ifdef CONFIG_SCHED_DEBUG
13         WARN_ON_ONCE(!rt_entity_is_task(rt_se));
14 #endif
15         return container_of(rt_se, struct task_struct, rt);
16 }
17
18 static inline struct rq *rq_of_rt_rq(struct rt_rq *rt_rq)
19 {
20         return rt_rq->rq;
21 }
22
23 static inline struct rt_rq *rt_rq_of_se(struct sched_rt_entity *rt_se)
24 {
25         return rt_se->rt_rq;
26 }
27
28 #else /* CONFIG_RT_GROUP_SCHED */
29
30 #define rt_entity_is_task(rt_se) (1)
31
32 static inline struct task_struct *rt_task_of(struct sched_rt_entity *rt_se)
33 {
34         return container_of(rt_se, struct task_struct, rt);
35 }
36
37 static inline struct rq *rq_of_rt_rq(struct rt_rq *rt_rq)
38 {
39         return container_of(rt_rq, struct rq, rt);
40 }
41
42 static inline struct rt_rq *rt_rq_of_se(struct sched_rt_entity *rt_se)
43 {
44         struct task_struct *p = rt_task_of(rt_se);
45         struct rq *rq = task_rq(p);
46
47         return &rq->rt;
48 }
49
50 #endif /* CONFIG_RT_GROUP_SCHED */
51
52 #ifdef CONFIG_SMP
53
54 static inline int rt_overloaded(struct rq *rq)
55 {
56         return atomic_read(&rq->rd->rto_count);
57 }
58
59 static inline void rt_set_overload(struct rq *rq)
60 {
61         if (!rq->online)
62                 return;
63
64         cpumask_set_cpu(rq->cpu, rq->rd->rto_mask);
65         /*
66          * Make sure the mask is visible before we set
67          * the overload count. That is checked to determine
68          * if we should look at the mask. It would be a shame
69          * if we looked at the mask, but the mask was not
70          * updated yet.
71          */
72         wmb();
73         atomic_inc(&rq->rd->rto_count);
74 }
75
76 static inline void rt_clear_overload(struct rq *rq)
77 {
78         if (!rq->online)
79                 return;
80
81         /* the order here really doesn't matter */
82         atomic_dec(&rq->rd->rto_count);
83         cpumask_clear_cpu(rq->cpu, rq->rd->rto_mask);
84 }
85
86 static void update_rt_migration(struct rt_rq *rt_rq)
87 {
88         if (rt_rq->rt_nr_migratory && rt_rq->rt_nr_total > 1) {
89                 if (!rt_rq->overloaded) {
90                         rt_set_overload(rq_of_rt_rq(rt_rq));
91                         rt_rq->overloaded = 1;
92                 }
93         } else if (rt_rq->overloaded) {
94                 rt_clear_overload(rq_of_rt_rq(rt_rq));
95                 rt_rq->overloaded = 0;
96         }
97 }
98
99 static void inc_rt_migration(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq)
100 {
101         if (!rt_entity_is_task(rt_se))
102                 return;
103
104         rt_rq = &rq_of_rt_rq(rt_rq)->rt;
105
106         rt_rq->rt_nr_total++;
107         if (rt_se->nr_cpus_allowed > 1)
108                 rt_rq->rt_nr_migratory++;
109
110         update_rt_migration(rt_rq);
111 }
112
113 static void dec_rt_migration(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq)
114 {
115         if (!rt_entity_is_task(rt_se))
116                 return;
117
118         rt_rq = &rq_of_rt_rq(rt_rq)->rt;
119
120         rt_rq->rt_nr_total--;
121         if (rt_se->nr_cpus_allowed > 1)
122                 rt_rq->rt_nr_migratory--;
123
124         update_rt_migration(rt_rq);
125 }
126
127 static void enqueue_pushable_task(struct rq *rq, struct task_struct *p)
128 {
129         plist_del(&p->pushable_tasks, &rq->rt.pushable_tasks);
130         plist_node_init(&p->pushable_tasks, p->prio);
131         plist_add(&p->pushable_tasks, &rq->rt.pushable_tasks);
132 }
133
134 static void dequeue_pushable_task(struct rq *rq, struct task_struct *p)
135 {
136         plist_del(&p->pushable_tasks, &rq->rt.pushable_tasks);
137 }
138
139 static inline int has_pushable_tasks(struct rq *rq)
140 {
141         return !plist_head_empty(&rq->rt.pushable_tasks);
142 }
143
144 #else
145
146 static inline void enqueue_pushable_task(struct rq *rq, struct task_struct *p)
147 {
148 }
149
150 static inline void dequeue_pushable_task(struct rq *rq, struct task_struct *p)
151 {
152 }
153
154 static inline
155 void inc_rt_migration(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq)
156 {
157 }
158
159 static inline
160 void dec_rt_migration(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq)
161 {
162 }
163
164 #endif /* CONFIG_SMP */
165
166 static inline int on_rt_rq(struct sched_rt_entity *rt_se)
167 {
168         return !list_empty(&rt_se->run_list);
169 }
170
171 #ifdef CONFIG_RT_GROUP_SCHED
172
173 static inline u64 sched_rt_runtime(struct rt_rq *rt_rq)
174 {
175         if (!rt_rq->tg)
176                 return RUNTIME_INF;
177
178         return rt_rq->rt_runtime;
179 }
180
181 static inline u64 sched_rt_period(struct rt_rq *rt_rq)
182 {
183         return ktime_to_ns(rt_rq->tg->rt_bandwidth.rt_period);
184 }
185
186 static inline void list_add_leaf_rt_rq(struct rt_rq *rt_rq)
187 {
188         list_add_rcu(&rt_rq->leaf_rt_rq_list,
189                         &rq_of_rt_rq(rt_rq)->leaf_rt_rq_list);
190 }
191
192 static inline void list_del_leaf_rt_rq(struct rt_rq *rt_rq)
193 {
194         list_del_rcu(&rt_rq->leaf_rt_rq_list);
195 }
196
197 #define for_each_leaf_rt_rq(rt_rq, rq) \
198         list_for_each_entry_rcu(rt_rq, &rq->leaf_rt_rq_list, leaf_rt_rq_list)
199
200 #define for_each_sched_rt_entity(rt_se) \
201         for (; rt_se; rt_se = rt_se->parent)
202
203 static inline struct rt_rq *group_rt_rq(struct sched_rt_entity *rt_se)
204 {
205         return rt_se->my_q;
206 }
207
208 static void enqueue_rt_entity(struct sched_rt_entity *rt_se, bool head);
209 static void dequeue_rt_entity(struct sched_rt_entity *rt_se);
210
211 static void sched_rt_rq_enqueue(struct rt_rq *rt_rq)
212 {
213         struct task_struct *curr = rq_of_rt_rq(rt_rq)->curr;
214         struct sched_rt_entity *rt_se;
215
216         int cpu = cpu_of(rq_of_rt_rq(rt_rq));
217
218         rt_se = rt_rq->tg->rt_se[cpu];
219
220         if (rt_rq->rt_nr_running) {
221                 if (rt_se && !on_rt_rq(rt_se))
222                         enqueue_rt_entity(rt_se, false);
223                 if (rt_rq->highest_prio.curr < curr->prio)
224                         resched_task(curr);
225         }
226 }
227
228 static void sched_rt_rq_dequeue(struct rt_rq *rt_rq)
229 {
230         struct sched_rt_entity *rt_se;
231         int cpu = cpu_of(rq_of_rt_rq(rt_rq));
232
233         rt_se = rt_rq->tg->rt_se[cpu];
234
235         if (rt_se && on_rt_rq(rt_se))
236                 dequeue_rt_entity(rt_se);
237 }
238
239 static inline int rt_rq_throttled(struct rt_rq *rt_rq)
240 {
241         return rt_rq->rt_throttled && !rt_rq->rt_nr_boosted;
242 }
243
244 static int rt_se_boosted(struct sched_rt_entity *rt_se)
245 {
246         struct rt_rq *rt_rq = group_rt_rq(rt_se);
247         struct task_struct *p;
248
249         if (rt_rq)
250                 return !!rt_rq->rt_nr_boosted;
251
252         p = rt_task_of(rt_se);
253         return p->prio != p->normal_prio;
254 }
255
256 #ifdef CONFIG_SMP
257 static inline const struct cpumask *sched_rt_period_mask(void)
258 {
259         return cpu_rq(smp_processor_id())->rd->span;
260 }
261 #else
262 static inline const struct cpumask *sched_rt_period_mask(void)
