]>
Commit | Line | Data |
---|---|---|
b2441318 | 1 | // SPDX-License-Identifier: GPL-2.0 |
bb44e5d1 IM |
2 | /* |
3 | * Real-Time Scheduling Class (mapped to the SCHED_FIFO and SCHED_RR | |
4 | * policies) | |
5 | */ | |
6 | ||
029632fb PZ |
7 | #include "sched.h" |
8 | ||
9 | #include <linux/slab.h> | |
b6366f04 | 10 | #include <linux/irq_work.h> |
029632fb | 11 | |
ce0dbbbb | 12 | int sched_rr_timeslice = RR_TIMESLICE; |
975e155e | 13 | int sysctl_sched_rr_timeslice = (MSEC_PER_SEC / HZ) * RR_TIMESLICE; |
ce0dbbbb | 14 | |
029632fb PZ |
15 | static int do_sched_rt_period_timer(struct rt_bandwidth *rt_b, int overrun); |
16 | ||
17 | struct rt_bandwidth def_rt_bandwidth; | |
18 | ||
19 | static enum hrtimer_restart sched_rt_period_timer(struct hrtimer *timer) | |
20 | { | |
21 | struct rt_bandwidth *rt_b = | |
22 | container_of(timer, struct rt_bandwidth, rt_period_timer); | |
029632fb | 23 | int idle = 0; |
77a4d1a1 | 24 | int overrun; |
029632fb | 25 | |
77a4d1a1 | 26 | raw_spin_lock(&rt_b->rt_runtime_lock); |
029632fb | 27 | for (;;) { |
77a4d1a1 | 28 | overrun = hrtimer_forward_now(timer, rt_b->rt_period); |
029632fb PZ |
29 | if (!overrun) |
30 | break; | |
31 | ||
77a4d1a1 | 32 | raw_spin_unlock(&rt_b->rt_runtime_lock); |
029632fb | 33 | idle = do_sched_rt_period_timer(rt_b, overrun); |
77a4d1a1 | 34 | raw_spin_lock(&rt_b->rt_runtime_lock); |
029632fb | 35 | } |
4cfafd30 PZ |
36 | if (idle) |
37 | rt_b->rt_period_active = 0; | |
77a4d1a1 | 38 | raw_spin_unlock(&rt_b->rt_runtime_lock); |
029632fb PZ |
39 | |
40 | return idle ? HRTIMER_NORESTART : HRTIMER_RESTART; | |
41 | } | |
42 | ||
43 | void init_rt_bandwidth(struct rt_bandwidth *rt_b, u64 period, u64 runtime) | |
44 | { | |
45 | rt_b->rt_period = ns_to_ktime(period); | |
46 | rt_b->rt_runtime = runtime; | |
47 | ||
48 | raw_spin_lock_init(&rt_b->rt_runtime_lock); | |
49 | ||
50 | hrtimer_init(&rt_b->rt_period_timer, | |
51 | CLOCK_MONOTONIC, HRTIMER_MODE_REL); | |
52 | rt_b->rt_period_timer.function = sched_rt_period_timer; | |
53 | } | |
54 | ||
55 | static void start_rt_bandwidth(struct rt_bandwidth *rt_b) | |
56 | { | |
57 | if (!rt_bandwidth_enabled() || rt_b->rt_runtime == RUNTIME_INF) | |
58 | return; | |
59 | ||
029632fb | 60 | raw_spin_lock(&rt_b->rt_runtime_lock); |
4cfafd30 PZ |
61 | if (!rt_b->rt_period_active) { |
62 | rt_b->rt_period_active = 1; | |
c3a990dc SR |
63 | /* |
64 | * SCHED_DEADLINE updates the bandwidth, as a run away | |
65 | * RT task with a DL task could hog a CPU. But DL does | |
66 | * not reset the period. If a deadline task was running | |
67 | * without an RT task running, it can cause RT tasks to | |
68 | * throttle when they start up. Kick the timer right away | |
69 | * to update the period. | |
70 | */ | |
71 | hrtimer_forward_now(&rt_b->rt_period_timer, ns_to_ktime(0)); | |
4cfafd30 PZ |
72 | hrtimer_start_expires(&rt_b->rt_period_timer, HRTIMER_MODE_ABS_PINNED); |
73 | } | |
029632fb PZ |
74 | raw_spin_unlock(&rt_b->rt_runtime_lock); |
75 | } | |
76 | ||
89b41108 | 77 | #if defined(CONFIG_SMP) && defined(HAVE_RT_PUSH_IPI) |
b6366f04 SR |
78 | static void push_irq_work_func(struct irq_work *work); |
79 | #endif | |
80 | ||
07c54f7a | 81 | void init_rt_rq(struct rt_rq *rt_rq) |
029632fb PZ |
82 | { |
83 | struct rt_prio_array *array; | |
84 | int i; | |
85 | ||
86 | array = &rt_rq->active; | |
87 | for (i = 0; i < MAX_RT_PRIO; i++) { | |
88 | INIT_LIST_HEAD(array->queue + i); | |
89 | __clear_bit(i, array->bitmap); | |
90 | } | |
91 | /* delimiter for bitsearch: */ | |
92 | __set_bit(MAX_RT_PRIO, array->bitmap); | |
93 | ||
94 | #if defined CONFIG_SMP | |
95 | rt_rq->highest_prio.curr = MAX_RT_PRIO; | |
96 | rt_rq->highest_prio.next = MAX_RT_PRIO; | |
97 | rt_rq->rt_nr_migratory = 0; | |
98 | rt_rq->overloaded = 0; | |
99 | plist_head_init(&rt_rq->pushable_tasks); | |
b6366f04 SR |
100 | |
101 | #ifdef HAVE_RT_PUSH_IPI | |
102 | rt_rq->push_flags = 0; | |
103 | rt_rq->push_cpu = nr_cpu_ids; | |
104 | raw_spin_lock_init(&rt_rq->push_lock); | |
105 | init_irq_work(&rt_rq->push_work, push_irq_work_func); | |
029632fb | 106 | #endif |
b6366f04 | 107 | #endif /* CONFIG_SMP */ |
f4ebcbc0 KT |
108 | /* We start is dequeued state, because no RT tasks are queued */ |
109 | rt_rq->rt_queued = 0; | |
029632fb PZ |
110 | |
111 | rt_rq->rt_time = 0; | |
112 | rt_rq->rt_throttled = 0; | |
113 | rt_rq->rt_runtime = 0; | |
114 | raw_spin_lock_init(&rt_rq->rt_runtime_lock); | |
115 | } | |
116 | ||
8f48894f | 117 | #ifdef CONFIG_RT_GROUP_SCHED |
029632fb PZ |
118 | static void destroy_rt_bandwidth(struct rt_bandwidth *rt_b) |
119 | { | |
120 | hrtimer_cancel(&rt_b->rt_period_timer); | |
121 | } | |
8f48894f PZ |
122 | |
123 | #define rt_entity_is_task(rt_se) (!(rt_se)->my_q) | |
124 | ||
398a153b GH |
125 | static inline struct task_struct *rt_task_of(struct sched_rt_entity *rt_se) |
126 | { | |
8f48894f PZ |
127 | #ifdef CONFIG_SCHED_DEBUG |
128 | WARN_ON_ONCE(!rt_entity_is_task(rt_se)); | |
129 | #endif | |
398a153b GH |
130 | return container_of(rt_se, struct task_struct, rt); |
131 | } | |
132 | ||
398a153b GH |
133 | static inline struct rq *rq_of_rt_rq(struct rt_rq *rt_rq) |
134 | { | |
135 | return rt_rq->rq; | |
136 | } | |
137 | ||
138 | static inline struct rt_rq *rt_rq_of_se(struct sched_rt_entity *rt_se) | |
139 | { | |
140 | return rt_se->rt_rq; | |
141 | } | |
142 | ||
653d07a6 KT |
143 | static inline struct rq *rq_of_rt_se(struct sched_rt_entity *rt_se) |
144 | { | |
145 | struct rt_rq *rt_rq = rt_se->rt_rq; | |
146 | ||
147 | return rt_rq->rq; | |
148 | } | |
149 | ||
029632fb PZ |
150 | void free_rt_sched_group(struct task_group *tg) |
151 | { | |
152 | int i; | |
153 | ||
154 | if (tg->rt_se) | |
155 | destroy_rt_bandwidth(&tg->rt_bandwidth); | |
156 | ||
157 | for_each_possible_cpu(i) { | |
158 | if (tg->rt_rq) | |
159 | kfree(tg->rt_rq[i]); | |
160 | if (tg->rt_se) | |
161 | kfree(tg->rt_se[i]); | |
162 | } | |
163 | ||
164 | kfree(tg->rt_rq); | |
165 | kfree(tg->rt_se); | |
166 | } | |
167 | ||
168 | void init_tg_rt_entry(struct task_group *tg, struct rt_rq *rt_rq, | |
169 | struct sched_rt_entity *rt_se, int cpu, | |
170 | struct sched_rt_entity *parent) | |
171 | { | |
172 | struct rq *rq = cpu_rq(cpu); | |
173 | ||
174 | rt_rq->highest_prio.curr = MAX_RT_PRIO; | |
175 | rt_rq->rt_nr_boosted = 0; | |
176 | rt_rq->rq = rq; | |
177 | rt_rq->tg = tg; | |
178 | ||
179 | tg->rt_rq[cpu] = rt_rq; | |
180 | tg->rt_se[cpu] = rt_se; | |
181 | ||
182 | if (!rt_se) | |
183 | return; | |
184 | ||
185 | if (!parent) | |
186 | rt_se->rt_rq = &rq->rt; | |
187 | else | |
188 | rt_se->rt_rq = parent->my_q; | |
189 | ||
190 | rt_se->my_q = rt_rq; | |
191 | rt_se->parent = parent; | |
192 | INIT_LIST_HEAD(&rt_se->run_list); | |
193 | } | |
194 | ||
195 | int alloc_rt_sched_group(struct task_group *tg, struct task_group *parent) | |
196 | { | |
197 | struct rt_rq *rt_rq; | |
198 | struct sched_rt_entity *rt_se; | |
199 | int i; | |
200 | ||
201 | tg->rt_rq = kzalloc(sizeof(rt_rq) * nr_cpu_ids, GFP_KERNEL); | |
202 | if (!tg->rt_rq) | |
203 | goto err; | |
204 | tg->rt_se = kzalloc(sizeof(rt_se) * nr_cpu_ids, GFP_KERNEL); | |
205 | if (!tg->rt_se) | |
206 | goto err; | |
207 | ||
208 | init_rt_bandwidth(&tg->rt_bandwidth, | |
209 | ktime_to_ns(def_rt_bandwidth.rt_period), 0); | |
210 | ||
211 | for_each_possible_cpu(i) { | |
212 | rt_rq = kzalloc_node(sizeof(struct rt_rq), | |
213 | GFP_KERNEL, cpu_to_node(i)); | |
214 | if (!rt_rq) | |
215 | goto err; | |
216 | ||
217 | rt_se = kzalloc_node(sizeof(struct sched_rt_entity), | |
218 | GFP_KERNEL, cpu_to_node(i)); | |
219 | if (!rt_se) | |
220 | goto err_free_rq; | |
221 | ||
07c54f7a | 222 | init_rt_rq(rt_rq); |
029632fb PZ |
223 | rt_rq->rt_runtime = tg->rt_bandwidth.rt_runtime; |
224 | init_tg_rt_entry(tg, rt_rq, rt_se, i, parent->rt_se[i]); | |
225 | } | |
226 | ||
227 | return 1; | |
228 | ||
229 | err_free_rq: | |
230 | kfree(rt_rq); | |
231 | err: | |
232 | return 0; | |
233 | } | |
234 | ||
398a153b GH |
235 | #else /* CONFIG_RT_GROUP_SCHED */ |
236 | ||
a1ba4d8b PZ |
237 | #define rt_entity_is_task(rt_se) (1) |
238 | ||
8f48894f PZ |
239 | static inline struct task_struct *rt_task_of(struct sched_rt_entity *rt_se) |
240 | { | |
241 | return container_of(rt_se, struct task_struct, rt); | |
242 | } | |
243 | ||
398a153b GH |
244 | static inline struct rq *rq_of_rt_rq(struct rt_rq *rt_rq) |
245 | { | |
246 | return container_of(rt_rq, struct rq, rt); | |
247 | } | |
248 | ||
653d07a6 | 249 | static inline struct rq *rq_of_rt_se(struct sched_rt_entity *rt_se) |
398a153b GH |
250 | { |
251 | struct task_struct *p = rt_task_of(rt_se); | |
653d07a6 KT |
252 | |
253 | return task_rq(p); | |
254 | } | |
255 | ||
256 | static inline struct rt_rq *rt_rq_of_se(struct sched_rt_entity *rt_se) | |
257 | { | |
258 | struct rq *rq = rq_of_rt_se(rt_se); | |
398a153b GH |
259 | |
260 | return &rq->rt; | |
261 | } | |
262 | ||
029632fb PZ |
263 | void free_rt_sched_group(struct task_group *tg) { } |
264 | ||
265 | int alloc_rt_sched_group(struct task_group *tg, struct task_group *parent) | |
266 | { | |
267 | return 1; | |
268 | } | |
398a153b GH |
269 | #endif /* CONFIG_RT_GROUP_SCHED */ |
270 | ||
4fd29176 | 271 | #ifdef CONFIG_SMP |
84de4274 | 272 | |
8046d680 | 273 | static void pull_rt_task(struct rq *this_rq); |
38033c37 | 274 | |
dc877341 PZ |
275 | static inline bool need_pull_rt_task(struct rq *rq, struct task_struct *prev) |
276 | { | |
277 | /* Try to pull RT tasks here if we lower this rq's prio */ | |
278 | return rq->rt.highest_prio.curr > prev->prio; | |
279 | } | |
280 | ||
637f5085 | 281 | static inline int rt_overloaded(struct rq *rq) |
4fd29176 | 282 | { |
637f5085 | 283 | return atomic_read(&rq->rd->rto_count); |
4fd29176 | 284 | } |
84de4274 | 285 | |
4fd29176 SR |
286 | static inline void rt_set_overload(struct rq *rq) |
287 | { | |
1f11eb6a GH |
288 | if (!rq->online) |
289 | return; | |
290 | ||
c6c4927b | 291 | cpumask_set_cpu(rq->cpu, rq->rd->rto_mask); |
4fd29176 SR |
292 | /* |
293 | * Make sure the mask is visible before we set | |
294 | * the overload count. That is checked to determine | |
295 | * if we should look at the mask. It would be a shame | |
296 | * if we looked at the mask, but the mask was not | |
297 | * updated yet. | |
7c3f2ab7 PZ |
298 | * |
