Root/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
10static 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
18static inline struct rq *rq_of_rt_rq(struct rt_rq *rt_rq)
19{
20    return rt_rq->rq;
21}
22
23static 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
32static 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
37static inline struct rq *rq_of_rt_rq(struct rt_rq *rt_rq)
38{
39    return container_of(rt_rq, struct rq, rt);
40}
41
42static 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
54static inline int rt_overloaded(struct rq *rq)
55{
56    return atomic_read(&rq->rd->rto_count);
57}
58
59static 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
76static 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
86static 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
99static 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
113static 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
127static 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
134static void dequeue_pushable_task(struct rq *rq, struct task_struct *p)
135{
136    plist_del(&p->pushable_tasks, &rq->rt.pushable_tasks);
137}
138
139static inline int has_pushable_tasks(struct rq *rq)
140{
141    return !plist_head_empty(&rq->rt.pushable_tasks);
142}
143
144#else
145
146static inline void enqueue_pushable_task(struct rq *rq, struct task_struct *p)
147{
148}
149
150static inline void dequeue_pushable_task(struct rq *rq, struct task_struct *p)
151{
152}
153
154static inline
155void inc_rt_migration(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq)
156{
157}
158
159static inline
160void dec_rt_migration(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq)
161{
162}
163
164#endif /* CONFIG_SMP */
165
166static 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
173static 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
181static 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
186static 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
192static 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
203static inline struct rt_rq *group_rt_rq(struct sched_rt_entity *rt_se)
204{
205    return rt_se->my_q;
206}
207
208static void enqueue_rt_entity(struct sched_rt_entity *rt_se, bool head);
209static void dequeue_rt_entity(struct sched_rt_entity *rt_se);
210
211static 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
228static 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
239static inline int rt_rq_throttled(struct rt_rq *rt_rq)
240{
241    return rt_rq->rt_throttled && !rt_rq->rt_nr_boosted;
242}
243
244static 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
257static inline const struct cpumask *sched_rt_period_mask(void)
258{
259    return cpu_rq(smp_processor_id())->rd->span;
260}
261#else
262static inline const struct cpumask *sched_rt_period_mask(void)
263{
264    return cpu_online_mask;
265}
266#endif
267
268static inline
269struct 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
274static 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
281static inline u64 sched_rt_runtime(struct rt_rq *rt_rq)
282{
283    return rt_rq->rt_runtime;
284}
285
286static inline u64 sched_rt_period(struct rt_rq *rt_rq)
287{
288    return ktime_to_ns(def_rt_bandwidth.rt_period);
289}
290
291static inline void list_add_leaf_rt_rq(struct rt_rq *rt_rq)
292{
293}
294
295static 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
305static inline struct rt_rq *group_rt_rq(struct sched_rt_entity *rt_se)
306{
307    return NULL;
308}
309
310static 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
316static inline void sched_rt_rq_dequeue(struct rt_rq *rt_rq)
317{
318}
319
320static inline int rt_rq_throttled(struct rt_rq *rt_rq)
321{
322    return rt_rq->rt_throttled;
323}
324
325static inline const struct cpumask *sched_rt_period_mask(void)
326{
327    return cpu_online_mask;
328}
329
330static inline
331struct rt_rq *sched_rt_period_rt_rq(struct rt_bandwidth *rt_b, int cpu)
332{
333    return &cpu_rq(cpu)->rt;
334}
335
336static 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 */
347static 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        }
391next:
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 */
402static 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);
468balanced:
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
479static 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
488static 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
511static 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
520static 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 */
533static inline int balance_runtime(struct rt_rq *rt_rq)
534{
535    return 0;
536}
537#endif /* CONFIG_SMP */
538
539static 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
583static 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
595static 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 */
625static 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
667static struct task_struct *pick_next_highest_task_rt(struct rq *rq, int cpu);
668
669static 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
679static void
680inc_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
710static void
711dec_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
724static inline
725void inc_rt_prio_smp(struct rt_rq *rt_rq, int prio, int prev_prio) {}
726static inline
727void 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
732static void
733inc_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
743static void
744dec_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
771static inline void inc_rt_prio(struct rt_rq *rt_rq, int prio) {}
772static 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
778static void
779inc_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
788static void
789dec_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
799static void
800inc_rt_group(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq)
801{
802    start_rt_bandwidth(&def_rt_bandwidth);
803}
804
805static inline
806void dec_rt_group(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq) {}
807
808#endif /* CONFIG_RT_GROUP_SCHED */
809