263 {
264         return cpu_online_mask;
265 }
266 #endif
267
268 static inline
269 struct rt_rq *sched_rt_period_rt_rq(struct rt_bandwidth *rt_b, int cpu)
270 {
271         return container_of(rt_b, struct task_group, rt_bandwidth)->rt_rq[cpu];
272 }
273
274 static inline struct rt_bandwidth *sched_rt_bandwidth(struct rt_rq *rt_rq)
275 {
276         return &rt_rq->tg->rt_bandwidth;
277 }
278
279 #else /* !CONFIG_RT_GROUP_SCHED */
280
281 static inline u64 sched_rt_runtime(struct rt_rq *rt_rq)
282 {
283         return rt_rq->rt_runtime;
284 }
285
286 static inline u64 sched_rt_period(struct rt_rq *rt_rq)
287 {
288         return ktime_to_ns(def_rt_bandwidth.rt_period);
289 }
290
291 static inline void list_add_leaf_rt_rq(struct rt_rq *rt_rq)
292 {
293 }
294
295 static inline void list_del_leaf_rt_rq(struct rt_rq *rt_rq)
296 {
297 }
298
299 #define for_each_leaf_rt_rq(rt_rq, rq) \
300         for (rt_rq = &rq->rt; rt_rq; rt_rq = NULL)
301
302 #define for_each_sched_rt_entity(rt_se) \
303         for (; rt_se; rt_se = NULL)
304
305 static inline struct rt_rq *group_rt_rq(struct sched_rt_entity *rt_se)
306 {
307         return NULL;
308 }
309
310 static inline void sched_rt_rq_enqueue(struct rt_rq *rt_rq)
311 {
312         if (rt_rq->rt_nr_running)
313                 resched_task(rq_of_rt_rq(rt_rq)->curr);
314 }
315
316 static inline void sched_rt_rq_dequeue(struct rt_rq *rt_rq)
317 {
318 }
319
320 static inline int rt_rq_throttled(struct rt_rq *rt_rq)
321 {
322         return rt_rq->rt_throttled;
323 }
324
325 static inline const struct cpumask *sched_rt_period_mask(void)
326 {
327         return cpu_online_mask;
328 }
329
330 static inline
331 struct rt_rq *sched_rt_period_rt_rq(struct rt_bandwidth *rt_b, int cpu)
332 {
333         return &cpu_rq(cpu)->rt;
334 }
335
336 static inline struct rt_bandwidth *sched_rt_bandwidth(struct rt_rq *rt_rq)
337 {
338         return &def_rt_bandwidth;
339 }
340
341 #endif /* CONFIG_RT_GROUP_SCHED */
342
343 #ifdef CONFIG_SMP
344 /*
345  * We ran out of runtime, see if we can borrow some from our neighbours.
346  */
347 static int do_balance_runtime(struct rt_rq *rt_rq)
348 {
349         struct rt_bandwidth *rt_b = sched_rt_bandwidth(rt_rq);
350         struct root_domain *rd = cpu_rq(smp_processor_id())->rd;
351         int i, weight, more = 0;
352         u64 rt_period;
353
354         weight = cpumask_weight(rd->span);
355
356         raw_spin_lock(&rt_b->rt_runtime_lock);
357         rt_period = ktime_to_ns(rt_b->rt_period);
358         for_each_cpu(i, rd->span) {
359                 struct rt_rq *iter = sched_rt_period_rt_rq(rt_b, i);
360                 s64 diff;
361
362                 if (iter == rt_rq)
363                         continue;
364
365                 raw_spin_lock(&iter->rt_runtime_lock);
366                 /*
367                  * Either all rqs have inf runtime and there's nothing to steal
368                  * or __disable_runtime() below sets a specific rq to inf to
369                  * indicate its been disabled and disalow stealing.
370                  */
371                 if (iter->rt_runtime == RUNTIME_INF)
372                         goto next;
373
374                 /*
375                  * From runqueues with spare time, take 1/n part of their
376                  * spare time, but no more than our period.
377                  */
378                 diff = iter->rt_runtime - iter->rt_time;
379                 if (diff > 0) {
380                         diff = div_u64((u64)diff, weight);
381                         if (rt_rq->rt_runtime + diff > rt_period)
382                                 diff = rt_period - rt_rq->rt_runtime;
383                         iter->rt_runtime -= diff;
384                         rt_rq->rt_runtime += diff;
385                         more = 1;
386                         if (rt_rq->rt_runtime == rt_period) {
387                                 raw_spin_unlock(&iter->rt_runtime_lock);
388                                 break;
389                         }
390                 }
391 next:
392                 raw_spin_unlock(&iter->rt_runtime_lock);
393         }
394         raw_spin_unlock(&rt_b->rt_runtime_lock);
395
396         return more;
397 }
398
399 /*
400  * Ensure this RQ takes back all the runtime it lend to its neighbours.
401  */
402 static void __disable_runtime(struct rq *rq)
403 {
404         struct root_domain *rd = rq->rd;
405         struct rt_rq *rt_rq;
406
407         if (unlikely(!scheduler_running))
408                 return;
409
410         for_each_leaf_rt_rq(rt_rq, rq) {
411                 struct rt_bandwidth *rt_b = sched_rt_bandwidth(rt_rq);
412                 s64 want;
413                 int i;
414
415                 raw_spin_lock(&rt_b->rt_runtime_lock);
416                 raw_spin_lock(&rt_rq->rt_runtime_lock);
417                 /*
418                  * Either we're all inf and nobody needs to borrow, or we're
419                  * already disabled and thus have nothing to do, or we have
420                  * exactly the right amount of runtime to take out.
421                  */
422                 if (rt_rq->rt_runtime == RUNTIME_INF ||
423                                 rt_rq->rt_runtime == rt_b->rt_runtime)
424                         goto balanced;
425                 raw_spin_unlock(&rt_rq->rt_runtime_lock);
426
427                 /*
428                  * Calculate the difference between what we started out with
429                  * and what we current have, that's the amount of runtime
430                  * we lend and now have to reclaim.
431                  */
432                 want = rt_b->rt_runtime - rt_rq->rt_runtime;
433
434                 /*
435                  * Greedy reclaim, take back as much as we can.
436                  */
437                 for_each_cpu(i, rd->span) {
438                         struct rt_rq *iter = sched_rt_period_rt_rq(rt_b, i);
439                         s64 diff;
440
441                         /*
442                          * Can't reclaim from ourselves or disabled runqueues.
443                          */
444                         if (iter == rt_rq || iter->rt_runtime == RUNTIME_INF)
445                                 continue;
446
447                         raw_spin_lock(&iter->rt_runtime_lock);
448                         if (want > 0) {
449                                 diff = min_t(s64, iter->rt_runtime, want);
450                                 iter->rt_runtime -= diff;
451                                 want -= diff;
452                         } else {
453                                 iter->rt_runtime -= want;
454                                 want -= want;
455                         }
456                         raw_spin_unlock(&iter->rt_runtime_lock);
457
458                         if (!want)
459                                 break;
460                 }
461
462                 raw_spin_lock(&rt_rq->rt_runtime_lock);
463                 /*
464                  * We cannot be left wanting - that would mean some runtime
465                  * leaked out of the system.