299 | * Matched by the barrier in pull_rt_task(). | |
4fd29176 | 300 | */ |
7c3f2ab7 | 301 | smp_wmb(); |
637f5085 | 302 | atomic_inc(&rq->rd->rto_count); |
4fd29176 | 303 | } |
84de4274 | 304 | |
4fd29176 SR |
305 | static inline void rt_clear_overload(struct rq *rq) |
306 | { | |
1f11eb6a GH |
307 | if (!rq->online) |
308 | return; | |
309 | ||
4fd29176 | 310 | /* the order here really doesn't matter */ |
637f5085 | 311 | atomic_dec(&rq->rd->rto_count); |
c6c4927b | 312 | cpumask_clear_cpu(rq->cpu, rq->rd->rto_mask); |
4fd29176 | 313 | } |
73fe6aae | 314 | |
398a153b | 315 | static void update_rt_migration(struct rt_rq *rt_rq) |
73fe6aae | 316 | { |
a1ba4d8b | 317 | if (rt_rq->rt_nr_migratory && rt_rq->rt_nr_total > 1) { |
398a153b GH |
318 | if (!rt_rq->overloaded) { |
319 | rt_set_overload(rq_of_rt_rq(rt_rq)); | |
320 | rt_rq->overloaded = 1; | |
cdc8eb98 | 321 | } |
398a153b GH |
322 | } else if (rt_rq->overloaded) { |
323 | rt_clear_overload(rq_of_rt_rq(rt_rq)); | |
324 | rt_rq->overloaded = 0; | |
637f5085 | 325 | } |
73fe6aae | 326 | } |
4fd29176 | 327 | |
398a153b GH |
328 | static void inc_rt_migration(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq) |
329 | { | |
29baa747 PZ |
330 | struct task_struct *p; |
331 | ||
a1ba4d8b PZ |
332 | if (!rt_entity_is_task(rt_se)) |
333 | return; | |
334 | ||
29baa747 | 335 | p = rt_task_of(rt_se); |
a1ba4d8b PZ |
336 | rt_rq = &rq_of_rt_rq(rt_rq)->rt; |
337 | ||
338 | rt_rq->rt_nr_total++; | |
4b53a341 | 339 | if (p->nr_cpus_allowed > 1) |
398a153b GH |
340 | rt_rq->rt_nr_migratory++; |
341 | ||
342 | update_rt_migration(rt_rq); | |
343 | } | |
344 | ||
345 | static void dec_rt_migration(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq) | |
346 | { | |
29baa747 PZ |
347 | struct task_struct *p; |
348 | ||
a1ba4d8b PZ |
349 | if (!rt_entity_is_task(rt_se)) |
350 | return; | |
351 | ||
29baa747 | 352 | p = rt_task_of(rt_se); |
a1ba4d8b PZ |
353 | rt_rq = &rq_of_rt_rq(rt_rq)->rt; |
354 | ||
355 | rt_rq->rt_nr_total--; | |
4b53a341 | 356 | if (p->nr_cpus_allowed > 1) |
398a153b GH |
357 | rt_rq->rt_nr_migratory--; |
358 | ||
359 | update_rt_migration(rt_rq); | |
360 | } | |
361 | ||
5181f4a4 SR |
362 | static inline int has_pushable_tasks(struct rq *rq) |
363 | { | |
364 | return !plist_head_empty(&rq->rt.pushable_tasks); | |
365 | } | |
366 | ||
fd7a4bed PZ |
367 | static DEFINE_PER_CPU(struct callback_head, rt_push_head); |
368 | static DEFINE_PER_CPU(struct callback_head, rt_pull_head); | |
e3fca9e7 PZ |
369 | |
370 | static void push_rt_tasks(struct rq *); | |
fd7a4bed | 371 | static void pull_rt_task(struct rq *); |
e3fca9e7 PZ |
372 | |
373 | static inline void queue_push_tasks(struct rq *rq) | |
dc877341 | 374 | { |
e3fca9e7 PZ |
375 | if (!has_pushable_tasks(rq)) |
376 | return; | |
377 | ||
fd7a4bed PZ |
378 | queue_balance_callback(rq, &per_cpu(rt_push_head, rq->cpu), push_rt_tasks); |
379 | } | |
380 | ||
381 | static inline void queue_pull_task(struct rq *rq) | |
382 | { | |
383 | queue_balance_callback(rq, &per_cpu(rt_pull_head, rq->cpu), pull_rt_task); | |
dc877341 PZ |
384 | } |
385 | ||
917b627d GH |
386 | static void enqueue_pushable_task(struct rq *rq, struct task_struct *p) |
387 | { | |
388 | plist_del(&p->pushable_tasks, &rq->rt.pushable_tasks); | |
389 | plist_node_init(&p->pushable_tasks, p->prio); | |
390 | plist_add(&p->pushable_tasks, &rq->rt.pushable_tasks); | |
5181f4a4 SR |
391 | |
392 | /* Update the highest prio pushable task */ | |
393 | if (p->prio < rq->rt.highest_prio.next) | |
394 | rq->rt.highest_prio.next = p->prio; | |
917b627d GH |
395 | } |
396 | ||
397 | static void dequeue_pushable_task(struct rq *rq, struct task_struct *p) | |
398 | { | |
399 | plist_del(&p->pushable_tasks, &rq->rt.pushable_tasks); | |
917b627d | 400 | |
5181f4a4 SR |
401 | /* Update the new highest prio pushable task */ |
402 | if (has_pushable_tasks(rq)) { | |
403 | p = plist_first_entry(&rq->rt.pushable_tasks, | |
404 | struct task_struct, pushable_tasks); | |
405 | rq->rt.highest_prio.next = p->prio; | |
406 | } else | |
407 | rq->rt.highest_prio.next = MAX_RT_PRIO; | |
bcf08df3 IM |
408 | } |
409 | ||
917b627d GH |
410 | #else |
411 | ||
ceacc2c1 | 412 | static inline void enqueue_pushable_task(struct rq *rq, struct task_struct *p) |
fa85ae24 | 413 | { |
6f505b16 PZ |
414 | } |
415 | ||
ceacc2c1 PZ |
416 | static inline void dequeue_pushable_task(struct rq *rq, struct task_struct *p) |
417 | { | |
418 | } | |
419 | ||
b07430ac | 420 | static inline |
ceacc2c1 PZ |
421 | void inc_rt_migration(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq) |
422 | { | |
423 | } | |
424 | ||
398a153b | 425 | static inline |
ceacc2c1 PZ |
426 | void dec_rt_migration(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq) |
427 | { | |
428 | } | |
917b627d | 429 | |
dc877341 PZ |
430 | static inline bool need_pull_rt_task(struct rq *rq, struct task_struct *prev) |
431 | { | |
432 | return false; | |
433 | } | |
434 | ||
8046d680 | 435 | static inline void pull_rt_task(struct rq *this_rq) |
dc877341 | 436 | { |
dc877341 PZ |
437 | } |
438 | ||
e3fca9e7 | 439 | static inline void queue_push_tasks(struct rq *rq) |
dc877341 PZ |
440 | { |
441 | } | |
4fd29176 SR |
442 | #endif /* CONFIG_SMP */ |
443 | ||
f4ebcbc0 KT |
444 | static void enqueue_top_rt_rq(struct rt_rq *rt_rq); |
445 | static void dequeue_top_rt_rq(struct rt_rq *rt_rq); | |
446 | ||
6f505b16 PZ |
447 | static inline int on_rt_rq(struct sched_rt_entity *rt_se) |
448 | { | |
ff77e468 | 449 | return rt_se->on_rq; |
6f505b16 PZ |
450 | } |
451 | ||
052f1dc7 | 452 | #ifdef CONFIG_RT_GROUP_SCHED |
6f505b16 | 453 | |
9f0c1e56 | 454 | static inline u64 sched_rt_runtime(struct rt_rq *rt_rq) |
6f505b16 PZ |
455 | { |
456 | if (!rt_rq->tg) | |
9f0c1e56 | 457 | return RUNTIME_INF; |
6f505b16 | 458 | |
ac086bc2 PZ |
459 | return rt_rq->rt_runtime; |
460 | } | |
461 | ||
462 | static inline u64 sched_rt_period(struct rt_rq *rt_rq) | |
463 | { | |
464 | return ktime_to_ns(rt_rq->tg->rt_bandwidth.rt_period); | |
6f505b16 PZ |
465 | } |
466 | ||
ec514c48 CX |
467 | typedef struct task_group *rt_rq_iter_t; |
468 | ||
1c09ab0d YZ |
469 | static inline struct task_group *next_task_group(struct task_group *tg) |
470 | { | |
471 | do { | |
472 | tg = list_entry_rcu(tg->list.next, | |
473 | typeof(struct task_group), list); | |
474 | } while (&tg->list != &task_groups && task_group_is_autogroup(tg)); | |
475 | ||
476 | if (&tg->list == &task_groups) | |
477 | tg = NULL; | |
478 | ||
479 | return tg; | |
480 | } | |
481 | ||
482 | #define for_each_rt_rq(rt_rq, iter, rq) \ | |
483 | for (iter = container_of(&task_groups, typeof(*iter), list); \ | |
484 | (iter = next_task_group(iter)) && \ | |
485 | (rt_rq = iter->rt_rq[cpu_of(rq)]);) | |
ec514c48 | 486 | |
6f505b16 PZ |
487 | #define for_each_sched_rt_entity(rt_se) \ |
488 | for (; rt_se; rt_se = rt_se->parent) | |
489 | ||
490 | static inline struct rt_rq *group_rt_rq(struct sched_rt_entity *rt_se) | |
491 | { | |
492 | return rt_se->my_q; | |
493 | } | |
494 | ||
ff77e468 PZ |
495 | static void enqueue_rt_entity(struct sched_rt_entity *rt_se, unsigned int flags); |
496 | static void dequeue_rt_entity(struct sched_rt_entity *rt_se, unsigned int flags); | |
6f505b16 | 497 | |
9f0c1e56 | 498 | static void sched_rt_rq_enqueue(struct rt_rq *rt_rq) |
6f505b16 | 499 | { |
f6121f4f | 500 | struct task_struct *curr = rq_of_rt_rq(rt_rq)->curr; |
8875125e | 501 | struct rq *rq = rq_of_rt_rq(rt_rq); |
74b7eb58 YZ |
502 | struct sched_rt_entity *rt_se; |
503 | ||
8875125e | 504 | int cpu = cpu_of(rq); |
0c3b9168 BS |
505 | |
506 | rt_se = rt_rq->tg->rt_se[cpu]; | |
6f505b16 | 507 | |
f6121f4f | 508 | if (rt_rq->rt_nr_running) { |
f4ebcbc0 KT |
509 | if (!rt_se) |
510 | enqueue_top_rt_rq(rt_rq); | |
511 | else if (!on_rt_rq(rt_se)) | |
ff77e468 | 512 | enqueue_rt_entity(rt_se, 0); |
f4ebcbc0 | 513 | |
e864c499 | 514 | if (rt_rq->highest_prio.curr < curr->prio) |
8875125e | 515 | resched_curr(rq); |
6f505b16 PZ |
516 | } |
517 | } | |
518 | ||
9f0c1e56 | 519 | static void sched_rt_rq_dequeue(struct rt_rq *rt_rq) |
6f505b16 | 520 | { |
74b7eb58 | 521 | struct sched_rt_entity *rt_se; |
0c3b9168 | 522 | int cpu = cpu_of(rq_of_rt_rq(rt_rq)); |
74b7eb58 | 523 | |
0c3b9168 | 524 | rt_se = rt_rq->tg->rt_se[cpu]; |
6f505b16 | 525 | |
f4ebcbc0 KT |
526 | if (!rt_se) |
527 | dequeue_top_rt_rq(rt_rq); | |
528 | else if (on_rt_rq(rt_se)) | |
ff77e468 | 529 | dequeue_rt_entity(rt_se, 0); |
6f505b16 PZ |
530 | } |
531 | ||
46383648 KT |
532 | static inline int rt_rq_throttled(struct rt_rq *rt_rq) |
533 | { | |
534 | return rt_rq->rt_throttled && !rt_rq->rt_nr_boosted; | |
535 | } | |
536 | ||
23b0fdfc PZ |
537 | static int rt_se_boosted(struct sched_rt_entity *rt_se) |
538 | { | |
539 | struct rt_rq *rt_rq = group_rt_rq(rt_se); | |
540 | struct task_struct *p; | |
541 | ||
542 | if (rt_rq) | |
543 | return !!rt_rq->rt_nr_boosted; | |
544 | ||
545 | p = rt_task_of(rt_se); | |
546 | return p->prio != p->normal_prio; | |
547 | } | |
548 | ||
d0b27fa7 | 549 | #ifdef CONFIG_SMP |
c6c4927b | 550 | static inline const struct cpumask *sched_rt_period_mask(void) |
d0b27fa7 | 551 | { |
424c93fe | 552 | return this_rq()->rd->span; |
d0b27fa7 | 553 | } |
6f505b16 | 554 | #else |
c6c4927b | 555 | static inline const struct cpumask *sched_rt_period_mask(void) |
d0b27fa7 | 556 | { |
c6c4927b | 557 | return cpu_online_mask; |
d0b27fa7 PZ |
558 | } |
559 | #endif | |
6f505b16 | 560 | |
d0b27fa7 PZ |
561 | static inline |
562 | struct rt_rq *sched_rt_period_rt_rq(struct rt_bandwidth *rt_b, int cpu) | |
6f505b16 | 563 | { |
d0b27fa7 PZ |
564 | return container_of(rt_b, struct task_group, rt_bandwidth)->rt_rq[cpu]; |
565 | } | |
9f0c1e56 | 566 | |
ac086bc2 PZ |
567 | static inline struct rt_bandwidth *sched_rt_bandwidth(struct rt_rq *rt_rq) |
568 | { | |
569 | return &rt_rq->tg->rt_bandwidth; | |
570 | } | |
571 | ||
55e12e5e | 572 | #else /* !CONFIG_RT_GROUP_SCHED */ |
d0b27fa7 PZ |
573 | |
574 | static inline u64 sched_rt_runtime(struct rt_rq *rt_rq) | |
575 | { | |
ac086bc2 PZ |
576 | return rt_rq->rt_runtime; |
577 | } | |
578 | ||
579 | static inline u64 sched_rt_period(struct rt_rq *rt_rq) | |
580 | { | |
581 | return ktime_to_ns(def_rt_bandwidth.rt_period); | |
6f505b16 PZ |
582 | } |
583 | ||
ec514c48 CX |
584 | typedef struct rt_rq *rt_rq_iter_t; |