810static inline
811void 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
823static inline
824void 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
835static 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
863static 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 */
881static 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
896static 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
903static 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 */
918static void
919enqueue_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
932static 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 */
946static void
947requeue_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
960static 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
971static void yield_task_rt(struct rq *rq)
972{
973    requeue_task_rt(rq, rq->curr, 0);
974}
975
976#ifdef CONFIG_SMP
977static int find_lowest_rq(struct task_struct *task);
978
979static int
980select_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
1017static 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 */
1043static 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
1068static 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
1085static 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
1111static 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
1130static 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 (p->se.on_rq && 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
1148static void deactivate_task(struct rq *rq, struct task_struct *p, int sleep);
1149
1150static 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 */
1160static 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);
1171next_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
1197static DEFINE_PER_CPU(cpumask_var_t, local_cpu_mask);
1198
1199static 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 */
1264static 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->se.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
1310static 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->se.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 */
1335static 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
1347retry:
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 hasnt 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
1410out:
1411    put_task_struct(next_task);
1412
1413    return 1;
1414}
1415
1416static 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
1423static 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->se.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 likelyhood
1492             * but possible)
1493             */
1494        }
1495skip:
1496        double_unlock_balance(this_rq, src_rq);
1497    }
1498
1499    return ret;
1500}
1501
1502static 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
1509static 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 */
1518static 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
1530static 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->se.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 */
1577static 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 */
1588static 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 */
1602static void switched_from_rt(struct rq *rq, struct task_struct *p,
1603               int running)
1604{
1605    /*
1606     * If there are other RT tasks then we will reschedule
1607     * and the scheduling of the other RT tasks will handle
1608     * the balancing. But if we are the last RT task
1609     * we may need to handle the pulling of RT tasks
1610     * now.
1611     */
1612    if (!rq->rt.rt_nr_running)
1613        pull_rt_task(rq);
1614}
1615
1616static inline void init_sched_rt_class(void)
1617{
1618    unsigned int i;
1619
1620    for_each_possible_cpu(i)
1621        zalloc_cpumask_var_node(&per_cpu(local_cpu_mask, i),
1622                    GFP_KERNEL, cpu_to_node(i));
1623}
1624#endif /* CONFIG_SMP */
1625
1626/*
1627 * When switching a task to RT, we may overload the runqueue
1628 * with RT tasks. In this case we try to push them off to
1629 * other runqueues.
1630 */
1631static void switched_to_rt(struct rq *rq, struct task_struct *p,
1632               int running)
1633{
1634    int check_resched = 1;
1635
1636    /*
1637     * If we are already running, then there's nothing
1638     * that needs to be done. But if we are not running
1639     * we may need to preempt the current running task.
1640     * If that current running task is also an RT task
1641     * then see if we can move to another run queue.
1642     */
1643    if (!running) {
1644#ifdef CONFIG_SMP
1645        if (rq->rt.overloaded && push_rt_task(rq) &&
1646            /* Don't resched if we changed runqueues */
1647            rq != task_rq(p))
1648            check_resched = 0;
1649#endif /* CONFIG_SMP */
1650        if (check_resched && p->prio < rq->curr->prio)
1651            resched_task(rq->curr);
1652    }
1653}
1654
1655/*
1656 * Priority of the task has changed. This may cause
1657 * us to initiate a push or pull.
1658 */
1659static void prio_changed_rt(struct rq *rq, struct task_struct *p,
1660                int oldprio, int running)
1661{
1662    if (running) {
1663#ifdef CONFIG_SMP
1664        /*
1665         * If our priority decreases while running, we
1666         * may need to pull tasks to this runqueue.
1667         */
1668        if (oldprio < p->prio)
1669            pull_rt_task(rq);
1670        /*
1671         * If there's a higher priority task waiting to run
1672         * then reschedule. Note, the above pull_rt_task
1673         * can release the rq lock and p could migrate.
1674         * Only reschedule if p is still on the same runqueue.
1675         */
1676        if (p->prio > rq->rt.highest_prio.curr && rq->curr == p)
1677            resched_task(p);
1678#else
1679        /* For UP simply resched on drop of prio */
1680        if (oldprio < p->prio)
1681            resched_task(p);
1682#endif /* CONFIG_SMP */
1683    } else {
1684        /*
1685         * This task is not running, but if it is
1686         * greater than the current running task
1687         * then reschedule.