466                  */
467                 BUG_ON(want);
468 balanced:
469                 /*
470                  * Disable all the borrow logic by pretending we have inf
471                  * runtime - in which case borrowing doesn't make sense.
472                  */
473                 rt_rq->rt_runtime = RUNTIME_INF;
474                 raw_spin_unlock(&rt_rq->rt_runtime_lock);
475                 raw_spin_unlock(&rt_b->rt_runtime_lock);
476         }
477 }
478
479 static void disable_runtime(struct rq *rq)
480 {
481         unsigned long flags;
482
483         raw_spin_lock_irqsave(&rq->lock, flags);
484         __disable_runtime(rq);
485         raw_spin_unlock_irqrestore(&rq->lock, flags);
486 }
487
488 static void __enable_runtime(struct rq *rq)
489 {
490         struct rt_rq *rt_rq;
491
492         if (unlikely(!scheduler_running))
493                 return;
494
495         /*
496          * Reset each runqueue's bandwidth settings
497          */
498         for_each_leaf_rt_rq(rt_rq, rq) {
499                 struct rt_bandwidth *rt_b = sched_rt_bandwidth(rt_rq);
500
501                 raw_spin_lock(&rt_b->rt_runtime_lock);
502                 raw_spin_lock(&rt_rq->rt_runtime_lock);
503                 rt_rq->rt_runtime = rt_b->rt_runtime;
504                 rt_rq->rt_time = 0;
505                 rt_rq->rt_throttled = 0;
506                 raw_spin_unlock(&rt_rq->rt_runtime_lock);
507                 raw_spin_unlock(&rt_b->rt_runtime_lock);
508         }
509 }
510
511 static void enable_runtime(struct rq *rq)
512 {
513         unsigned long flags;
514
515         raw_spin_lock_irqsave(&rq->lock, flags);
516         __enable_runtime(rq);
517         raw_spin_unlock_irqrestore(&rq->lock, flags);
518 }
519
520 static int balance_runtime(struct rt_rq *rt_rq)
521 {
522         int more = 0;
523
524         if (rt_rq->rt_time > rt_rq->rt_runtime) {
525                 raw_spin_unlock(&rt_rq->rt_runtime_lock);
526                 more = do_balance_runtime(rt_rq);
527                 raw_spin_lock(&rt_rq->rt_runtime_lock);
528         }
529
530         return more;
531 }
532 #else /* !CONFIG_SMP */
533 static inline int balance_runtime(struct rt_rq *rt_rq)
534 {
535         return 0;
536 }
537 #endif /* CONFIG_SMP */
538
539 static int do_sched_rt_period_timer(struct rt_bandwidth *rt_b, int overrun)
540 {
541         int i, idle = 1;
542         const struct cpumask *span;
543
544         if (!rt_bandwidth_enabled() || rt_b->rt_runtime == RUNTIME_INF)
545                 return 1;
546
547         span = sched_rt_period_mask();
548         for_each_cpu(i, span) {
549                 int enqueue = 0;
550                 struct rt_rq *rt_rq = sched_rt_period_rt_rq(rt_b, i);
551                 struct rq *rq = rq_of_rt_rq(rt_rq);
552
553                 raw_spin_lock(&rq->lock);
554                 if (rt_rq->rt_time) {
555                         u64 runtime;
556
557                         raw_spin_lock(&rt_rq->rt_runtime_lock);
558                         if (rt_rq->rt_throttled)
559                                 balance_runtime(rt_rq);
560                         runtime = rt_rq->rt_runtime;
561                         rt_rq->rt_time -= min(rt_rq->rt_time, overrun*runtime);
562                         if (rt_rq->rt_throttled && rt_rq->rt_time < runtime) {
563                                 rt_rq->rt_throttled = 0;
564                                 enqueue = 1;
565                         }
566                         if (rt_rq->rt_time || rt_rq->rt_nr_running)
567                                 idle = 0;
568                         raw_spin_unlock(&rt_rq->rt_runtime_lock);
569                 } else if (rt_rq->rt_nr_running) {
570                         idle = 0;
571                         if (!rt_rq_throttled(rt_rq))
572                                 enqueue = 1;
573                 }
574
575                 if (enqueue)
576                         sched_rt_rq_enqueue(rt_rq);
577                 raw_spin_unlock(&rq->lock);
578         }
579
580         return idle;
581 }
582
583 static inline int rt_se_prio(struct sched_rt_entity *rt_se)
584 {
585 #ifdef CONFIG_RT_GROUP_SCHED
586         struct rt_rq *rt_rq = group_rt_rq(rt_se);
587
588         if (rt_rq)
589                 return rt_rq->highest_prio.curr;
590 #endif
591
592         return rt_task_of(rt_se)->prio;
593 }
594
595 static int sched_rt_runtime_exceeded(struct rt_rq *rt_rq)
596 {
597         u64 runtime = sched_rt_runtime(rt_rq);
598
599         if (rt_rq->rt_throttled)
600                 return rt_rq_throttled(rt_rq);
601
602         if (sched_rt_runtime(rt_rq) >= sched_rt_period(rt_rq))
603                 return 0;
604
605         balance_runtime(rt_rq);
606         runtime = sched_rt_runtime(rt_rq);
607         if (runtime == RUNTIME_INF)
608                 return 0;
609
610         if (rt_rq->rt_time > runtime) {
611                 rt_rq->rt_throttled = 1;
612                 if (rt_rq_throttled(rt_rq)) {
613                         sched_rt_rq_dequeue(rt_rq);
614                         return 1;
615                 }
616         }
617
618         return 0;
619 }
620
621 /*
622  * Update the current task's runtime statistics. Skip current tasks that
623  * are not in our scheduling class.
624  */
625 static void update_curr_rt(struct rq *rq)
626 {
627         struct task_struct *curr = rq->curr;
628         struct sched_rt_entity *rt_se = &curr->rt;
629         struct rt_rq *rt_rq = rt_rq_of_se(rt_se);
630         u64 delta_exec;
631
632         if (curr->sched_class != &rt_sched_class)
633                 return;
634
635         delta_exec = rq->clock_task - curr->se.exec_start;
636         if (unlikely((s64)delta_exec < 0))
637                 delta_exec = 0;
638
639         schedstat_set(curr->se.statistics.exec_max, max(curr->se.statistics.exec_max, delta_exec));
640
641         curr->se.sum_exec_runtime += delta_exec;
642         account_group_exec_runtime(curr, delta_exec);
643
644         curr->se.exec_start = rq->clock_task;
645         cpuacct_charge(curr, delta_exec);
646
647         sched_rt_avg_update(rq, delta_exec);
648
649         if (!rt_bandwidth_enabled())
650                 return;
651
652         for_each_sched_rt_entity(rt_se) {
653                 rt_rq = rt_rq_of_se(rt_se);
654
655                 if (sched_rt_runtime(rt_rq) != RUNTIME_INF) {
656                         raw_spin_lock(&rt_rq->rt_runtime_lock);
657                         rt_rq->rt_time += delta_exec;
658                         if (sched_rt_runtime_exceeded(rt_rq))
659                                 resched_task(curr);
660                         raw_spin_unlock(&rt_rq->rt_runtime_lock);
661                 }
662         }
663 }
664
665 #if defined CONFIG_SMP
666
667 static struct task_struct *pick_next_highest_task_rt(struct rq *rq, int cpu);
668
669 static inline int next_prio(struct rq *rq)
670 {
671         struct task_struct *next = pick_next_highest_task_rt(rq, rq->cpu);
672
673         if (next && rt_prio(next->prio))
674                 return next->prio;
675         else
676                 return MAX_RT_PRIO;
677 }
678
679 static void
680 inc_rt_prio_smp(struct rt_rq *rt_rq, int prio, int prev_prio)
681 {
682         struct rq *rq = rq_of_rt_rq(rt_rq);
683
684         if (prio < prev_prio) {
685
686                 /*
687                  * If the new task is higher in priority than anything on the
688                  * run-queue, we know that the previous high becomes our
689                  * next-highest.