585 | ||
586 | #define for_each_rt_rq(rt_rq, iter, rq) \ | |
587 | for ((void) iter, rt_rq = &rq->rt; rt_rq; rt_rq = NULL) | |
588 | ||
6f505b16 PZ |
589 | #define for_each_sched_rt_entity(rt_se) \ |
590 | for (; rt_se; rt_se = NULL) | |
591 | ||
592 | static inline struct rt_rq *group_rt_rq(struct sched_rt_entity *rt_se) | |
593 | { | |
594 | return NULL; | |
595 | } | |
596 | ||
9f0c1e56 | 597 | static inline void sched_rt_rq_enqueue(struct rt_rq *rt_rq) |
6f505b16 | 598 | { |
f4ebcbc0 KT |
599 | struct rq *rq = rq_of_rt_rq(rt_rq); |
600 | ||
601 | if (!rt_rq->rt_nr_running) | |
602 | return; | |
603 | ||
604 | enqueue_top_rt_rq(rt_rq); | |
8875125e | 605 | resched_curr(rq); |
6f505b16 PZ |
606 | } |
607 | ||
9f0c1e56 | 608 | static inline void sched_rt_rq_dequeue(struct rt_rq *rt_rq) |
6f505b16 | 609 | { |
f4ebcbc0 | 610 | dequeue_top_rt_rq(rt_rq); |
6f505b16 PZ |
611 | } |
612 | ||
46383648 KT |
613 | static inline int rt_rq_throttled(struct rt_rq *rt_rq) |
614 | { | |
615 | return rt_rq->rt_throttled; | |
616 | } | |
617 | ||
c6c4927b | 618 | static inline const struct cpumask *sched_rt_period_mask(void) |
d0b27fa7 | 619 | { |
c6c4927b | 620 | return cpu_online_mask; |
d0b27fa7 PZ |
621 | } |
622 | ||
623 | static inline | |
624 | struct rt_rq *sched_rt_period_rt_rq(struct rt_bandwidth *rt_b, int cpu) | |
625 | { | |
626 | return &cpu_rq(cpu)->rt; | |
627 | } | |
628 | ||
ac086bc2 PZ |
629 | static inline struct rt_bandwidth *sched_rt_bandwidth(struct rt_rq *rt_rq) |
630 | { | |
631 | return &def_rt_bandwidth; | |
632 | } | |
633 | ||
55e12e5e | 634 | #endif /* CONFIG_RT_GROUP_SCHED */ |
d0b27fa7 | 635 | |
faa59937 JL |
636 | bool sched_rt_bandwidth_account(struct rt_rq *rt_rq) |
637 | { | |
638 | struct rt_bandwidth *rt_b = sched_rt_bandwidth(rt_rq); | |
639 | ||
640 | return (hrtimer_active(&rt_b->rt_period_timer) || | |
641 | rt_rq->rt_time < rt_b->rt_runtime); | |
642 | } | |
643 | ||
ac086bc2 | 644 | #ifdef CONFIG_SMP |
78333cdd PZ |
645 | /* |
646 | * We ran out of runtime, see if we can borrow some from our neighbours. | |
647 | */ | |
269b26a5 | 648 | static void do_balance_runtime(struct rt_rq *rt_rq) |
ac086bc2 PZ |
649 | { |
650 | struct rt_bandwidth *rt_b = sched_rt_bandwidth(rt_rq); | |
aa7f6730 | 651 | struct root_domain *rd = rq_of_rt_rq(rt_rq)->rd; |
269b26a5 | 652 | int i, weight; |
ac086bc2 PZ |
653 | u64 rt_period; |
654 | ||
c6c4927b | 655 | weight = cpumask_weight(rd->span); |
ac086bc2 | 656 | |
0986b11b | 657 | raw_spin_lock(&rt_b->rt_runtime_lock); |
ac086bc2 | 658 | rt_period = ktime_to_ns(rt_b->rt_period); |
c6c4927b | 659 | for_each_cpu(i, rd->span) { |
ac086bc2 PZ |
660 | struct rt_rq *iter = sched_rt_period_rt_rq(rt_b, i); |
661 | s64 diff; | |
662 | ||
663 | if (iter == rt_rq) | |
664 | continue; | |
665 | ||
0986b11b | 666 | raw_spin_lock(&iter->rt_runtime_lock); |
78333cdd PZ |
667 | /* |
668 | * Either all rqs have inf runtime and there's nothing to steal | |
669 | * or __disable_runtime() below sets a specific rq to inf to | |
670 | * indicate its been disabled and disalow stealing. | |
671 | */ | |
7def2be1 PZ |
672 | if (iter->rt_runtime == RUNTIME_INF) |
673 | goto next; | |
674 | ||
78333cdd PZ |
675 | /* |
676 | * From runqueues with spare time, take 1/n part of their | |
677 | * spare time, but no more than our period. | |
678 | */ | |
ac086bc2 PZ |
679 | diff = iter->rt_runtime - iter->rt_time; |
680 | if (diff > 0) { | |
58838cf3 | 681 | diff = div_u64((u64)diff, weight); |
ac086bc2 PZ |
682 | if (rt_rq->rt_runtime + diff > rt_period) |
683 | diff = rt_period - rt_rq->rt_runtime; | |
684 | iter->rt_runtime -= diff; | |
685 | rt_rq->rt_runtime += diff; | |
ac086bc2 | 686 | if (rt_rq->rt_runtime == rt_period) { |
0986b11b | 687 | raw_spin_unlock(&iter->rt_runtime_lock); |
ac086bc2 PZ |
688 | break; |
689 | } | |
690 | } | |
7def2be1 | 691 | next: |
0986b11b | 692 | raw_spin_unlock(&iter->rt_runtime_lock); |
ac086bc2 | 693 | } |
0986b11b | 694 | raw_spin_unlock(&rt_b->rt_runtime_lock); |
ac086bc2 | 695 | } |
7def2be1 | 696 | |
78333cdd PZ |
697 | /* |
698 | * Ensure this RQ takes back all the runtime it lend to its neighbours. | |
699 | */ | |
7def2be1 PZ |
700 | static void __disable_runtime(struct rq *rq) |
701 | { | |
702 | struct root_domain *rd = rq->rd; | |
ec514c48 | 703 | rt_rq_iter_t iter; |
7def2be1 PZ |
704 | struct rt_rq *rt_rq; |
705 | ||
706 | if (unlikely(!scheduler_running)) | |
707 | return; | |
708 | ||
ec514c48 | 709 | for_each_rt_rq(rt_rq, iter, rq) { |
7def2be1 PZ |
710 | struct rt_bandwidth *rt_b = sched_rt_bandwidth(rt_rq); |
711 | s64 want; | |
712 | int i; | |
713 | ||
0986b11b TG |
714 | raw_spin_lock(&rt_b->rt_runtime_lock); |
715 | raw_spin_lock(&rt_rq->rt_runtime_lock); | |
78333cdd PZ |
716 | /* |
717 | * Either we're all inf and nobody needs to borrow, or we're | |
718 | * already disabled and thus have nothing to do, or we have | |
719 | * exactly the right amount of runtime to take out. | |
720 | */ | |
7def2be1 PZ |
721 | if (rt_rq->rt_runtime == RUNTIME_INF || |
722 | rt_rq->rt_runtime == rt_b->rt_runtime) | |
723 | goto balanced; | |
0986b11b | 724 | raw_spin_unlock(&rt_rq->rt_runtime_lock); |
7def2be1 | 725 | |
78333cdd PZ |
726 | /* |
727 | * Calculate the difference between what we started out with | |
728 | * and what we current have, that's the amount of runtime | |
729 | * we lend and now have to reclaim. | |
730 | */ | |
7def2be1 PZ |
731 | want = rt_b->rt_runtime - rt_rq->rt_runtime; |
732 | ||
78333cdd PZ |
733 | /* |
734 | * Greedy reclaim, take back as much as we can. | |
735 | */ | |
c6c4927b | 736 | for_each_cpu(i, rd->span) { |
7def2be1 PZ |
737 | struct rt_rq *iter = sched_rt_period_rt_rq(rt_b, i); |
738 | s64 diff; | |
739 | ||
78333cdd PZ |
740 | /* |
741 | * Can't reclaim from ourselves or disabled runqueues. | |
742 | */ | |
f1679d08 | 743 | if (iter == rt_rq || iter->rt_runtime == RUNTIME_INF) |
7def2be1 PZ |
744 | continue; |
745 | ||
0986b11b | 746 | raw_spin_lock(&iter->rt_runtime_lock); |
7def2be1 PZ |
747 | if (want > 0) { |
748 | diff = min_t(s64, iter->rt_runtime, want); | |
749 | iter->rt_runtime -= diff; | |
750 | want -= diff; | |
751 | } else { | |
752 | iter->rt_runtime -= want; | |
753 | want -= want; | |
754 | } | |
0986b11b | 755 | raw_spin_unlock(&iter->rt_runtime_lock); |
7def2be1 PZ |
756 | |
757 | if (!want) | |
758 | break; | |
759 | } | |
760 | ||
0986b11b | 761 | raw_spin_lock(&rt_rq->rt_runtime_lock); |
78333cdd PZ |
762 | /* |
763 | * We cannot be left wanting - that would mean some runtime | |
764 | * leaked out of the system. | |
765 | */ | |
7def2be1 PZ |
766 | BUG_ON(want); |
767 | balanced: | |
78333cdd PZ |
768 | /* |
769 | * Disable all the borrow logic by pretending we have inf | |
770 | * runtime - in which case borrowing doesn't make sense. | |
771 | */ | |
7def2be1 | 772 | rt_rq->rt_runtime = RUNTIME_INF; |
a4c96ae3 | 773 | rt_rq->rt_throttled = 0; |
0986b11b TG |
774 | raw_spin_unlock(&rt_rq->rt_runtime_lock); |
775 | raw_spin_unlock(&rt_b->rt_runtime_lock); | |
99b62567 KT |
776 | |
777 | /* Make rt_rq available for pick_next_task() */ | |
778 | sched_rt_rq_enqueue(rt_rq); | |
7def2be1 PZ |
779 | } |
780 | } | |
781 | ||
7def2be1 PZ |
782 | static void __enable_runtime(struct rq *rq) |
783 | { | |
ec514c48 | 784 | rt_rq_iter_t iter; |
7def2be1 PZ |
785 | struct rt_rq *rt_rq; |
786 | ||
787 | if (unlikely(!scheduler_running)) | |
788 | return; | |
789 | ||
78333cdd PZ |
790 | /* |
791 | * Reset each runqueue's bandwidth settings | |
792 | */ | |
ec514c48 | 793 | for_each_rt_rq(rt_rq, iter, rq) { |
7def2be1 PZ |
794 | struct rt_bandwidth *rt_b = sched_rt_bandwidth(rt_rq); |
795 | ||
0986b11b TG |
796 | raw_spin_lock(&rt_b->rt_runtime_lock); |
797 | raw_spin_lock(&rt_rq->rt_runtime_lock); | |
7def2be1 PZ |
798 | rt_rq->rt_runtime = rt_b->rt_runtime; |
799 | rt_rq->rt_time = 0; | |
baf25731 | 800 | rt_rq->rt_throttled = 0; |
0986b11b TG |
801 | raw_spin_unlock(&rt_rq->rt_runtime_lock); |
802 | raw_spin_unlock(&rt_b->rt_runtime_lock); | |
7def2be1 PZ |
803 | } |
804 | } | |
805 | ||
269b26a5 | 806 | static void balance_runtime(struct rt_rq *rt_rq) |
eff6549b | 807 | { |
4a6184ce | 808 | if (!sched_feat(RT_RUNTIME_SHARE)) |
269b26a5 | 809 | return; |
4a6184ce | 810 | |
eff6549b | 811 | if (rt_rq->rt_time > rt_rq->rt_runtime) { |
0986b11b | 812 | raw_spin_unlock(&rt_rq->rt_runtime_lock); |
269b26a5 | 813 | do_balance_runtime(rt_rq); |
0986b11b | 814 | raw_spin_lock(&rt_rq->rt_runtime_lock); |
eff6549b | 815 | } |
eff6549b | 816 | } |
55e12e5e | 817 | #else /* !CONFIG_SMP */ |
269b26a5 | 818 | static inline void balance_runtime(struct rt_rq *rt_rq) {} |
55e12e5e | 819 | #endif /* CONFIG_SMP */ |
ac086bc2 | 820 | |
eff6549b PZ |
821 | static int do_sched_rt_period_timer(struct rt_bandwidth *rt_b, int overrun) |
822 | { | |
42c62a58 | 823 | int i, idle = 1, throttled = 0; |
c6c4927b | 824 | const struct cpumask *span; |
eff6549b | 825 | |
eff6549b | 826 | span = sched_rt_period_mask(); |
e221d028 MG |
827 | #ifdef CONFIG_RT_GROUP_SCHED |
828 | /* | |
829 | * FIXME: isolated CPUs should really leave the root task group, | |
830 | * whether they are isolcpus or were isolated via cpusets, lest | |
831 | * the timer run on a CPU which does not service all runqueues, | |
832 | * potentially leaving other CPUs indefinitely throttled. If | |
833 | * isolation is really required, the user will turn the throttle | |
834 | * off to kill the perturbations it causes anyway. Meanwhile, | |
835 | * this maintains functionality for boot and/or troubleshooting. | |
836 | */ | |
837 | if (rt_b == &root_task_group.rt_bandwidth) | |
838 | span = cpu_online_mask; | |
839 | #endif | |
c6c4927b | 840 | for_each_cpu(i, span) { |
eff6549b PZ |
841 | int enqueue = 0; |
842 | struct rt_rq *rt_rq = sched_rt_period_rt_rq(rt_b, i); | |
843 | struct rq *rq = rq_of_rt_rq(rt_rq); | |
c249f255 DK |
844 | int skip; |
845 | ||
846 | /* | |
847 | * When span == cpu_online_mask, taking each rq->lock | |
848 | * can be time-consuming. Try to avoid it when possible. | |
849 | */ | |
850 | raw_spin_lock(&rt_rq->rt_runtime_lock); | |
851 | skip = !rt_rq->rt_time && !rt_rq->rt_nr_running; | |
852 | raw_spin_unlock(&rt_rq->rt_runtime_lock); | |
853 | if (skip) | |
854 | continue; | |
eff6549b | 855 | |
05fa785c | 856 | raw_spin_lock(&rq->lock); |