1688         */
1689        if (p->prio < rq->curr->prio)
1690            resched_task(rq->curr);
1691    }
1692}
1693
1694static void watchdog(struct rq *rq, struct task_struct *p)
1695{
1696    unsigned long soft, hard;
1697
1698    /* max may change after cur was read, this will be fixed next tick */
1699    soft = task_rlimit(p, RLIMIT_RTTIME);
1700    hard = task_rlimit_max(p, RLIMIT_RTTIME);
1701
1702    if (soft != RLIM_INFINITY) {
1703        unsigned long next;
1704
1705        p->rt.timeout++;
1706        next = DIV_ROUND_UP(min(soft, hard), USEC_PER_SEC/HZ);
1707        if (p->rt.timeout > next)
1708            p->cputime_expires.sched_exp = p->se.sum_exec_runtime;
1709    }
1710}
1711
1712static void task_tick_rt(struct rq *rq, struct task_struct *p, int queued)
1713{
1714    update_curr_rt(rq);
1715
1716    watchdog(rq, p);
1717
1718    /*
1719     * RR tasks need a special form of timeslice management.
1720     * FIFO tasks have no timeslices.
1721     */
1722    if (p->policy != SCHED_RR)
1723        return;
1724
1725    if (--p->rt.time_slice)
1726        return;
1727
1728    p->rt.time_slice = DEF_TIMESLICE;
1729
1730    /*
1731     * Requeue to the end of queue if we are not the only element
1732     * on the queue:
1733     */
1734    if (p->rt.run_list.prev != p->rt.run_list.next) {
1735        requeue_task_rt(rq, p, 0);
1736        set_tsk_need_resched(p);
1737    }
1738}
1739
1740static void set_curr_task_rt(struct rq *rq)
1741{
1742    struct task_struct *p = rq->curr;
1743
1744    p->se.exec_start = rq->clock_task;
1745
1746    /* The running task is never eligible for pushing */
1747    dequeue_pushable_task(rq, p);
1748}
1749
1750static unsigned int get_rr_interval_rt(struct rq *rq, struct task_struct *task)
1751{
1752    /*
1753     * Time slice is 0 for SCHED_FIFO tasks
1754     */
1755    if (task->policy == SCHED_RR)
1756        return DEF_TIMESLICE;
1757    else
1758        return 0;
1759}
1760
1761static const struct sched_class rt_sched_class = {
1762    .next = &fair_sched_class,
1763    .enqueue_task = enqueue_task_rt,
1764    .dequeue_task = dequeue_task_rt,
1765    .yield_task = yield_task_rt,
1766
1767    .check_preempt_curr = check_preempt_curr_rt,
1768
1769    .pick_next_task = pick_next_task_rt,
1770    .put_prev_task = put_prev_task_rt,
1771
1772#ifdef CONFIG_SMP
1773    .select_task_rq = select_task_rq_rt,
1774
1775    .set_cpus_allowed = set_cpus_allowed_rt,
1776    .rq_online = rq_online_rt,
1777    .rq_offline = rq_offline_rt,
1778    .pre_schedule = pre_schedule_rt,
1779    .post_schedule = post_schedule_rt,
1780    .task_woken = task_woken_rt,
1781    .switched_from = switched_from_rt,
1782#endif
1783
1784    .set_curr_task = set_curr_task_rt,
1785    .task_tick = task_tick_rt,
1786
1787    .get_rr_interval = get_rr_interval_rt,
1788
1789    .prio_changed = prio_changed_rt,
1790    .switched_to = switched_to_rt,
1791};
1792
1793#ifdef CONFIG_SCHED_DEBUG
1794extern void print_rt_rq(struct seq_file *m, int cpu, struct rt_rq *rt_rq);
1795
1796static void print_rt_stats(struct seq_file *m, int cpu)
1797{
1798    struct rt_rq *rt_rq;
1799
1800    rcu_read_lock();
1801    for_each_leaf_rt_rq(rt_rq, cpu_rq(cpu))
1802        print_rt_rq(m, cpu, rt_rq);
1803    rcu_read_unlock();
1804}
1805#endif /* CONFIG_SCHED_DEBUG */
1806
1807

Archive Download this file

Branches:
ben-wpan
ben-wpan-stefan
javiroman/ks7010
jz-2.6.34
jz-2.6.34-rc5
jz-2.6.34-rc6
jz-2.6.34-rc7
jz-2.6.35
jz-2.6.36
jz-2.6.37
jz-2.6.38
jz-2.6.39
jz-3.0
jz-3.1
jz-3.11
jz-3.12
jz-3.13
jz-3.15
jz-3.16
jz-3.18-dt
jz-3.2
jz-3.3
jz-3.4
jz-3.5
jz-3.6
jz-3.6-rc2-pwm
jz-3.9
jz-3.9-clk
jz-3.9-rc8
jz47xx
jz47xx-2.6.38
master

Tags:
od-2011-09-04
od-2011-09-18
v2.6.34-rc5
v2.6.34-rc6
v2.6.34-rc7
v3.9



interactive