690                  */
691                 rt_rq->highest_prio.next = prev_prio;
692
693                 if (rq->online)
694                         cpupri_set(&rq->rd->cpupri, rq->cpu, prio);
695
696         } else if (prio == rt_rq->highest_prio.curr)
697                 /*
698                  * If the next task is equal in priority to the highest on
699                  * the run-queue, then we implicitly know that the next highest
700                  * task cannot be any lower than current
701                  */
702                 rt_rq->highest_prio.next = prio;
703         else if (prio < rt_rq->highest_prio.next)
704                 /*
705                  * Otherwise, we need to recompute next-highest
706                  */
707                 rt_rq->highest_prio.next = next_prio(rq);
708 }
709
710 static void
711 dec_rt_prio_smp(struct rt_rq *rt_rq, int prio, int prev_prio)
712 {
713         struct rq *rq = rq_of_rt_rq(rt_rq);
714
715         if (rt_rq->rt_nr_running && (prio <= rt_rq->highest_prio.next))
716                 rt_rq->highest_prio.next = next_prio(rq);
717
718         if (rq->online && rt_rq->highest_prio.curr != prev_prio)
719                 cpupri_set(&rq->rd->cpupri, rq->cpu, rt_rq->highest_prio.curr);
720 }
721
722 #else /* CONFIG_SMP */
723
724 static inline
725 void inc_rt_prio_smp(struct rt_rq *rt_rq, int prio, int prev_prio) {}
726 static inline
727 void dec_rt_prio_smp(struct rt_rq *rt_rq, int prio, int prev_prio) {}
728
729 #endif /* CONFIG_SMP */
730
731 #if defined CONFIG_SMP || defined CONFIG_RT_GROUP_SCHED
732 static void
733 inc_rt_prio(struct rt_rq *rt_rq, int prio)
734 {
735         int prev_prio = rt_rq->highest_prio.curr;
736
737         if (prio < prev_prio)
738                 rt_rq->highest_prio.curr = prio;
739
740         inc_rt_prio_smp(rt_rq, prio, prev_prio);
741 }
742
743 static void
744 dec_rt_prio(struct rt_rq *rt_rq, int prio)
745 {
746         int prev_prio = rt_rq->highest_prio.curr;
747
748         if (rt_rq->rt_nr_running) {
749
750                 WARN_ON(prio < prev_prio);
751
752                 /*
753                  * This may have been our highest task, and therefore
754                  * we may have some recomputation to do
755                  */
756                 if (prio == prev_prio) {
757                         struct rt_prio_array *array = &rt_rq->active;
758
759                         rt_rq->highest_prio.curr =
760                                 sched_find_first_bit(array->bitmap);
761                 }
762
763         } else
764                 rt_rq->highest_prio.curr = MAX_RT_PRIO;
765
766         dec_rt_prio_smp(rt_rq, prio, prev_prio);
767 }
768
769 #else
770
771 static inline void inc_rt_prio(struct rt_rq *rt_rq, int prio) {}
772 static inline void dec_rt_prio(struct rt_rq *rt_rq, int prio) {}
773
774 #endif /* CONFIG_SMP || CONFIG_RT_GROUP_SCHED */
775
776 #ifdef CONFIG_RT_GROUP_SCHED
777
778 static void
779 inc_rt_group(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq)
780 {
781         if (rt_se_boosted(rt_se))
782                 rt_rq->rt_nr_boosted++;
783
784         if (rt_rq->tg)
785                 start_rt_bandwidth(&rt_rq->tg->rt_bandwidth);
786 }
787
788 static void
789 dec_rt_group(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq)
790 {
791         if (rt_se_boosted(rt_se))
792                 rt_rq->rt_nr_boosted--;
793
794         WARN_ON(!rt_rq->rt_nr_running && rt_rq->rt_nr_boosted);
795 }
796
797 #else /* CONFIG_RT_GROUP_SCHED */
798
799 static void
800 inc_rt_group(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq)
801 {
802         start_rt_bandwidth(&def_rt_bandwidth);
803 }
804
805 static inline
806 void dec_rt_group(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq) {}
807
808 #endif /* CONFIG_RT_GROUP_SCHED */
809
810 static inline
811 void inc_rt_tasks(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq)
812 {
813         int prio = rt_se_prio(rt_se);
814
815         WARN_ON(!rt_prio(prio));
816         rt_rq->rt_nr_running++;
817
818         inc_rt_prio(rt_rq, prio);
819         inc_rt_migration(rt_se, rt_rq);
820         inc_rt_group(rt_se, rt_rq);
821 }
822
823 static inline
824 void dec_rt_tasks(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq)
825 {
826         WARN_ON(!rt_prio(rt_se_prio(rt_se)));
827         WARN_ON(!rt_rq->rt_nr_running);
828         rt_rq->rt_nr_running--;
829
830         dec_rt_prio(rt_rq, rt_se_prio(rt_se));
831         dec_rt_migration(rt_se, rt_rq);
832         dec_rt_group(rt_se, rt_rq);
833 }
834
835 static void __enqueue_rt_entity(struct sched_rt_entity *rt_se, bool head)
836 {
837         struct rt_rq *rt_rq = rt_rq_of_se(rt_se);
838         struct rt_prio_array *array = &rt_rq->active;
839         struct rt_rq *group_rq = group_rt_rq(rt_se);
840         struct list_head *queue = array->queue + rt_se_prio(rt_se);
841
842         /*
843          * Don't enqueue the group if its throttled, or when empty.
844          * The latter is a consequence of the former when a child group
845          * get throttled and the current group doesn't have any other
846          * active members.
847          */
848         if (group_rq && (rt_rq_throttled(group_rq) || !group_rq->rt_nr_running))
849                 return;
850
851         if (!rt_rq->rt_nr_running)
852                 list_add_leaf_rt_rq(rt_rq);
853
854         if (head)
855                 list_add(&rt_se->run_list, queue);
856         else
857                 list_add_tail(&rt_se->run_list, queue);
858         __set_bit(rt_se_prio(rt_se), array->bitmap);
859
860         inc_rt_tasks(rt_se, rt_rq);
861 }
862
863 static void __dequeue_rt_entity(struct sched_rt_entity *rt_se)
864 {
865         struct rt_rq *rt_rq = rt_rq_of_se(rt_se);
866         struct rt_prio_array *array = &rt_rq->active;
867
868         list_del_init(&rt_se->run_list);
869         if (list_empty(array->queue + rt_se_prio(rt_se)))
870                 __clear_bit(rt_se_prio(rt_se), array->bitmap);
871
872         dec_rt_tasks(rt_se, rt_rq);
873         if (!rt_rq->rt_nr_running)
874                 list_del_leaf_rt_rq(rt_rq);
875 }
876
877 /*
878  * Because the prio of an upper entry depends on the lower
879  * entries, we must remove entries top - down.