eff6549b PZ |
857 | if (rt_rq->rt_time) { |
858 | u64 runtime; | |
859 | ||
0986b11b | 860 | raw_spin_lock(&rt_rq->rt_runtime_lock); |
eff6549b PZ |
861 | if (rt_rq->rt_throttled) |
862 | balance_runtime(rt_rq); | |
863 | runtime = rt_rq->rt_runtime; | |
864 | rt_rq->rt_time -= min(rt_rq->rt_time, overrun*runtime); | |
865 | if (rt_rq->rt_throttled && rt_rq->rt_time < runtime) { | |
866 | rt_rq->rt_throttled = 0; | |
867 | enqueue = 1; | |
61eadef6 MG |
868 | |
869 | /* | |
9edfbfed PZ |
870 | * When we're idle and a woken (rt) task is |
871 | * throttled check_preempt_curr() will set | |
872 | * skip_update and the time between the wakeup | |
873 | * and this unthrottle will get accounted as | |
874 | * 'runtime'. | |
61eadef6 MG |
875 | */ |
876 | if (rt_rq->rt_nr_running && rq->curr == rq->idle) | |
9edfbfed | 877 | rq_clock_skip_update(rq, false); |
eff6549b PZ |
878 | } |
879 | if (rt_rq->rt_time || rt_rq->rt_nr_running) | |
880 | idle = 0; | |
0986b11b | 881 | raw_spin_unlock(&rt_rq->rt_runtime_lock); |
0c3b9168 | 882 | } else if (rt_rq->rt_nr_running) { |
6c3df255 | 883 | idle = 0; |
0c3b9168 BS |
884 | if (!rt_rq_throttled(rt_rq)) |
885 | enqueue = 1; | |
886 | } | |
42c62a58 PZ |
887 | if (rt_rq->rt_throttled) |
888 | throttled = 1; | |
eff6549b PZ |
889 | |
890 | if (enqueue) | |
891 | sched_rt_rq_enqueue(rt_rq); | |
05fa785c | 892 | raw_spin_unlock(&rq->lock); |
eff6549b PZ |
893 | } |
894 | ||
42c62a58 PZ |
895 | if (!throttled && (!rt_bandwidth_enabled() || rt_b->rt_runtime == RUNTIME_INF)) |
896 | return 1; | |
897 | ||
eff6549b PZ |
898 | return idle; |
899 | } | |
ac086bc2 | 900 | |
6f505b16 PZ |
901 | static inline int rt_se_prio(struct sched_rt_entity *rt_se) |
902 | { | |
052f1dc7 | 903 | #ifdef CONFIG_RT_GROUP_SCHED |
6f505b16 PZ |
904 | struct rt_rq *rt_rq = group_rt_rq(rt_se); |
905 | ||
906 | if (rt_rq) | |
e864c499 | 907 | return rt_rq->highest_prio.curr; |
6f505b16 PZ |
908 | #endif |
909 | ||
910 | return rt_task_of(rt_se)->prio; | |
911 | } | |
912 | ||
9f0c1e56 | 913 | static int sched_rt_runtime_exceeded(struct rt_rq *rt_rq) |
6f505b16 | 914 | { |
9f0c1e56 | 915 | u64 runtime = sched_rt_runtime(rt_rq); |
fa85ae24 | 916 | |
fa85ae24 | 917 | if (rt_rq->rt_throttled) |
23b0fdfc | 918 | return rt_rq_throttled(rt_rq); |
fa85ae24 | 919 | |
5b680fd6 | 920 | if (runtime >= sched_rt_period(rt_rq)) |
ac086bc2 PZ |
921 | return 0; |
922 | ||
b79f3833 PZ |
923 | balance_runtime(rt_rq); |
924 | runtime = sched_rt_runtime(rt_rq); | |
925 | if (runtime == RUNTIME_INF) | |
926 | return 0; | |
ac086bc2 | 927 | |
9f0c1e56 | 928 | if (rt_rq->rt_time > runtime) { |
7abc63b1 PZ |
929 | struct rt_bandwidth *rt_b = sched_rt_bandwidth(rt_rq); |
930 | ||
931 | /* | |
932 | * Don't actually throttle groups that have no runtime assigned | |
933 | * but accrue some time due to boosting. | |
934 | */ | |
935 | if (likely(rt_b->rt_runtime)) { | |
936 | rt_rq->rt_throttled = 1; | |
c224815d | 937 | printk_deferred_once("sched: RT throttling activated\n"); |
7abc63b1 PZ |
938 | } else { |
939 | /* | |
940 | * In case we did anyway, make it go away, | |
941 | * replenishment is a joke, since it will replenish us | |
942 | * with exactly 0 ns. | |
943 | */ | |
944 | rt_rq->rt_time = 0; | |
945 | } | |
946 | ||
23b0fdfc | 947 | if (rt_rq_throttled(rt_rq)) { |
9f0c1e56 | 948 | sched_rt_rq_dequeue(rt_rq); |
23b0fdfc PZ |
949 | return 1; |
950 | } | |
fa85ae24 PZ |
951 | } |
952 | ||
953 | return 0; | |
954 | } | |
955 | ||
bb44e5d1 IM |
956 | /* |
957 | * Update the current task's runtime statistics. Skip current tasks that | |
958 | * are not in our scheduling class. | |
959 | */ | |
a9957449 | 960 | static void update_curr_rt(struct rq *rq) |
bb44e5d1 IM |
961 | { |
962 | struct task_struct *curr = rq->curr; | |
6f505b16 | 963 | struct sched_rt_entity *rt_se = &curr->rt; |
bb44e5d1 IM |
964 | u64 delta_exec; |
965 | ||
06c3bc65 | 966 | if (curr->sched_class != &rt_sched_class) |
bb44e5d1 IM |
967 | return; |
968 | ||
78becc27 | 969 | delta_exec = rq_clock_task(rq) - curr->se.exec_start; |
fc79e240 KT |
970 | if (unlikely((s64)delta_exec <= 0)) |
971 | return; | |
6cfb0d5d | 972 | |
58919e83 | 973 | /* Kick cpufreq (see the comment in kernel/sched/sched.h). */ |
674e7541 | 974 | cpufreq_update_util(rq, SCHED_CPUFREQ_RT); |
594dd290 | 975 | |
42c62a58 PZ |
976 | schedstat_set(curr->se.statistics.exec_max, |
977 | max(curr->se.statistics.exec_max, delta_exec)); | |
bb44e5d1 IM |
978 | |
979 | curr->se.sum_exec_runtime += delta_exec; | |
f06febc9 FM |
980 | account_group_exec_runtime(curr, delta_exec); |
981 | ||
78becc27 | 982 | curr->se.exec_start = rq_clock_task(rq); |
d842de87 | 983 | cpuacct_charge(curr, delta_exec); |
fa85ae24 | 984 | |
e9e9250b PZ |
985 | sched_rt_avg_update(rq, delta_exec); |
986 | ||
0b148fa0 PZ |
987 | if (!rt_bandwidth_enabled()) |
988 | return; | |
989 | ||
354d60c2 | 990 | for_each_sched_rt_entity(rt_se) { |
0b07939c | 991 | struct rt_rq *rt_rq = rt_rq_of_se(rt_se); |
354d60c2 | 992 | |
cc2991cf | 993 | if (sched_rt_runtime(rt_rq) != RUNTIME_INF) { |
0986b11b | 994 | raw_spin_lock(&rt_rq->rt_runtime_lock); |
cc2991cf PZ |
995 | rt_rq->rt_time += delta_exec; |
996 | if (sched_rt_runtime_exceeded(rt_rq)) | |
8875125e | 997 | resched_curr(rq); |
0986b11b | 998 | raw_spin_unlock(&rt_rq->rt_runtime_lock); |
cc2991cf | 999 | } |
354d60c2 | 1000 | } |
bb44e5d1 IM |
1001 | } |
1002 | ||
f4ebcbc0 KT |
1003 | static void |
1004 | dequeue_top_rt_rq(struct rt_rq *rt_rq) | |
1005 | { | |
1006 | struct rq *rq = rq_of_rt_rq(rt_rq); | |
1007 | ||
1008 | BUG_ON(&rq->rt != rt_rq); | |
1009 | ||
1010 | if (!rt_rq->rt_queued) | |
1011 | return; | |
1012 | ||
1013 | BUG_ON(!rq->nr_running); | |
1014 | ||
72465447 | 1015 | sub_nr_running(rq, rt_rq->rt_nr_running); |
f4ebcbc0 KT |
1016 | rt_rq->rt_queued = 0; |
1017 | } | |
1018 | ||
1019 | static void | |
1020 | enqueue_top_rt_rq(struct rt_rq *rt_rq) | |
1021 | { | |
1022 | struct rq *rq = rq_of_rt_rq(rt_rq); | |
1023 | ||
1024 | BUG_ON(&rq->rt != rt_rq); | |
1025 | ||
1026 | if (rt_rq->rt_queued) | |
1027 | return; | |
1028 | if (rt_rq_throttled(rt_rq) || !rt_rq->rt_nr_running) | |
1029 | return; | |
1030 | ||
72465447 | 1031 | add_nr_running(rq, rt_rq->rt_nr_running); |
f4ebcbc0 KT |
1032 | rt_rq->rt_queued = 1; |
1033 | } | |
1034 | ||
398a153b | 1035 | #if defined CONFIG_SMP |
e864c499 | 1036 | |
398a153b GH |
1037 | static void |
1038 | inc_rt_prio_smp(struct rt_rq *rt_rq, int prio, int prev_prio) | |
63489e45 | 1039 | { |
4d984277 | 1040 | struct rq *rq = rq_of_rt_rq(rt_rq); |
1f11eb6a | 1041 | |
757dfcaa KT |
1042 | #ifdef CONFIG_RT_GROUP_SCHED |
1043 | /* | |
1044 | * Change rq's cpupri only if rt_rq is the top queue. | |
1045 | */ | |
1046 | if (&rq->rt != rt_rq) | |
1047 | return; | |
1048 | #endif | |
5181f4a4 SR |
1049 | if (rq->online && prio < prev_prio) |
1050 | cpupri_set(&rq->rd->cpupri, rq->cpu, prio); | |
398a153b | 1051 | } |
73fe6aae | 1052 | |
398a153b GH |
1053 | static void |
1054 | dec_rt_prio_smp(struct rt_rq *rt_rq, int prio, int prev_prio) | |
1055 | { | |
1056 | struct rq *rq = rq_of_rt_rq(rt_rq); | |
d0b27fa7 | 1057 | |
757dfcaa KT |
1058 | #ifdef CONFIG_RT_GROUP_SCHED |
1059 | /* | |
1060 | * Change rq's cpupri only if rt_rq is the top queue. | |
1061 | */ | |
1062 | if (&rq->rt != rt_rq) | |
1063 | return; | |
1064 | #endif | |
398a153b GH |
1065 | if (rq->online && rt_rq->highest_prio.curr != prev_prio) |
1066 | cpupri_set(&rq->rd->cpupri, rq->cpu, rt_rq->highest_prio.curr); | |
63489e45 SR |
1067 | } |
1068 | ||
398a153b GH |
1069 | #else /* CONFIG_SMP */ |
1070 | ||
6f505b16 | 1071 | static inline |
398a153b GH |
1072 | void inc_rt_prio_smp(struct rt_rq *rt_rq, int prio, int prev_prio) {} |
1073 | static inline | |
1074 | void dec_rt_prio_smp(struct rt_rq *rt_rq, int prio, int prev_prio) {} | |
1075 | ||
1076 | #endif /* CONFIG_SMP */ | |
6e0534f2 | 1077 | |
052f1dc7 | 1078 | #if defined CONFIG_SMP || defined CONFIG_RT_GROUP_SCHED |
398a153b GH |
1079 | static void |
1080 | inc_rt_prio(struct rt_rq *rt_rq, int prio) | |
1081 | { | |
1082 | int prev_prio = rt_rq->highest_prio.curr; | |
1083 | ||
1084 | if (prio < prev_prio) | |
1085 | rt_rq->highest_prio.curr = prio; | |
1086 | ||
1087 | inc_rt_prio_smp(rt_rq, prio, prev_prio); | |
1088 | } | |
1089 | ||
1090 | static void | |
1091 | dec_rt_prio(struct rt_rq *rt_rq, int prio) | |
1092 | { | |
1093 | int prev_prio = rt_rq->highest_prio.curr; | |
1094 | ||
6f505b16 | 1095 | if (rt_rq->rt_nr_running) { |
764a9d6f | 1096 | |
398a153b | 1097 | WARN_ON(prio < prev_prio); |
764a9d6f | 1098 | |
e864c499 | 1099 | /* |
398a153b GH |
1100 | * This may have been our highest task, and therefore |
1101 | * we may have some recomputation to do | |
e864c499 | 1102 | */ |
398a153b | 1103 | if (prio == prev_prio) { |
e864c499 GH |
1104 | struct rt_prio_array *array = &rt_rq->active; |
1105 | ||
1106 | rt_rq->highest_prio.curr = | |
764a9d6f | 1107 | sched_find_first_bit(array->bitmap); |
e864c499 GH |
1108 | } |
1109 | ||
764a9d6f | 1110 | } else |
e864c499 | 1111 | rt_rq->highest_prio.curr = MAX_RT_PRIO; |
73fe6aae | 1112 | |
398a153b GH |
1113 | dec_rt_prio_smp(rt_rq, prio, prev_prio); |
1114 | } | |
1f11eb6a | 1115 | |
398a153b GH |
1116 | #else |
1117 | ||
1118 | static inline void inc_rt_prio(struct rt_rq *rt_rq, int prio) {} | |
1119 | static inline void dec_rt_prio(struct rt_rq *rt_rq, int prio) {} | |
1120 | ||
1121 | #endif /* CONFIG_SMP || CONFIG_RT_GROUP_SCHED */ | |
6e0534f2 | 1122 | |
052f1dc7 | 1123 | #ifdef CONFIG_RT_GROUP_SCHED |
398a153b GH |
1124 | |
1125 | static void | |
1126 | inc_rt_group(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq) | |
1127 | { | |
1128 | if (rt_se_boosted(rt_se)) | |
1129 | rt_rq->rt_nr_boosted++; | |
1130 | ||
1131 | if (rt_rq->tg) | |
1132 | start_rt_bandwidth(&rt_rq->tg->rt_bandwidth); | |
1133 | } | |
1134 | ||
1135 | static void | |
1136 | dec_rt_group(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq) | |
1137 | { | |
23b0fdfc PZ |
1138 | if (rt_se_boosted(rt_se)) |
1139 | rt_rq->rt_nr_boosted--; | |
1140 | ||
1141 | WARN_ON(!rt_rq->rt_nr_running && rt_rq->rt_nr_boosted); | |
398a153b GH |
1142 | } |
1143 | ||
1144 | #else /* CONFIG_RT_GROUP_SCHED */ | |
1145 | ||
1146 | static void | |
1147 | inc_rt_group(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq) | |