880  */
881 static void dequeue_rt_stack(struct sched_rt_entity *rt_se)
882 {
883         struct sched_rt_entity *back = NULL;
884
885         for_each_sched_rt_entity(rt_se) {
886                 rt_se->back = back;
887                 back = rt_se;
888         }
889
890         for (rt_se = back; rt_se; rt_se = rt_se->back) {
891                 if (on_rt_rq(rt_se))
892                         __dequeue_rt_entity(rt_se);
893         }
894 }
895
896 static void enqueue_rt_entity(struct sched_rt_entity *rt_se, bool head)
897 {
898         dequeue_rt_stack(rt_se);
899         for_each_sched_rt_entity(rt_se)
900                 __enqueue_rt_entity(rt_se, head);
901 }
902
903 static void dequeue_rt_entity(struct sched_rt_entity *rt_se)
904 {
905         dequeue_rt_stack(rt_se);
906
907         for_each_sched_rt_entity(rt_se) {
908                 struct rt_rq *rt_rq = group_rt_rq(rt_se);
909
910                 if (rt_rq && rt_rq->rt_nr_running)
911                         __enqueue_rt_entity(rt_se, false);
912         }
913 }
914
915 /*
916  * Adding/removing a task to/from a priority array:
917  */
918 static void
919 enqueue_task_rt(struct rq *rq, struct task_struct *p, int flags)
920 {
921         struct sched_rt_entity *rt_se = &p->rt;
922
923         if (flags & ENQUEUE_WAKEUP)
924                 rt_se->timeout = 0;
925
926         enqueue_rt_entity(rt_se, flags & ENQUEUE_HEAD);
927
928         if (!task_current(rq, p) && p->rt.nr_cpus_allowed > 1)
929                 enqueue_pushable_task(rq, p);
930 }
931
932 static void dequeue_task_rt(struct rq *rq, struct task_struct *p, int flags)
933 {
934         struct sched_rt_entity *rt_se = &p->rt;
935
936         update_curr_rt(rq);
937         dequeue_rt_entity(rt_se);
938
939         dequeue_pushable_task(rq, p);
940 }
941
942 /*
943  * Put task to the end of the run list without the overhead of dequeue
944  * followed by enqueue.
945  */
946 static void
947 requeue_rt_entity(struct rt_rq *rt_rq, struct sched_rt_entity *rt_se, int head)
948 {
949         if (on_rt_rq(rt_se)) {
950                 struct rt_prio_array *array = &rt_rq->active;
951                 struct list_head *queue = array->queue + rt_se_prio(rt_se);
952
953                 if (head)
954                         list_move(&rt_se->run_list, queue);
955                 else
956                         list_move_tail(&rt_se->run_list, queue);
957         }
958 }
959
960 static void requeue_task_rt(struct rq *rq, struct task_struct *p, int head)
961 {
962         struct sched_rt_entity *rt_se = &p->rt;
963         struct rt_rq *rt_rq;
964
965         for_each_sched_rt_entity(rt_se) {
966                 rt_rq = rt_rq_of_se(rt_se);
967                 requeue_rt_entity(rt_rq, rt_se, head);
968         }
969 }
970
971 static void yield_task_rt(struct rq *rq)
972 {
973         requeue_task_rt(rq, rq->curr, 0);
974 }
975
976 #ifdef CONFIG_SMP
977 static int find_lowest_rq(struct task_struct *task);
978
979 static int
980 select_task_rq_rt(struct rq *rq, struct task_struct *p, int sd_flag, int flags)
981 {
982         if (sd_flag != SD_BALANCE_WAKE)
983                 return smp_processor_id();
984
985         /*
986          * If the current task is an RT task, then
987          * try to see if we can wake this RT task up on another
988          * runqueue. Otherwise simply start this RT task
989          * on its current runqueue.
990          *
991          * We want to avoid overloading runqueues. If the woken
992          * task is a higher priority, then it will stay on this CPU
993          * and the lower prio task should be moved to another CPU.
994          * Even though this will probably make the lower prio task
995          * lose its cache, we do not want to bounce a higher task
996          * around just because it gave up its CPU, perhaps for a
997          * lock?
998          *
999          * For equal prio tasks, we just let the scheduler sort it out.
1000          */
1001         if (unlikely(rt_task(rq->curr)) &&
1002             (rq->curr->rt.nr_cpus_allowed < 2 ||
1003              rq->curr->prio < p->prio) &&
1004             (p->rt.nr_cpus_allowed > 1)) {
1005                 int cpu = find_lowest_rq(p);
1006
1007                 return (cpu == -1) ? task_cpu(p) : cpu;
1008         }
1009
1010         /*
1011          * Otherwise, just let it ride on the affined RQ and the
1012          * post-schedule router will push the preempted task away
1013          */
1014         return task_cpu(p);
1015 }
1016
1017 static void check_preempt_equal_prio(struct rq *rq, struct task_struct *p)
1018 {
1019         if (rq->curr->rt.nr_cpus_allowed == 1)
1020                 return;
1021
1022         if (p->rt.nr_cpus_allowed != 1
1023             && cpupri_find(&rq->rd->cpupri, p, NULL))
1024                 return;
1025
1026         if (!cpupri_find(&rq->rd->cpupri, rq->curr, NULL))
1027                 return;
1028
1029         /*
1030          * There appears to be other cpus that can accept
1031          * current and none to run 'p', so lets reschedule
1032          * to try and push current away:
1033          */
1034         requeue_task_rt(rq, p, 1);
1035         resched_task(rq->curr);
1036 }
1037
1038 #endif /* CONFIG_SMP */
1039
1040 /*
1041  * Preempt the current task with a newly woken task if needed:
1042  */
1043 static void check_preempt_curr_rt(struct rq *rq, struct task_struct *p, int flags)
1044 {
1045         if (p->prio < rq->curr->prio) {
1046                 resched_task(rq->curr);
1047                 return;
1048         }
1049
1050 #ifdef CONFIG_SMP
1051         /*
1052          * If:
1053          *
1054          * - the newly woken task is of equal priority to the current task
1055          * - the newly woken task is non-migratable while current is migratable
1056          * - current will be preempted on the next reschedule
1057          *
1058          * we should check to see if current can readily move to a different
1059          * cpu.  If so, we will reschedule to allow the push logic to try
1060          * to move current somewhere else, making room for our non-migratable
1061          * task.
1062          */
1063         if (p->prio == rq->curr->prio && !need_resched())
1064                 check_preempt_equal_prio(rq, p);
1065 #endif
1066 }
1067
1068 static struct sched_rt_entity *pick_next_rt_entity(struct rq *rq,
1069                                                    struct rt_rq *rt_rq)
1070 {
1071         struct rt_prio_array *array = &rt_rq->active;
1072         struct sched_rt_entity *next = NULL;
1073         struct list_head *queue;
1074         int idx;
1075
1076         idx = sched_find_first_bit(array->bitmap);
1077         BUG_ON(idx >= MAX_RT_PRIO);
1078
1079         queue = array->queue + idx;
1080         next = list_entry(queue->next, struct sched_rt_entity, run_list);
1081
1082         return next;
1083 }
1084
1085 static struct task_struct *_pick_next_task_rt(struct rq *rq)
1086 {
1087         struct sched_rt_entity *rt_se;
1088         struct task_struct *p;
1089         struct rt_rq *rt_rq;
1090
1091         rt_rq = &rq->rt;
1092
1093         if (unlikely(!rt_rq->rt_nr_running))
1094                 return NULL;
1095
1096         if (rt_rq_throttled(rt_rq))
1097                 return NULL;
1098
1099         do {
1100                 rt_se = pick_next_rt_entity(rq, rt_rq);
1101                 BUG_ON(!rt_se);
1102                 rt_rq = group_rt_rq(rt_se);
1103         } while (rt_rq);
1104
1105         p = rt_task_of(rt_se);
1106         p->se.exec_start = rq->clock_task;
1107
1108         return p;
1109 }
1110
1111 static struct task_struct *pick_next_task_rt(struct rq *rq)
1112 {
1113         struct task_struct *p = _pick_next_task_rt(rq);
1114
1115         /* The running task is never eligible for pushing */
1116         if (p)
1117                 dequeue_pushable_task(rq, p);
1118
1119 #ifdef CONFIG_SMP
1120         /*
1121          * We detect this state here so that we can avoid taking the RQ
1122          * lock again later if there is no need to push
1123          */
1124         rq->post_schedule = has_pushable_tasks(rq);
1125 #endif
1126
1127         return p;
1128 }
1129
1130 static void put_prev_task_rt(struct rq *rq, struct task_struct *p)
1131 {