1148 | { | |
1149 | start_rt_bandwidth(&def_rt_bandwidth); | |
1150 | } | |
1151 | ||
1152 | static inline | |
1153 | void dec_rt_group(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq) {} | |
1154 | ||
1155 | #endif /* CONFIG_RT_GROUP_SCHED */ | |
1156 | ||
22abdef3 KT |
1157 | static inline |
1158 | unsigned int rt_se_nr_running(struct sched_rt_entity *rt_se) | |
1159 | { | |
1160 | struct rt_rq *group_rq = group_rt_rq(rt_se); | |
1161 | ||
1162 | if (group_rq) | |
1163 | return group_rq->rt_nr_running; | |
1164 | else | |
1165 | return 1; | |
1166 | } | |
1167 | ||
01d36d0a FW |
1168 | static inline |
1169 | unsigned int rt_se_rr_nr_running(struct sched_rt_entity *rt_se) | |
1170 | { | |
1171 | struct rt_rq *group_rq = group_rt_rq(rt_se); | |
1172 | struct task_struct *tsk; | |
1173 | ||
1174 | if (group_rq) | |
1175 | return group_rq->rr_nr_running; | |
1176 | ||
1177 | tsk = rt_task_of(rt_se); | |
1178 | ||
1179 | return (tsk->policy == SCHED_RR) ? 1 : 0; | |
1180 | } | |
1181 | ||
398a153b GH |
1182 | static inline |
1183 | void inc_rt_tasks(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq) | |
1184 | { | |
1185 | int prio = rt_se_prio(rt_se); | |
1186 | ||
1187 | WARN_ON(!rt_prio(prio)); | |
22abdef3 | 1188 | rt_rq->rt_nr_running += rt_se_nr_running(rt_se); |
01d36d0a | 1189 | rt_rq->rr_nr_running += rt_se_rr_nr_running(rt_se); |
398a153b GH |
1190 | |
1191 | inc_rt_prio(rt_rq, prio); | |
1192 | inc_rt_migration(rt_se, rt_rq); | |
1193 | inc_rt_group(rt_se, rt_rq); | |
1194 | } | |
1195 | ||
1196 | static inline | |
1197 | void dec_rt_tasks(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq) | |
1198 | { | |
1199 | WARN_ON(!rt_prio(rt_se_prio(rt_se))); | |
1200 | WARN_ON(!rt_rq->rt_nr_running); | |
22abdef3 | 1201 | rt_rq->rt_nr_running -= rt_se_nr_running(rt_se); |
01d36d0a | 1202 | rt_rq->rr_nr_running -= rt_se_rr_nr_running(rt_se); |
398a153b GH |
1203 | |
1204 | dec_rt_prio(rt_rq, rt_se_prio(rt_se)); | |
1205 | dec_rt_migration(rt_se, rt_rq); | |
1206 | dec_rt_group(rt_se, rt_rq); | |
63489e45 SR |
1207 | } |
1208 | ||
ff77e468 PZ |
1209 | /* |
1210 | * Change rt_se->run_list location unless SAVE && !MOVE | |
1211 | * | |
1212 | * assumes ENQUEUE/DEQUEUE flags match | |
1213 | */ | |
1214 | static inline bool move_entity(unsigned int flags) | |
1215 | { | |
1216 | if ((flags & (DEQUEUE_SAVE | DEQUEUE_MOVE)) == DEQUEUE_SAVE) | |
1217 | return false; | |
1218 | ||
1219 | return true; | |
1220 | } | |
1221 | ||
1222 | static void __delist_rt_entity(struct sched_rt_entity *rt_se, struct rt_prio_array *array) | |
1223 | { | |
1224 | list_del_init(&rt_se->run_list); | |
1225 | ||
1226 | if (list_empty(array->queue + rt_se_prio(rt_se))) | |
1227 | __clear_bit(rt_se_prio(rt_se), array->bitmap); | |
1228 | ||
1229 | rt_se->on_list = 0; | |
1230 | } | |
1231 | ||
1232 | static void __enqueue_rt_entity(struct sched_rt_entity *rt_se, unsigned int flags) | |
bb44e5d1 | 1233 | { |
6f505b16 PZ |
1234 | struct rt_rq *rt_rq = rt_rq_of_se(rt_se); |
1235 | struct rt_prio_array *array = &rt_rq->active; | |
1236 | struct rt_rq *group_rq = group_rt_rq(rt_se); | |
20b6331b | 1237 | struct list_head *queue = array->queue + rt_se_prio(rt_se); |
bb44e5d1 | 1238 | |
ad2a3f13 PZ |
1239 | /* |
1240 | * Don't enqueue the group if its throttled, or when empty. | |
1241 | * The latter is a consequence of the former when a child group | |
1242 | * get throttled and the current group doesn't have any other | |
1243 | * active members. | |
1244 | */ | |
ff77e468 PZ |
1245 | if (group_rq && (rt_rq_throttled(group_rq) || !group_rq->rt_nr_running)) { |
1246 | if (rt_se->on_list) | |
1247 | __delist_rt_entity(rt_se, array); | |
6f505b16 | 1248 | return; |
ff77e468 | 1249 | } |
63489e45 | 1250 | |
ff77e468 PZ |
1251 | if (move_entity(flags)) { |
1252 | WARN_ON_ONCE(rt_se->on_list); | |
1253 | if (flags & ENQUEUE_HEAD) | |
1254 | list_add(&rt_se->run_list, queue); | |
1255 | else | |
1256 | list_add_tail(&rt_se->run_list, queue); | |
1257 | ||
1258 | __set_bit(rt_se_prio(rt_se), array->bitmap); | |
1259 | rt_se->on_list = 1; | |
1260 | } | |
1261 | rt_se->on_rq = 1; | |
78f2c7db | 1262 | |
6f505b16 PZ |
1263 | inc_rt_tasks(rt_se, rt_rq); |
1264 | } | |
1265 | ||
ff77e468 | 1266 | static void __dequeue_rt_entity(struct sched_rt_entity *rt_se, unsigned int flags) |
6f505b16 PZ |
1267 | { |
1268 | struct rt_rq *rt_rq = rt_rq_of_se(rt_se); | |
1269 | struct rt_prio_array *array = &rt_rq->active; | |
1270 | ||
ff77e468 PZ |
1271 | if (move_entity(flags)) { |
1272 | WARN_ON_ONCE(!rt_se->on_list); | |
1273 | __delist_rt_entity(rt_se, array); | |
1274 | } | |
1275 | rt_se->on_rq = 0; | |
6f505b16 PZ |
1276 | |
1277 | dec_rt_tasks(rt_se, rt_rq); | |
1278 | } | |
1279 | ||
1280 | /* | |
1281 | * Because the prio of an upper entry depends on the lower | |
1282 | * entries, we must remove entries top - down. | |
6f505b16 | 1283 | */ |
ff77e468 | 1284 | static void dequeue_rt_stack(struct sched_rt_entity *rt_se, unsigned int flags) |
6f505b16 | 1285 | { |
ad2a3f13 | 1286 | struct sched_rt_entity *back = NULL; |
6f505b16 | 1287 | |
58d6c2d7 PZ |
1288 | for_each_sched_rt_entity(rt_se) { |
1289 | rt_se->back = back; | |
1290 | back = rt_se; | |
1291 | } | |
1292 | ||
f4ebcbc0 KT |
1293 | dequeue_top_rt_rq(rt_rq_of_se(back)); |
1294 | ||
58d6c2d7 PZ |
1295 | for (rt_se = back; rt_se; rt_se = rt_se->back) { |
1296 | if (on_rt_rq(rt_se)) | |
ff77e468 | 1297 | __dequeue_rt_entity(rt_se, flags); |
ad2a3f13 PZ |
1298 | } |
1299 | } | |
1300 | ||
ff77e468 | 1301 | static void enqueue_rt_entity(struct sched_rt_entity *rt_se, unsigned int flags) |
ad2a3f13 | 1302 | { |
f4ebcbc0 KT |
1303 | struct rq *rq = rq_of_rt_se(rt_se); |
1304 | ||
ff77e468 | 1305 | dequeue_rt_stack(rt_se, flags); |
ad2a3f13 | 1306 | for_each_sched_rt_entity(rt_se) |
ff77e468 | 1307 | __enqueue_rt_entity(rt_se, flags); |
f4ebcbc0 | 1308 | enqueue_top_rt_rq(&rq->rt); |
ad2a3f13 PZ |
1309 | } |
1310 | ||
ff77e468 | 1311 | static void dequeue_rt_entity(struct sched_rt_entity *rt_se, unsigned int flags) |
ad2a3f13 | 1312 | { |
f4ebcbc0 KT |
1313 | struct rq *rq = rq_of_rt_se(rt_se); |
1314 | ||
ff77e468 | 1315 | dequeue_rt_stack(rt_se, flags); |
ad2a3f13 PZ |
1316 | |
1317 | for_each_sched_rt_entity(rt_se) { | |
1318 | struct rt_rq *rt_rq = group_rt_rq(rt_se); | |
1319 | ||
1320 | if (rt_rq && rt_rq->rt_nr_running) | |
ff77e468 | 1321 | __enqueue_rt_entity(rt_se, flags); |
58d6c2d7 | 1322 | } |
f4ebcbc0 | 1323 | enqueue_top_rt_rq(&rq->rt); |
bb44e5d1 IM |
1324 | } |
1325 | ||
1326 | /* | |
1327 | * Adding/removing a task to/from a priority array: | |
1328 | */ | |
ea87bb78 | 1329 | static void |
371fd7e7 | 1330 | enqueue_task_rt(struct rq *rq, struct task_struct *p, int flags) |
6f505b16 PZ |
1331 | { |
1332 | struct sched_rt_entity *rt_se = &p->rt; | |
1333 | ||
371fd7e7 | 1334 | if (flags & ENQUEUE_WAKEUP) |
6f505b16 PZ |
1335 | rt_se->timeout = 0; |
1336 | ||
ff77e468 | 1337 | enqueue_rt_entity(rt_se, flags); |
c09595f6 | 1338 | |
4b53a341 | 1339 | if (!task_current(rq, p) && p->nr_cpus_allowed > 1) |
917b627d | 1340 | enqueue_pushable_task(rq, p); |
6f505b16 PZ |
1341 | } |
1342 | ||
371fd7e7 | 1343 | static void dequeue_task_rt(struct rq *rq, struct task_struct *p, int flags) |
bb44e5d1 | 1344 | { |
6f505b16 | 1345 | struct sched_rt_entity *rt_se = &p->rt; |
bb44e5d1 | 1346 | |
f1e14ef6 | 1347 | update_curr_rt(rq); |
ff77e468 | 1348 | dequeue_rt_entity(rt_se, flags); |
c09595f6 | 1349 | |
917b627d | 1350 | dequeue_pushable_task(rq, p); |
bb44e5d1 IM |
1351 | } |
1352 | ||
1353 | /* | |
60686317 RW |
1354 | * Put task to the head or the end of the run list without the overhead of |
1355 | * dequeue followed by enqueue. | |
bb44e5d1 | 1356 | */ |
7ebefa8c DA |
1357 | static void |
1358 | requeue_rt_entity(struct rt_rq *rt_rq, struct sched_rt_entity *rt_se, int head) | |
6f505b16 | 1359 | { |
1cdad715 | 1360 | if (on_rt_rq(rt_se)) { |
7ebefa8c DA |
1361 | struct rt_prio_array *array = &rt_rq->active; |
1362 | struct list_head *queue = array->queue + rt_se_prio(rt_se); | |
1363 | ||
1364 | if (head) | |
1365 | list_move(&rt_se->run_list, queue); | |
1366 | else | |
1367 | list_move_tail(&rt_se->run_list, queue); | |
1cdad715 | 1368 | } |
6f505b16 PZ |
1369 | } |
1370 | ||
7ebefa8c | 1371 | static void requeue_task_rt(struct rq *rq, struct task_struct *p, int head) |
bb44e5d1 | 1372 | { |
6f505b16 PZ |
1373 | struct sched_rt_entity *rt_se = &p->rt; |
1374 | struct rt_rq *rt_rq; | |
bb44e5d1 | 1375 | |
6f505b16 PZ |
1376 | for_each_sched_rt_entity(rt_se) { |
1377 | rt_rq = rt_rq_of_se(rt_se); | |
7ebefa8c | 1378 | requeue_rt_entity(rt_rq, rt_se, head); |
6f505b16 | 1379 | } |
bb44e5d1 IM |
1380 | } |
1381 | ||
6f505b16 | 1382 | static void yield_task_rt(struct rq *rq) |
bb44e5d1 | 1383 | { |
7ebefa8c | 1384 | requeue_task_rt(rq, rq->curr, 0); |
bb44e5d1 IM |
1385 | } |
1386 | ||
e7693a36 | 1387 | #ifdef CONFIG_SMP |
318e0893 GH |
1388 | static int find_lowest_rq(struct task_struct *task); |
1389 | ||
0017d735 | 1390 | static int |
ac66f547 | 1391 | select_task_rq_rt(struct task_struct *p, int cpu, int sd_flag, int flags) |
e7693a36 | 1392 | { |
7608dec2 PZ |
1393 | struct task_struct *curr; |
1394 | struct rq *rq; | |
c37495fd SR |
1395 | |
1396 | /* For anything but wake ups, just return the task_cpu */ | |
1397 | if (sd_flag != SD_BALANCE_WAKE && sd_flag != SD_BALANCE_FORK) | |
1398 | goto out; | |
1399 | ||
7608dec2 PZ |
1400 | rq = cpu_rq(cpu); |
1401 | ||
1402 | rcu_read_lock(); | |
316c1608 | 1403 | curr = READ_ONCE(rq->curr); /* unlocked access */ |
7608dec2 | 1404 | |
318e0893 | 1405 | /* |
7608dec2 | 1406 | * If the current task on @p's runqueue is an RT task, then |
e1f47d89 SR |
1407 | * try to see if we can wake this RT task up on another |
1408 | * runqueue. Otherwise simply start this RT task | |
1409 | * on its current runqueue. | |
1410 | * | |
43fa5460 SR |
1411 | * We want to avoid overloading runqueues. If the woken |
1412 | * task is a higher priority, then it will stay on this CPU | |
1413 | * and the lower prio task should be moved to another CPU. | |
1414 | * Even though this will probably make the lower prio task | |
1415 | * lose its cache, we do not want to bounce a higher task | |
1416 | * around just because it gave up its CPU, perhaps for a | |