1132         update_curr_rt(rq);
1133         p->se.exec_start = 0;
1134
1135         /*
1136          * The previous task needs to be made eligible for pushing
1137          * if it is still active
1138          */
1139         if (on_rt_rq(&p->rt) && p->rt.nr_cpus_allowed > 1)
1140                 enqueue_pushable_task(rq, p);
1141 }
1142
1143 #ifdef CONFIG_SMP
1144
1145 /* Only try algorithms three times */
1146 #define RT_MAX_TRIES 3
1147
1148 static void deactivate_task(struct rq *rq, struct task_struct *p, int sleep);
1149
1150 static int pick_rt_task(struct rq *rq, struct task_struct *p, int cpu)
1151 {
1152         if (!task_running(rq, p) &&
1153             (cpu < 0 || cpumask_test_cpu(cpu, &p->cpus_allowed)) &&
1154             (p->rt.nr_cpus_allowed > 1))
1155                 return 1;
1156         return 0;
1157 }
1158
1159 /* Return the second highest RT task, NULL otherwise */
1160 static struct task_struct *pick_next_highest_task_rt(struct rq *rq, int cpu)
1161 {
1162         struct task_struct *next = NULL;
1163         struct sched_rt_entity *rt_se;
1164         struct rt_prio_array *array;
1165         struct rt_rq *rt_rq;
1166         int idx;
1167
1168         for_each_leaf_rt_rq(rt_rq, rq) {
1169                 array = &rt_rq->active;
1170                 idx = sched_find_first_bit(array->bitmap);
1171 next_idx:
1172                 if (idx >= MAX_RT_PRIO)
1173                         continue;
1174                 if (next && next->prio < idx)
1175                         continue;
1176                 list_for_each_entry(rt_se, array->queue + idx, run_list) {
1177                         struct task_struct *p;
1178
1179                         if (!rt_entity_is_task(rt_se))
1180                                 continue;
1181
1182                         p = rt_task_of(rt_se);
1183                         if (pick_rt_task(rq, p, cpu)) {
1184                                 next = p;
1185                                 break;
1186                         }
1187                 }
1188                 if (!next) {
1189                         idx = find_next_bit(array->bitmap, MAX_RT_PRIO, idx+1);
1190                         goto next_idx;
1191                 }
1192         }
1193
1194         return next;
1195 }
1196
1197 static DEFINE_PER_CPU(cpumask_var_t, local_cpu_mask);
1198
1199 static int find_lowest_rq(struct task_struct *task)
1200 {
1201         struct sched_domain *sd;
1202         struct cpumask *lowest_mask = __get_cpu_var(local_cpu_mask);
1203         int this_cpu = smp_processor_id();
1204         int cpu      = task_cpu(task);
1205
1206         if (task->rt.nr_cpus_allowed == 1)
1207                 return -1; /* No other targets possible */
1208
1209         if (!cpupri_find(&task_rq(task)->rd->cpupri, task, lowest_mask))
1210                 return -1; /* No targets found */
1211
1212         /*
1213          * At this point we have built a mask of cpus representing the
1214          * lowest priority tasks in the system.  Now we want to elect
1215          * the best one based on our affinity and topology.
1216          *
1217          * We prioritize the last cpu that the task executed on since
1218          * it is most likely cache-hot in that location.
1219          */
1220         if (cpumask_test_cpu(cpu, lowest_mask))
1221                 return cpu;
1222
1223         /*
1224          * Otherwise, we consult the sched_domains span maps to figure
1225          * out which cpu is logically closest to our hot cache data.
1226          */
1227         if (!cpumask_test_cpu(this_cpu, lowest_mask))
1228                 this_cpu = -1; /* Skip this_cpu opt if not among lowest */
1229
1230         for_each_domain(cpu, sd) {
1231                 if (sd->flags & SD_WAKE_AFFINE) {
1232                         int best_cpu;
1233
1234                         /*
1235                          * "this_cpu" is cheaper to preempt than a
1236                          * remote processor.
1237                          */
1238                         if (this_cpu != -1 &&
1239                             cpumask_test_cpu(this_cpu, sched_domain_span(sd)))
1240                                 return this_cpu;
1241
1242                         best_cpu = cpumask_first_and(lowest_mask,
1243                                                      sched_domain_span(sd));
1244                         if (best_cpu < nr_cpu_ids)
1245                                 return best_cpu;
1246                 }
1247         }
1248
1249         /*
1250          * And finally, if there were no matches within the domains
1251          * just give the caller *something* to work with from the compatible
1252          * locations.
1253          */
1254         if (this_cpu != -1)
1255                 return this_cpu;
1256
1257         cpu = cpumask_any(lowest_mask);
1258         if (cpu < nr_cpu_ids)
1259                 return cpu;
1260         return -1;
1261 }
1262
1263 /* Will lock the rq it finds */
1264 static struct rq *find_lock_lowest_rq(struct task_struct *task, struct rq *rq)
1265 {
1266         struct rq *lowest_rq = NULL;
1267         int tries;
1268         int cpu;
1269
1270         for (tries = 0; tries < RT_MAX_TRIES; tries++) {
1271                 cpu = find_lowest_rq(task);
1272
1273                 if ((cpu == -1) || (cpu == rq->cpu))
1274                         break;
1275
1276                 lowest_rq = cpu_rq(cpu);
1277
1278                 /* if the prio of this runqueue changed, try again */
1279                 if (double_lock_balance(rq, lowest_rq)) {
1280                         /*
1281                          * We had to unlock the run queue. In
1282                          * the mean time, task could have
1283                          * migrated already or had its affinity changed.
1284                          * Also make sure that it wasn't scheduled on its rq.
1285                          */
1286                         if (unlikely(task_rq(task) != rq ||
1287                                      !cpumask_test_cpu(lowest_rq->cpu,
1288                                                        &task->cpus_allowed) ||
1289                                      task_running(rq, task) ||
1290                                      !task->on_rq)) {
1291
1292                                 raw_spin_unlock(&lowest_rq->lock);
1293                                 lowest_rq = NULL;
1294                                 break;
1295                         }
1296                 }
1297
1298                 /* If this rq is still suitable use it. */
1299                 if (lowest_rq->rt.highest_prio.curr > task->prio)
1300                         break;
1301
1302                 /* try again */
1303                 double_unlock_balance(rq, lowest_rq);
1304                 lowest_rq = NULL;
1305         }
1306
1307         return lowest_rq;
1308 }
1309
1310 static struct task_struct *pick_next_pushable_task(struct rq *rq)
1311 {
1312         struct task_struct *p;
1313
1314         if (!has_pushable_tasks(rq))
1315                 return NULL;
1316
1317         p = plist_first_entry(&rq->rt.pushable_tasks,
1318                               struct task_struct, pushable_tasks);
1319
1320         BUG_ON(rq->cpu != task_cpu(p));
1321         BUG_ON(task_current(rq, p));
1322         BUG_ON(p->rt.nr_cpus_allowed <= 1);
1323
1324         BUG_ON(!p->on_rq);
1325         BUG_ON(!rt_task(p));
1326
1327         return p;
1328 }
1329
1330 /*
1331  * If the current CPU has more than one RT task, see if the non
1332  * running task can migrate over to a CPU that is running a task
1333  * of lesser priority.
1334  */
1335 static int push_rt_task(struct rq *rq)
1336 {
1337         struct task_struct *next_task;
1338         struct rq *lowest_rq;
1339
1340         if (!rq->rt.overloaded)
1341                 return 0;
1342
1343         next_task = pick_next_pushable_task(rq);
1344         if (!next_task)
1345                 return 0;
1346
1347 retry:
1348         if (unlikely(next_task == rq->curr)) {
1349                 WARN_ON(1);
1350                 return 0;
1351         }
1352
1353         /*
1354          * It's possible that the next_task slipped in of
1355          * higher priority than current. If that's the case
1356          * just reschedule current.