1417 | * lock? | |
1418 | * | |
1419 | * For equal prio tasks, we just let the scheduler sort it out. | |
7608dec2 PZ |
1420 | * |
1421 | * Otherwise, just let it ride on the affined RQ and the | |
1422 | * post-schedule router will push the preempted task away | |
1423 | * | |
1424 | * This test is optimistic, if we get it wrong the load-balancer | |
1425 | * will have to sort it out. | |
318e0893 | 1426 | */ |
7608dec2 | 1427 | if (curr && unlikely(rt_task(curr)) && |
4b53a341 | 1428 | (curr->nr_cpus_allowed < 2 || |
6bfa687c | 1429 | curr->prio <= p->prio)) { |
7608dec2 | 1430 | int target = find_lowest_rq(p); |
318e0893 | 1431 | |
80e3d87b TC |
1432 | /* |
1433 | * Don't bother moving it if the destination CPU is | |
1434 | * not running a lower priority task. | |
1435 | */ | |
1436 | if (target != -1 && | |
1437 | p->prio < cpu_rq(target)->rt.highest_prio.curr) | |
7608dec2 | 1438 | cpu = target; |
318e0893 | 1439 | } |
7608dec2 | 1440 | rcu_read_unlock(); |
318e0893 | 1441 | |
c37495fd | 1442 | out: |
7608dec2 | 1443 | return cpu; |
e7693a36 | 1444 | } |
7ebefa8c DA |
1445 | |
1446 | static void check_preempt_equal_prio(struct rq *rq, struct task_struct *p) | |
1447 | { | |
308a623a WL |
1448 | /* |
1449 | * Current can't be migrated, useless to reschedule, | |
1450 | * let's hope p can move out. | |
1451 | */ | |
4b53a341 | 1452 | if (rq->curr->nr_cpus_allowed == 1 || |
308a623a | 1453 | !cpupri_find(&rq->rd->cpupri, rq->curr, NULL)) |
7ebefa8c DA |
1454 | return; |
1455 | ||
308a623a WL |
1456 | /* |
1457 | * p is migratable, so let's not schedule it and | |
1458 | * see if it is pushed or pulled somewhere else. | |
1459 | */ | |
4b53a341 | 1460 | if (p->nr_cpus_allowed != 1 |
13b8bd0a RR |
1461 | && cpupri_find(&rq->rd->cpupri, p, NULL)) |
1462 | return; | |
24600ce8 | 1463 | |
7ebefa8c DA |
1464 | /* |
1465 | * There appears to be other cpus that can accept | |
1466 | * current and none to run 'p', so lets reschedule | |
1467 | * to try and push current away: | |
1468 | */ | |
1469 | requeue_task_rt(rq, p, 1); | |
8875125e | 1470 | resched_curr(rq); |
7ebefa8c DA |
1471 | } |
1472 | ||
e7693a36 GH |
1473 | #endif /* CONFIG_SMP */ |
1474 | ||
bb44e5d1 IM |
1475 | /* |
1476 | * Preempt the current task with a newly woken task if needed: | |
1477 | */ | |
7d478721 | 1478 | static void check_preempt_curr_rt(struct rq *rq, struct task_struct *p, int flags) |
bb44e5d1 | 1479 | { |
45c01e82 | 1480 | if (p->prio < rq->curr->prio) { |
8875125e | 1481 | resched_curr(rq); |
45c01e82 GH |
1482 | return; |
1483 | } | |
1484 | ||
1485 | #ifdef CONFIG_SMP | |
1486 | /* | |
1487 | * If: | |
1488 | * | |
1489 | * - the newly woken task is of equal priority to the current task | |
1490 | * - the newly woken task is non-migratable while current is migratable | |
1491 | * - current will be preempted on the next reschedule | |
1492 | * | |
1493 | * we should check to see if current can readily move to a different | |
1494 | * cpu. If so, we will reschedule to allow the push logic to try | |
1495 | * to move current somewhere else, making room for our non-migratable | |
1496 | * task. | |
1497 | */ | |
8dd0de8b | 1498 | if (p->prio == rq->curr->prio && !test_tsk_need_resched(rq->curr)) |
7ebefa8c | 1499 | check_preempt_equal_prio(rq, p); |
45c01e82 | 1500 | #endif |
bb44e5d1 IM |
1501 | } |
1502 | ||
6f505b16 PZ |
1503 | static struct sched_rt_entity *pick_next_rt_entity(struct rq *rq, |
1504 | struct rt_rq *rt_rq) | |
bb44e5d1 | 1505 | { |
6f505b16 PZ |
1506 | struct rt_prio_array *array = &rt_rq->active; |
1507 | struct sched_rt_entity *next = NULL; | |
bb44e5d1 IM |
1508 | struct list_head *queue; |
1509 | int idx; | |
1510 | ||
1511 | idx = sched_find_first_bit(array->bitmap); | |
6f505b16 | 1512 | BUG_ON(idx >= MAX_RT_PRIO); |
bb44e5d1 IM |
1513 | |
1514 | queue = array->queue + idx; | |
6f505b16 | 1515 | next = list_entry(queue->next, struct sched_rt_entity, run_list); |
326587b8 | 1516 | |
6f505b16 PZ |
1517 | return next; |
1518 | } | |
bb44e5d1 | 1519 | |
917b627d | 1520 | static struct task_struct *_pick_next_task_rt(struct rq *rq) |
6f505b16 PZ |
1521 | { |
1522 | struct sched_rt_entity *rt_se; | |
1523 | struct task_struct *p; | |
606dba2e | 1524 | struct rt_rq *rt_rq = &rq->rt; |
6f505b16 PZ |
1525 | |
1526 | do { | |
1527 | rt_se = pick_next_rt_entity(rq, rt_rq); | |
326587b8 | 1528 | BUG_ON(!rt_se); |
6f505b16 PZ |
1529 | rt_rq = group_rt_rq(rt_se); |
1530 | } while (rt_rq); | |
1531 | ||
1532 | p = rt_task_of(rt_se); | |
78becc27 | 1533 | p->se.exec_start = rq_clock_task(rq); |
917b627d GH |
1534 | |
1535 | return p; | |
1536 | } | |
1537 | ||
606dba2e | 1538 | static struct task_struct * |
d8ac8971 | 1539 | pick_next_task_rt(struct rq *rq, struct task_struct *prev, struct rq_flags *rf) |
917b627d | 1540 | { |
606dba2e PZ |
1541 | struct task_struct *p; |
1542 | struct rt_rq *rt_rq = &rq->rt; | |
1543 | ||
37e117c0 | 1544 | if (need_pull_rt_task(rq, prev)) { |
cbce1a68 PZ |
1545 | /* |
1546 | * This is OK, because current is on_cpu, which avoids it being | |
1547 | * picked for load-balance and preemption/IRQs are still | |
1548 | * disabled avoiding further scheduler activity on it and we're | |
1549 | * being very careful to re-start the picking loop. | |
1550 | */ | |
d8ac8971 | 1551 | rq_unpin_lock(rq, rf); |
38033c37 | 1552 | pull_rt_task(rq); |
d8ac8971 | 1553 | rq_repin_lock(rq, rf); |
37e117c0 PZ |
1554 | /* |
1555 | * pull_rt_task() can drop (and re-acquire) rq->lock; this | |
a1d9a323 KT |
1556 | * means a dl or stop task can slip in, in which case we need |
1557 | * to re-start task selection. | |
37e117c0 | 1558 | */ |
da0c1e65 | 1559 | if (unlikely((rq->stop && task_on_rq_queued(rq->stop)) || |
a1d9a323 | 1560 | rq->dl.dl_nr_running)) |
37e117c0 PZ |
1561 | return RETRY_TASK; |
1562 | } | |
38033c37 | 1563 | |
734ff2a7 KT |
1564 | /* |
1565 | * We may dequeue prev's rt_rq in put_prev_task(). | |
1566 | * So, we update time before rt_nr_running check. | |
1567 | */ | |
1568 | if (prev->sched_class == &rt_sched_class) | |
1569 | update_curr_rt(rq); | |
1570 | ||
f4ebcbc0 | 1571 | if (!rt_rq->rt_queued) |
606dba2e PZ |
1572 | return NULL; |
1573 | ||
3f1d2a31 | 1574 | put_prev_task(rq, prev); |
606dba2e PZ |
1575 | |
1576 | p = _pick_next_task_rt(rq); | |
917b627d GH |
1577 | |
1578 | /* The running task is never eligible for pushing */ | |
f3f1768f | 1579 | dequeue_pushable_task(rq, p); |
917b627d | 1580 | |
e3fca9e7 | 1581 | queue_push_tasks(rq); |
3f029d3c | 1582 | |
6f505b16 | 1583 | return p; |
bb44e5d1 IM |
1584 | } |
1585 | ||
31ee529c | 1586 | static void put_prev_task_rt(struct rq *rq, struct task_struct *p) |
bb44e5d1 | 1587 | { |
f1e14ef6 | 1588 | update_curr_rt(rq); |
917b627d GH |
1589 | |
1590 | /* | |
1591 | * The previous task needs to be made eligible for pushing | |
1592 | * if it is still active | |
1593 | */ | |
4b53a341 | 1594 | if (on_rt_rq(&p->rt) && p->nr_cpus_allowed > 1) |
917b627d | 1595 | enqueue_pushable_task(rq, p); |
bb44e5d1 IM |
1596 | } |
1597 | ||
681f3e68 | 1598 | #ifdef CONFIG_SMP |
6f505b16 | 1599 | |
e8fa1362 SR |
1600 | /* Only try algorithms three times */ |
1601 | #define RT_MAX_TRIES 3 | |
1602 | ||
f65eda4f SR |
1603 | static int pick_rt_task(struct rq *rq, struct task_struct *p, int cpu) |
1604 | { | |
1605 | if (!task_running(rq, p) && | |
0c98d344 | 1606 | cpumask_test_cpu(cpu, &p->cpus_allowed)) |
f65eda4f SR |
1607 | return 1; |
1608 | return 0; | |
1609 | } | |
1610 | ||
e23ee747 KT |
1611 | /* |
1612 | * Return the highest pushable rq's task, which is suitable to be executed | |
1613 | * on the cpu, NULL otherwise | |
1614 | */ | |
1615 | static struct task_struct *pick_highest_pushable_task(struct rq *rq, int cpu) | |
e8fa1362 | 1616 | { |
e23ee747 KT |
1617 | struct plist_head *head = &rq->rt.pushable_tasks; |
1618 | struct task_struct *p; | |
3d07467b | 1619 | |
e23ee747 KT |
1620 | if (!has_pushable_tasks(rq)) |
1621 | return NULL; | |
3d07467b | 1622 | |
e23ee747 KT |
1623 | plist_for_each_entry(p, head, pushable_tasks) { |
1624 | if (pick_rt_task(rq, p, cpu)) | |
1625 | return p; | |
f65eda4f SR |
1626 | } |
1627 | ||
e23ee747 | 1628 | return NULL; |
e8fa1362 SR |
1629 | } |
1630 | ||
0e3900e6 | 1631 | static DEFINE_PER_CPU(cpumask_var_t, local_cpu_mask); |
e8fa1362 | 1632 | |
6e1254d2 GH |
1633 | static int find_lowest_rq(struct task_struct *task) |
1634 | { | |
1635 | struct sched_domain *sd; | |
4ba29684 | 1636 | struct cpumask *lowest_mask = this_cpu_cpumask_var_ptr(local_cpu_mask); |
6e1254d2 GH |
1637 | int this_cpu = smp_processor_id(); |
1638 | int cpu = task_cpu(task); | |
06f90dbd | 1639 | |
0da938c4 SR |
1640 | /* Make sure the mask is initialized first */ |
1641 | if (unlikely(!lowest_mask)) | |
1642 | return -1; | |
1643 | ||
4b53a341 | 1644 | if (task->nr_cpus_allowed == 1) |
6e0534f2 | 1645 | return -1; /* No other targets possible */ |
6e1254d2 | 1646 | |
6e0534f2 GH |
1647 | if (!cpupri_find(&task_rq(task)->rd->cpupri, task, lowest_mask)) |
1648 | return -1; /* No targets found */ | |
6e1254d2 GH |
1649 | |
1650 | /* | |
1651 | * At this point we have built a mask of cpus representing the | |
1652 | * lowest priority tasks in the system. Now we want to elect | |
1653 | * the best one based on our affinity and topology. | |
1654 | * | |
1655 | * We prioritize the last cpu that the task executed on since | |
1656 | * it is most likely cache-hot in that location. | |
1657 | */ | |
96f874e2 | 1658 | if (cpumask_test_cpu(cpu, lowest_mask)) |
6e1254d2 GH |
1659 | return cpu; |
1660 | ||
1661 | /* | |
1662 | * Otherwise, we consult the sched_domains span maps to figure | |
1663 | * out which cpu is logically closest to our hot cache data. | |
1664 | */ | |
e2c88063 RR |
1665 | if (!cpumask_test_cpu(this_cpu, lowest_mask)) |
1666 | this_cpu = -1; /* Skip this_cpu opt if not among lowest */ | |
6e1254d2 | 1667 | |
cd4ae6ad | 1668 | rcu_read_lock(); |
e2c88063 RR |
1669 | for_each_domain(cpu, sd) { |
1670 | if (sd->flags & SD_WAKE_AFFINE) { | |
1671 | int best_cpu; | |
6e1254d2 | 1672 | |
e2c88063 RR |
1673 | /* |
1674 | * "this_cpu" is cheaper to preempt than a | |
1675 | * remote processor. | |
1676 | */ | |
1677 | if (this_cpu != -1 && | |
cd4ae6ad XF |
1678 | cpumask_test_cpu(this_cpu, sched_domain_span(sd))) { |
1679 | rcu_read_unlock(); | |
e2c88063 | 1680 | return this_cpu; |
cd4ae6ad | 1681 | } |
e2c88063 RR |
1682 | |
1683 | best_cpu = cpumask_first_and(lowest_mask, | |
1684 | sched_domain_span(sd)); | |
cd4ae6ad XF |
1685 | if (best_cpu < nr_cpu_ids) { |
1686 | rcu_read_unlock(); | |
e2c88063 | 1687 | return best_cpu; |
cd4ae6ad | 1688 | } |
6e1254d2 GH |
1689 | } |
1690 | } | |
cd4ae6ad | 1691 | rcu_read_unlock(); |
6e1254d2 GH |
1692 | |
1693 | /* | |
1694 | * And finally, if there were no matches within the domains | |
1695 | * just give the caller *something* to work with from the compatible | |
1696 | * locations. | |
1697 | */ | |
e2c88063 RR |
1698 | if (this_cpu != -1) |
1699 | return this_cpu; | |
1700 | ||
1701 | cpu = cpumask_any(lowest_mask); | |
1702 | if (cpu < nr_cpu_ids) | |
1703 | return cpu; | |
1704 | return -1; | |
07b4032c GH |
1705 | } |
1706 | ||
1707 | /* Will lock the rq it finds */ | |
4df64c0b | 1708 | static struct rq *find_lock_lowest_rq(struct task_struct *task, struct rq *rq) |
07b4032c GH |
1709 | { |
1710 | struct rq *lowest_rq = NULL; | |
07b4032c | 1711 | int tries; |
4df64c0b | 1712 | int cpu; |
e8fa1362 | 1713 | |
07b4032c GH |
1714 | for (tries = 0; tries < RT_MAX_TRIES; tries++) { |
1715 | cpu = find_lowest_rq(task); | |
1716 | ||
2de0b463 | 1717 | if ((cpu == -1) || (cpu == rq->cpu)) |
e8fa1362 SR |
1718 | break; |
1719 | ||
07b4032c GH |
1720 | lowest_rq = cpu_rq(cpu); |
1721 | ||
80e3d87b TC |
1722 | if (lowest_rq->rt.highest_prio.curr <= task->prio) { |
1723 | /* | |
1724 | * Target rq has tasks of equal or higher priority, | |
1725 | * retrying does not release any lock and is unlikely | |
1726 | * to yield a different result. | |
1727 | */ | |
1728 | lowest_rq = NULL; | |
1729 | break; | |
1730 | } | |
1731 | ||
e8fa1362 | 1732 | /* if the prio of this runqueue changed, try again */ |
07b4032c | 1733 | if (double_lock_balance(rq, lowest_rq)) { |
e8fa1362 SR |
1734 | /* |
1735 | * We had to unlock the run queue. In | |
1736 | * the mean time, task could have | |
1737 | * migrated already or had its affinity changed. | |
1738 | * Also make sure that it wasn't scheduled on its rq. | |
1739 | */ | |
07b4032c | 1740 | if (unlikely(task_rq(task) != rq || |
0c98d344 | 1741 | !cpumask_test_cpu(lowest_rq->cpu, &task->cpus_allowed) || |
07b4032c | 1742 | task_running(rq, task) || |
13b5ab02 | 1743 | !rt_task(task) || |
da0c1e65 | 1744 | !task_on_rq_queued(task))) { |
4df64c0b | 1745 | |
7f1b4393 | 1746 | double_unlock_balance(rq, lowest_rq); |
e8fa1362 SR |
1747 | lowest_rq = NULL; |
1748 | break; | |
1749 | } | |
1750 | } | |
1751 | ||
1752 | /* If this rq is still suitable use it. */ | |
e864c499 | 1753 | if (lowest_rq->rt.highest_prio.curr > task->prio) |
e8fa1362 SR |
1754 | break; |
1755 | ||
1756 | /* try again */ | |
1b12bbc7 | 1757 | double_unlock_balance(rq, lowest_rq); |
e8fa1362 SR |
1758 | lowest_rq = NULL; |
1759 | } | |
1760 | ||
1761 | return lowest_rq; | |
1762 | } | |
1763 | ||
917b627d GH |
1764 | static struct task_struct *pick_next_pushable_task(struct rq *rq) |
1765 | { | |
1766 | struct task_struct *p; | |
1767 | ||
1768 | if (!has_pushable_tasks(rq)) | |
1769 | return NULL; | |
1770 | ||
1771 | p = plist_first_entry(&rq->rt.pushable_tasks, | |
1772 | struct task_struct, pushable_tasks); | |
1773 | ||
1774 | BUG_ON(rq->cpu != task_cpu(p)); | |
1775 | BUG_ON(task_current(rq, p)); | |
4b53a341 | 1776 | BUG_ON(p->nr_cpus_allowed <= 1); |
917b627d | 1777 | |
da0c1e65 | 1778 | BUG_ON(!task_on_rq_queued(p)); |
917b627d GH |
1779 | BUG_ON(!rt_task(p)); |
1780 | ||
1781 | return p; | |
1782 | } | |
1783 | ||
e8fa1362 SR |
1784 | /* |
1785 | * If the current CPU has more than one RT task, see if the non | |
1786 | * running task can migrate over to a CPU that is running a task | |
1787 | * of lesser priority. | |
1788 | */ | |
697f0a48 | 1789 | static int push_rt_task(struct rq *rq) |
e8fa1362 SR |
1790 | { |
1791 | struct task_struct *next_task; | |
1792 | struct rq *lowest_rq; | |
311e800e | 1793 | int ret = 0; |
e8fa1362 | 1794 | |
a22d7fc1 GH |
1795 | if (!rq->rt.overloaded) |
1796 | return 0; | |
1797 | ||
917b627d | 1798 | next_task = pick_next_pushable_task(rq); |
e8fa1362 SR |
1799 | if (!next_task) |
1800 | return 0; | |
1801 | ||
49246274 | 1802 | retry: |
697f0a48 | 1803 | if (unlikely(next_task == rq->curr)) { |
f65eda4f | 1804 | WARN_ON(1); |
e8fa1362 | 1805 | return 0; |
f65eda4f | 1806 | } |
e8fa1362 SR |
1807 | |
1808 | /* | |
1809 | * It's possible that the next_task slipped in of | |
1810 | * higher priority than current. If that's the case | |
1811 | * just reschedule current. | |
1812 | */ | |
697f0a48 | 1813 | if (unlikely(next_task->prio < rq->curr->prio)) { |
8875125e | 1814 | resched_curr(rq); |
e8fa1362 SR |
1815 | return 0; |
1816 | } | |
1817 | ||
697f0a48 | 1818 | /* We might release rq lock */ |
e8fa1362 SR |
1819 | get_task_struct(next_task); |
1820 | ||
1821 | /* find_lock_lowest_rq locks the rq if found */ | |
697f0a48 | 1822 | lowest_rq = find_lock_lowest_rq(next_task, rq); |
e8fa1362 SR |
1823 | if (!lowest_rq) { |
1824 | struct task_struct *task; | |
1825 | /* | |
311e800e | 1826 | * find_lock_lowest_rq releases rq->lock |
1563513d GH |
1827 | * so it is possible that next_task has migrated. |
1828 | * | |
1829 | * We need to make sure that the task is still on the same | |
1830 | * run-queue and is also still the next task eligible for | |
1831 | * pushing. | |
e8fa1362 | 1832 | */ |
917b627d | 1833 | task = pick_next_pushable_task(rq); |
de16b91e | 1834 | if (task == next_task) { |
1563513d | 1835 | /* |
311e800e HD |
1836 | * The task hasn't migrated, and is still the next |
1837 | * eligible task, but we failed to find a run-queue | |
1838 | * to push it to. Do not retry in this case, since | |
1839 | * other cpus will pull from us when ready. | |
1563513d | 1840 | */ |
1563513d | 1841 | goto out; |
e8fa1362 | 1842 | } |
917b627d | 1843 | |
1563513d GH |
1844 | if (!task) |
1845 | /* No more tasks, just exit */ | |
1846 | goto out; | |
1847 | ||
917b627d | 1848 | /* |
1563513d | 1849 | * Something has shifted, try again. |
917b627d | 1850 | */ |
1563513d GH |
1851 | put_task_struct(next_task); |
1852 | next_task = task; | |
1853 | goto retry; | |
e8fa1362 SR |
1854 | } |
1855 | ||
697f0a48 | 1856 | deactivate_task(rq, next_task, 0); |
e8fa1362 SR |
1857 | set_task_cpu(next_task, lowest_rq->cpu); |
1858 | activate_task(lowest_rq, next_task, 0); | |
311e800e | 1859 | ret = 1; |
e8fa1362 | 1860 | |
8875125e | 1861 | resched_curr(lowest_rq); |
e8fa1362 | 1862 | |
1b12bbc7 | 1863 | double_unlock_balance(rq, lowest_rq); |
e8fa1362 | 1864 | |
e8fa1362 SR |
1865 | out: |
1866 | put_task_struct(next_task); | |
1867 | ||
311e800e | 1868 | return ret; |
e8fa1362 SR |
1869 | } |
1870 | ||
e8fa1362 SR |
1871 | static void push_rt_tasks(struct rq *rq) |
1872 | { | |
1873 | /* push_rt_task will return true if it moved an RT */ | |
1874 | while (push_rt_task(rq)) | |
1875 | ; | |
1876 | } | |
1877 | ||
b6366f04 SR |
1878 | #ifdef HAVE_RT_PUSH_IPI |
1879 | /* | |
1880 | * The search for the next cpu always starts at rq->cpu and ends | |
1881 | * when we reach rq->cpu again. It will never return rq->cpu. | |
1882 | * This returns the next cpu to check, or nr_cpu_ids if the loop | |
1883 | * is complete. | |
1884 | * | |
1885 | * rq->rt.push_cpu holds the last cpu returned by this function, | |
1886 | * or if this is the first instance, it must hold rq->cpu. | |
1887 | */ | |
1888 | static int rto_next_cpu(struct rq *rq) | |
1889 | { | |
1890 | int prev_cpu = rq->rt.push_cpu; | |
1891 | int cpu; | |
1892 | ||
1893 | cpu = cpumask_next(prev_cpu, rq->rd->rto_mask); | |
1894 | ||
1895 | /* | |
1896 | * If the previous cpu is less than the rq's CPU, then it already | |
1897 | * passed the end of the mask, and has started from the beginning. | |
1898 | * We end if the next CPU is greater or equal to rq's CPU. | |
1899 | */ | |
1900 | if (prev_cpu < rq->cpu) { | |
1901 | if (cpu >= rq->cpu) | |
1902 | return nr_cpu_ids; | |
1903 | ||
1904 | } else if (cpu >= nr_cpu_ids) { | |
1905 | /* | |
1906 | * We passed the end of the mask, start at the beginning. | |
1907 | * If the result is greater or equal to the rq's CPU, then | |
1908 | * the loop is finished. | |
1909 | */ | |
1910 | cpu = cpumask_first(rq->rd->rto_mask); | |
1911 | if (cpu >= rq->cpu) | |
1912 | return nr_cpu_ids; | |
1913 | } | |
1914 | rq->rt.push_cpu = cpu; | |
1915 | ||
1916 | /* Return cpu to let the caller know if the loop is finished or not */ | |
1917 | return cpu; | |
1918 | } | |
1919 | ||
1920 | static int find_next_push_cpu(struct rq *rq) | |
1921 | { | |
1922 | struct rq *next_rq; | |
1923 | int cpu; | |
1924 | ||
1925 | while (1) { | |
1926 | cpu = rto_next_cpu(rq); | |
1927 | if (cpu >= nr_cpu_ids) | |
1928 | break; | |
1929 | next_rq = cpu_rq(cpu); | |
1930 | ||
1931 | /* Make sure the next rq can push to this rq */ | |
1932 | if (next_rq->rt.highest_prio.next < rq->rt.highest_prio.curr) | |
1933 | break; | |
1934 | } | |
1935 | ||
1936 | return cpu; | |
1937 | } | |
1938 | ||
1939 | #define RT_PUSH_IPI_EXECUTING 1 | |
1940 | #define RT_PUSH_IPI_RESTART 2 | |
1941 | ||
3e777f99 SRV |
1942 | /* |
1943 | * When a high priority task schedules out from a CPU and a lower priority | |
1944 | * task is scheduled in, a check is made to see if there's any RT tasks | |
1945 | * on other CPUs that are waiting to run because a higher priority RT task | |
1946 | * is currently running on its CPU. In this case, the CPU with multiple RT | |
1947 | * tasks queued on it (overloaded) needs to be notified that a CPU has opened | |
1948 | * up that may be able to run one of its non-running queued RT tasks. | |
1949 | * | |
1950 | * On large CPU boxes, there's the case that several CPUs could schedule | |
1951 | * a lower priority task at the same time, in which case it will look for | |
1952 | * any overloaded CPUs that it could pull a task from. To do this, the runqueue | |
1953 | * lock must be taken from that overloaded CPU. Having 10s of CPUs all fighting | |
1954 | * for a single overloaded CPU's runqueue lock can produce a large latency. | |
1955 | * (This has actually been observed on large boxes running cyclictest). | |
1956 | * Instead of taking the runqueue lock of the overloaded CPU, each of the | |