1357          */
1358         if (unlikely(next_task->prio < rq->curr->prio)) {
1359                 resched_task(rq->curr);
1360                 return 0;
1361         }
1362
1363         /* We might release rq lock */
1364         get_task_struct(next_task);
1365
1366         /* find_lock_lowest_rq locks the rq if found */
1367         lowest_rq = find_lock_lowest_rq(next_task, rq);
1368         if (!lowest_rq) {
1369                 struct task_struct *task;
1370                 /*
1371                  * find lock_lowest_rq releases rq->lock
1372                  * so it is possible that next_task has migrated.
1373                  *
1374                  * We need to make sure that the task is still on the same
1375                  * run-queue and is also still the next task eligible for
1376                  * pushing.
1377                  */
1378                 task = pick_next_pushable_task(rq);
1379                 if (task_cpu(next_task) == rq->cpu && task == next_task) {
1380                         /*
1381                          * If we get here, the task hasn't moved at all, but
1382                          * it has failed to push.  We will not try again,
1383                          * since the other cpus will pull from us when they
1384                          * are ready.
1385                          */
1386                         dequeue_pushable_task(rq, next_task);
1387                         goto out;
1388                 }
1389
1390                 if (!task)
1391                         /* No more tasks, just exit */
1392                         goto out;
1393
1394                 /*
1395                  * Something has shifted, try again.
1396                  */
1397                 put_task_struct(next_task);
1398                 next_task = task;
1399                 goto retry;
1400         }
1401
1402         deactivate_task(rq, next_task, 0);
1403         set_task_cpu(next_task, lowest_rq->cpu);
1404         activate_task(lowest_rq, next_task, 0);
1405
1406         resched_task(lowest_rq->curr);
1407
1408         double_unlock_balance(rq, lowest_rq);
1409
1410 out:
1411         put_task_struct(next_task);
1412
1413         return 1;
1414 }
1415
1416 static void push_rt_tasks(struct rq *rq)
1417 {
1418         /* push_rt_task will return true if it moved an RT */
1419         while (push_rt_task(rq))
1420                 ;
1421 }
1422
1423 static int pull_rt_task(struct rq *this_rq)
1424 {
1425         int this_cpu = this_rq->cpu, ret = 0, cpu;
1426         struct task_struct *p;
1427         struct rq *src_rq;
1428
1429         if (likely(!rt_overloaded(this_rq)))
1430                 return 0;
1431
1432         for_each_cpu(cpu, this_rq->rd->rto_mask) {
1433                 if (this_cpu == cpu)
1434                         continue;
1435
1436                 src_rq = cpu_rq(cpu);
1437
1438                 /*
1439                  * Don't bother taking the src_rq->lock if the next highest
1440                  * task is known to be lower-priority than our current task.
1441                  * This may look racy, but if this value is about to go
1442                  * logically higher, the src_rq will push this task away.
1443                  * And if its going logically lower, we do not care
1444                  */
1445                 if (src_rq->rt.highest_prio.next >=
1446                     this_rq->rt.highest_prio.curr)
1447                         continue;
1448
1449                 /*
1450                  * We can potentially drop this_rq's lock in
1451                  * double_lock_balance, and another CPU could
1452                  * alter this_rq
1453                  */
1454                 double_lock_balance(this_rq, src_rq);
1455
1456                 /*
1457                  * Are there still pullable RT tasks?
1458                  */
1459                 if (src_rq->rt.rt_nr_running <= 1)
1460                         goto skip;
1461
1462                 p = pick_next_highest_task_rt(src_rq, this_cpu);
1463
1464                 /*
1465                  * Do we have an RT task that preempts
1466                  * the to-be-scheduled task?
1467                  */
1468                 if (p && (p->prio < this_rq->rt.highest_prio.curr)) {
1469                         WARN_ON(p == src_rq->curr);
1470                         WARN_ON(!p->on_rq);
1471
1472                         /*
1473                          * There's a chance that p is higher in priority
1474                          * than what's currently running on its cpu.
1475                          * This is just that p is wakeing up and hasn't
1476                          * had a chance to schedule. We only pull
1477                          * p if it is lower in priority than the
1478                          * current task on the run queue
1479                          */
1480                         if (p->prio < src_rq->curr->prio)
1481                                 goto skip;
1482
1483                         ret = 1;
1484
1485                         deactivate_task(src_rq, p, 0);
1486                         set_task_cpu(p, this_cpu);
1487                         activate_task(this_rq, p, 0);
1488                         /*
1489                          * We continue with the search, just in
1490                          * case there's an even higher prio task
1491                          * in another runqueue. (low likelihood
1492                          * but possible)
1493                          */
1494                 }
1495 skip:
1496                 double_unlock_balance(this_rq, src_rq);
1497         }
1498
1499         return ret;
1500 }
1501
1502 static void pre_schedule_rt(struct rq *rq, struct task_struct *prev)
1503 {
1504         /* Try to pull RT tasks here if we lower this rq's prio */
1505         if (unlikely(rt_task(prev)) && rq->rt.highest_prio.curr > prev->prio)
1506                 pull_rt_task(rq);
1507 }
1508
1509 static void post_schedule_rt(struct rq *rq)
1510 {
1511         push_rt_tasks(rq);
1512 }
1513
1514 /*
1515  * If we are not running and we are not going to reschedule soon, we should
1516  * try to push tasks away now
1517  */
1518 static void task_woken_rt(struct rq *rq, struct task_struct *p)
1519 {
1520         if (!task_running(rq, p) &&
1521             !test_tsk_need_resched(rq->curr) &&
1522             has_pushable_tasks(rq) &&
1523             p->rt.nr_cpus_allowed > 1 &&
1524             rt_task(rq->curr) &&
1525             (rq->curr->rt.nr_cpus_allowed < 2 ||
1526              rq->curr->prio < p->prio))
1527                 push_rt_tasks(rq);
1528 }
1529
1530 static void set_cpus_allowed_rt(struct task_struct *p,
1531                                 const struct cpumask *new_mask)
1532 {
1533         int weight = cpumask_weight(new_mask);
1534
1535         BUG_ON(!rt_task(p));
1536
1537         /*
1538          * Update the migration status of the RQ if we have an RT task
1539          * which is running AND changing its weight value.
1540          */
1541         if (p->on_rq && (weight != p->rt.nr_cpus_allowed)) {
1542                 struct rq *rq = task_rq(p);
1543
1544                 if (!task_current(rq, p)) {
1545                         /*
1546                          * Make sure we dequeue this task from the pushable list
1547                          * before going further.  It will either remain off of
1548                          * the list because we are no longer pushable, or it
1549                          * will be requeued.
1550                          */
1551                         if (p->rt.nr_cpus_allowed > 1)
1552                                 dequeue_pushable_task(rq, p);
1553
1554                         /*
1555                          * Requeue if our weight is changing and still > 1
1556                          */
1557                         if (weight > 1)
1558                                 enqueue_pushable_task(rq, p);
1559
1560                 }
1561
1562                 if ((p->rt.nr_cpus_allowed <= 1) && (weight > 1)) {
1563                         rq->rt.rt_nr_migratory++;
1564                 } else if ((p->rt.nr_cpus_allowed > 1) && (weight <= 1)) {
1565                         BUG_ON(!rq->rt.rt_nr_migratory);
1566                         rq->rt.rt_nr_migratory--;
1567                 }
1568
1569                 update_rt_migration(&rq->rt);
1570         }
1571
1572         cpumask_copy(&p->cpus_allowed, new_mask);
1573         p->rt.nr_cpus_allowed = weight;
1574 }
1575
1576 /* Assumes rq->lock is held */
1577 static void rq_online_rt(struct rq *rq)
1578 {
1579         if (rq->rt.overloaded)
1580                 rt_set_overload(rq);
1581
1582         __enable_runtime(rq);
1583
1584         cpupri_set(&rq->rd->cpupri, rq->cpu, rq->rt.highest_prio.curr);
1585 }
1586
1587 /* Assumes rq->lock is held */
1588 static void rq_offline_rt(struct rq *rq)
1589 {
1590         if (rq->rt.overloaded)
1591                 rt_clear_overload(rq);
1592
1593         __disable_runtime(rq);
1594
1595         cpupri_set(&rq->rd->cpupri, rq->cpu, CPUPRI_INVALID);
1596 }
1597
1598 /*
1599  * When switch from the rt queue, we bring ourselves to a position
1600  * that we might want to pull RT tasks from other runqueues.
1601  */
1602 static void switched_from_rt(struct rq *rq, struct task_struct *p)
1603 {
1604         /*
1605          * If there are other RT tasks then we will reschedule
1606          * and the scheduling of the other RT tasks will handle
1607          * the balancing. But if we are the last RT task
1608          * we may need to handle the pulling of RT tasks
1609          * now.
1610          */
1611         if (p->on_rq && !rq->rt.rt_nr_running)
1612                 pull_rt_task(rq);
1613 }
1614
1615 static inline void init_sched_rt_class(void)
1616 {
1617         unsigned int i;
1618
1619         for_each_possible_cpu(i)
1620                 zalloc_cpumask_var_node(&per_cpu(local_cpu_mask, i),
1621                                         GFP_KERNEL, cpu_to_node(i));
1622 }
1623 #endif /* CONFIG_SMP */
1624
1625 /*
1626  * When switching a task to RT, we may overload the runqueue
1627  * with RT tasks. In this case we try to push them off to
1628  * other runqueues.
1629  */
1630 static void switched_to_rt(struct rq *rq, struct task_struct *p)
1631 {
1632         int check_resched = 1;
1633
1634         /*
1635          * If we are already running, then there's nothing
1636          * that needs to be done. But if we are not running
1637          * we may need to preempt the current running task.
1638          * If that current running task is also an RT task
1639          * then see if we can move to another run queue.
1640          */
1641         if (p->on_rq && rq->curr != p) {
1642 #ifdef CONFIG_SMP
1643                 if (rq->rt.overloaded && push_rt_task(rq) &&
1644                     /* Don't resched if we changed runqueues */
1645                     rq != task_rq(p))
1646                         check_resched = 0;
1647 #endif /* CONFIG_SMP */
1648                 if (check_resched && p->prio < rq->curr->prio)
1649                         resched_task(rq->curr);
1650         }
1651 }
1652
1653 /*
1654  * Priority of the task has changed. This may cause
1655  * us to initiate a push or pull.
1656  */
1657 static void
1658 prio_changed_rt(struct rq *rq, struct task_struct *p, int oldprio)
1659 {
1660         if (!p->on_rq)
1661                 return;
1662
1663         if (rq->curr == p) {
1664 #ifdef CONFIG_SMP
1665                 /*
1666                  * If our priority decreases while running, we
1667                  * may need to pull tasks to this runqueue.
1668                  */
1669                 if (oldprio < p->prio)
1670                         pull_rt_task(rq);
1671                 /*
1672                  * If there's a higher priority task waiting to run
1673                  * then reschedule. Note, the above pull_rt_task
1674                  * can release the rq lock and p could migrate.
1675                  * Only reschedule if p is still on the same runqueue.
1676                  */
1677                 if (p->prio > rq->rt.highest_prio.curr && rq->curr == p)
1678                         resched_task(p);
1679 #else
1680                 /* For UP simply resched on drop of prio */
1681                 if (oldprio < p->prio)
1682                         resched_task(p);
1683 #endif /* CONFIG_SMP */
1684         } else {
1685                 /*
1686                  * This task is not running, but if it is
1687                  * greater than the current running task
1688                  * then reschedule.
1689                  */
1690                 if (p->prio < rq->curr->prio)
1691                         resched_task(rq->curr);
1692         }
1693 }
1694
1695 static void watchdog(struct rq *rq, struct task_struct *p)
1696 {
1697         unsigned long soft, hard;
1698
1699         /* max may change after cur was read, this will be fixed next tick */
1700         soft = task_rlimit(p, RLIMIT_RTTIME);
1701         hard = task_rlimit_max(p, RLIMIT_RTTIME);
1702
1703         if (soft != RLIM_INFINITY) {
1704                 unsigned long next;
1705
1706                 p->rt.timeout++;
1707                 next = DIV_ROUND_UP(min(soft, hard), USEC_PER_SEC/HZ);
1708                 if (p->rt.timeout > next)
1709                         p->cputime_expires.sched_exp = p->se.sum_exec_runtime;
1710         }
1711 }
1712
1713 static void task_tick_rt(struct rq *rq, struct task_struct *p, int queued)
1714 {
1715         update_curr_rt(rq);
1716
1717         watchdog(rq, p);
1718
1719         /*
1720          * RR tasks need a special form of timeslice management.
1721          * FIFO tasks have no timeslices.
1722          */
1723         if (p->policy != SCHED_RR)
1724                 return;
1725
1726         if (--p->rt.time_slice)
1727                 return;
1728
1729         p->rt.time_slice = DEF_TIMESLICE;
1730
1731         /*
1732          * Requeue to the end of queue if we are not the only element
1733          * on the queue:
1734          */
1735         if (p->rt.run_list.prev != p->rt.run_list.next) {
1736                 requeue_task_rt(rq, p, 0);
1737                 set_tsk_need_resched(p);
1738         }
1739 }
1740
1741 static void set_curr_task_rt(struct rq *rq)
1742 {
1743         struct task_struct *p = rq->curr;
1744
1745         p->se.exec_start = rq->clock_task;
1746
1747         /* The running task is never eligible for pushing */
1748         dequeue_pushable_task(rq, p);
1749 }
1750
1751 static unsigned int get_rr_interval_rt(struct rq *rq, struct task_struct *task)
1752 {
1753         /*
1754          * Time slice is 0 for SCHED_FIFO tasks
1755          */
1756         if (task->policy == SCHED_RR)
1757                 return DEF_TIMESLICE;
1758         else
1759                 return 0;
1760 }
1761
1762 static const struct sched_class rt_sched_class = {
1763         .next                   = &fair_sched_class,
1764         .enqueue_task           = enqueue_task_rt,
1765         .dequeue_task           = dequeue_task_rt,
1766         .yield_task             = yield_task_rt,
1767
1768         .check_preempt_curr     = check_preempt_curr_rt,
1769
1770         .pick_next_task         = pick_next_task_rt,
1771         .put_prev_task          = put_prev_task_rt,
1772
1773 #ifdef CONFIG_SMP
1774         .select_task_rq         = select_task_rq_rt,
1775
1776         .set_cpus_allowed       = set_cpus_allowed_rt,
1777         .rq_online              = rq_online_rt,
1778         .rq_offline             = rq_offline_rt,
1779         .pre_schedule           = pre_schedule_rt,
1780         .post_schedule          = post_schedule_rt,
1781         .task_woken             = task_woken_rt,
1782         .switched_from          = switched_from_rt,
1783 #endif
1784
1785         .set_curr_task          = set_curr_task_rt,
1786         .task_tick              = task_tick_rt,
1787
1788         .get_rr_interval        = get_rr_interval_rt,
1789
1790         .prio_changed           = prio_changed_rt,
1791         .switched_to            = switched_to_rt,
1792 };
1793
1794 #ifdef CONFIG_SCHED_DEBUG
1795 extern void print_rt_rq(struct seq_file *m, int cpu, struct rt_rq *rt_rq);
1796
1797 static void print_rt_stats(struct seq_file *m, int cpu)
1798 {
1799         struct rt_rq *rt_rq;
1800
1801         rcu_read_lock();
1802         for_each_leaf_rt_rq(rt_rq, cpu_rq(cpu))
1803                 print_rt_rq(m, cpu, rt_rq);
1804         rcu_read_unlock();
1805 }
1806 #endif /* CONFIG_SCHED_DEBUG */
1807