1957 | * CPUs that scheduled a lower priority task simply sends an IPI to the | |
1958 | * overloaded CPU. An IPI is much cheaper than taking an runqueue lock with | |
1959 | * lots of contention. The overloaded CPU will look to push its non-running | |
1960 | * RT task off, and if it does, it can then ignore the other IPIs coming | |
1961 | * in, and just pass those IPIs off to any other overloaded CPU. | |
1962 | * | |
1963 | * When a CPU schedules a lower priority task, it only sends an IPI to | |
1964 | * the "next" CPU that has overloaded RT tasks. This prevents IPI storms, | |
1965 | * as having 10 CPUs scheduling lower priority tasks and 10 CPUs with | |
1966 | * RT overloaded tasks, would cause 100 IPIs to go out at once. | |
1967 | * | |
1968 | * The overloaded RT CPU, when receiving an IPI, will try to push off its | |
1969 | * overloaded RT tasks and then send an IPI to the next CPU that has | |
1970 | * overloaded RT tasks. This stops when all CPUs with overloaded RT tasks | |
1971 | * have completed. Just because a CPU may have pushed off its own overloaded | |
1972 | * RT task does not mean it should stop sending the IPI around to other | |
1973 | * overloaded CPUs. There may be another RT task waiting to run on one of | |
1974 | * those CPUs that are of higher priority than the one that was just | |
1975 | * pushed. | |
1976 | * | |
1977 | * An optimization that could possibly be made is to make a CPU array similar | |
1978 | * to the cpupri array mask of all running RT tasks, but for the overloaded | |
1979 | * case, then the IPI could be sent to only the CPU with the highest priority | |
1980 | * RT task waiting, and that CPU could send off further IPIs to the CPU with | |
1981 | * the next highest waiting task. Since the overloaded case is much less likely | |
1982 | * to happen, the complexity of this implementation may not be worth it. | |
1983 | * Instead, just send an IPI around to all overloaded CPUs. | |
1984 | * | |
1985 | * The rq->rt.push_flags holds the status of the IPI that is going around. | |
1986 | * A run queue can only send out a single IPI at a time. The possible flags | |
1987 | * for rq->rt.push_flags are: | |
1988 | * | |
1989 | * (None or zero): No IPI is going around for the current rq | |
1990 | * RT_PUSH_IPI_EXECUTING: An IPI for the rq is being passed around | |
1991 | * RT_PUSH_IPI_RESTART: The priority of the running task for the rq | |
1992 | * has changed, and the IPI should restart | |
1993 | * circulating the overloaded CPUs again. | |
1994 | * | |
1995 | * rq->rt.push_cpu contains the CPU that is being sent the IPI. It is updated | |
1996 | * before sending to the next CPU. | |
1997 | * | |
1998 | * Instead of having all CPUs that schedule a lower priority task send | |
1999 | * an IPI to the same "first" CPU in the RT overload mask, they send it | |
2000 | * to the next overloaded CPU after their own CPU. This helps distribute | |
2001 | * the work when there's more than one overloaded CPU and multiple CPUs | |
2002 | * scheduling in lower priority tasks. | |
2003 | * | |
2004 | * When a rq schedules a lower priority task than what was currently | |
2005 | * running, the next CPU with overloaded RT tasks is examined first. | |
2006 | * That is, if CPU 1 and 5 are overloaded, and CPU 3 schedules a lower | |
2007 | * priority task, it will send an IPI first to CPU 5, then CPU 5 will | |
2008 | * send to CPU 1 if it is still overloaded. CPU 1 will clear the | |
2009 | * rq->rt.push_flags if RT_PUSH_IPI_RESTART is not set. | |
2010 | * | |
2011 | * The first CPU to notice IPI_RESTART is set, will clear that flag and then | |
2012 | * send an IPI to the next overloaded CPU after the rq->cpu and not the next | |
2013 | * CPU after push_cpu. That is, if CPU 1, 4 and 5 are overloaded when CPU 3 | |
2014 | * schedules a lower priority task, and the IPI_RESTART gets set while the | |
2015 | * handling is being done on CPU 5, it will clear the flag and send it back to | |
2016 | * CPU 4 instead of CPU 1. | |
2017 | * | |
2018 | * Note, the above logic can be disabled by turning off the sched_feature | |
2019 | * RT_PUSH_IPI. Then the rq lock of the overloaded CPU will simply be | |
2020 | * taken by the CPU requesting a pull and the waiting RT task will be pulled | |
2021 | * by that CPU. This may be fine for machines with few CPUs. | |
2022 | */ | |
b6366f04 SR |
2023 | static void tell_cpu_to_push(struct rq *rq) |
2024 | { | |
2025 | int cpu; | |
2026 | ||
2027 | if (rq->rt.push_flags & RT_PUSH_IPI_EXECUTING) { | |
2028 | raw_spin_lock(&rq->rt.push_lock); | |
2029 | /* Make sure it's still executing */ | |
2030 | if (rq->rt.push_flags & RT_PUSH_IPI_EXECUTING) { | |
2031 | /* | |
2032 | * Tell the IPI to restart the loop as things have | |
2033 | * changed since it started. | |
2034 | */ | |
2035 | rq->rt.push_flags |= RT_PUSH_IPI_RESTART; | |
2036 | raw_spin_unlock(&rq->rt.push_lock); | |
2037 | return; | |
2038 | } | |
2039 | raw_spin_unlock(&rq->rt.push_lock); | |
2040 | } | |
2041 | ||
2042 | /* When here, there's no IPI going around */ | |
2043 | ||
2044 | rq->rt.push_cpu = rq->cpu; | |
2045 | cpu = find_next_push_cpu(rq); | |
2046 | if (cpu >= nr_cpu_ids) | |
2047 | return; | |
2048 | ||
2049 | rq->rt.push_flags = RT_PUSH_IPI_EXECUTING; | |
2050 | ||
2051 | irq_work_queue_on(&rq->rt.push_work, cpu); | |
2052 | } | |
2053 | ||
2054 | /* Called from hardirq context */ | |
2055 | static void try_to_push_tasks(void *arg) | |
2056 | { | |
2057 | struct rt_rq *rt_rq = arg; | |
2058 | struct rq *rq, *src_rq; | |
2059 | int this_cpu; | |
2060 | int cpu; | |
2061 | ||
2062 | this_cpu = rt_rq->push_cpu; | |
2063 | ||
2064 | /* Paranoid check */ | |
2065 | BUG_ON(this_cpu != smp_processor_id()); | |
2066 | ||
2067 | rq = cpu_rq(this_cpu); | |
2068 | src_rq = rq_of_rt_rq(rt_rq); | |
2069 | ||
2070 | again: | |
2071 | if (has_pushable_tasks(rq)) { | |
2072 | raw_spin_lock(&rq->lock); | |
2073 | push_rt_task(rq); | |
2074 | raw_spin_unlock(&rq->lock); | |
2075 | } | |
2076 | ||
2077 | /* Pass the IPI to the next rt overloaded queue */ | |
2078 | raw_spin_lock(&rt_rq->push_lock); | |
2079 | /* | |
2080 | * If the source queue changed since the IPI went out, | |
2081 | * we need to restart the search from that CPU again. | |
2082 | */ | |
2083 | if (rt_rq->push_flags & RT_PUSH_IPI_RESTART) { | |
2084 | rt_rq->push_flags &= ~RT_PUSH_IPI_RESTART; | |
2085 | rt_rq->push_cpu = src_rq->cpu; | |
2086 | } | |
2087 | ||
2088 | cpu = find_next_push_cpu(src_rq); | |
2089 | ||
2090 | if (cpu >= nr_cpu_ids) | |
2091 | rt_rq->push_flags &= ~RT_PUSH_IPI_EXECUTING; | |
2092 | raw_spin_unlock(&rt_rq->push_lock); | |
2093 | ||
2094 | if (cpu >= nr_cpu_ids) | |
2095 | return; | |
2096 | ||
2097 | /* | |
2098 | * It is possible that a restart caused this CPU to be | |
2099 | * chosen again. Don't bother with an IPI, just see if we | |
2100 | * have more to push. | |
2101 | */ | |
2102 | if (unlikely(cpu == rq->cpu)) | |
2103 | goto again; | |
2104 | ||
2105 | /* Try the next RT overloaded CPU */ | |
2106 | irq_work_queue_on(&rt_rq->push_work, cpu); | |
2107 | } | |
2108 | ||
2109 | static void push_irq_work_func(struct irq_work *work) | |
2110 | { | |
2111 | struct rt_rq *rt_rq = container_of(work, struct rt_rq, push_work); | |
2112 | ||
2113 | try_to_push_tasks(rt_rq); | |
2114 | } | |
2115 | #endif /* HAVE_RT_PUSH_IPI */ | |
2116 | ||
8046d680 | 2117 | static void pull_rt_task(struct rq *this_rq) |
f65eda4f | 2118 | { |
8046d680 PZ |
2119 | int this_cpu = this_rq->cpu, cpu; |
2120 | bool resched = false; | |
a8728944 | 2121 | struct task_struct *p; |
f65eda4f | 2122 | struct rq *src_rq; |
f65eda4f | 2123 | |
637f5085 | 2124 | if (likely(!rt_overloaded(this_rq))) |
8046d680 | 2125 | return; |
f65eda4f | 2126 | |
7c3f2ab7 PZ |
2127 | /* |
2128 | * Match the barrier from rt_set_overloaded; this guarantees that if we | |
2129 | * see overloaded we must also see the rto_mask bit. | |
2130 | */ | |
2131 | smp_rmb(); | |
2132 | ||
b6366f04 SR |
2133 | #ifdef HAVE_RT_PUSH_IPI |
2134 | if (sched_feat(RT_PUSH_IPI)) { | |
2135 | tell_cpu_to_push(this_rq); | |
8046d680 | 2136 | return; |
b6366f04 SR |
2137 | } |
2138 | #endif | |
2139 | ||
c6c4927b | 2140 | for_each_cpu(cpu, this_rq->rd->rto_mask) { |
f65eda4f SR |
2141 | if (this_cpu == cpu) |
2142 | continue; | |
2143 | ||
2144 | src_rq = cpu_rq(cpu); | |
74ab8e4f GH |
2145 | |
2146 | /* | |
2147 | * Don't bother taking the src_rq->lock if the next highest | |
2148 | * task is known to be lower-priority than our current task. | |
2149 | * This may look racy, but if this value is about to go | |
2150 | * logically higher, the src_rq will push this task away. | |
2151 | * And if its going logically lower, we do not care | |
2152 | */ | |
2153 | if (src_rq->rt.highest_prio.next >= | |
2154 | this_rq->rt.highest_prio.curr) | |
2155 | continue; | |
2156 | ||
f65eda4f SR |
2157 | /* |
2158 | * We can potentially drop this_rq's lock in | |
2159 | * double_lock_balance, and another CPU could | |
a8728944 | 2160 | * alter this_rq |
f65eda4f | 2161 | */ |
a8728944 | 2162 | double_lock_balance(this_rq, src_rq); |
f65eda4f SR |
2163 | |
2164 | /* | |
e23ee747 KT |
2165 | * We can pull only a task, which is pushable |
2166 | * on its rq, and no others. | |
f65eda4f | 2167 | */ |
e23ee747 | 2168 | p = pick_highest_pushable_task(src_rq, this_cpu); |
f65eda4f SR |
2169 | |
2170 | /* | |
2171 | * Do we have an RT task that preempts | |
2172 | * the to-be-scheduled task? | |
2173 | */ | |
a8728944 | 2174 | if (p && (p->prio < this_rq->rt.highest_prio.curr)) { |
f65eda4f | 2175 | WARN_ON(p == src_rq->curr); |
da0c1e65 | 2176 | WARN_ON(!task_on_rq_queued(p)); |
f65eda4f SR |
2177 | |
2178 | /* | |
2179 | * There's a chance that p is higher in priority | |
2180 | * than what's currently running on its cpu. | |
2181 | * This is just that p is wakeing up and hasn't | |
2182 | * had a chance to schedule. We only pull | |
2183 | * p if it is lower in priority than the | |
a8728944 | 2184 | * current task on the run queue |
f65eda4f | 2185 | */ |
a8728944 | 2186 | if (p->prio < src_rq->curr->prio) |
614ee1f6 | 2187 | goto skip; |
f65eda4f | 2188 | |
8046d680 | 2189 | resched = true; |
f65eda4f SR |
2190 | |
2191 | deactivate_task(src_rq, p, 0); | |
2192 | set_task_cpu(p, this_cpu); | |
2193 | activate_task(this_rq, p, 0); | |
2194 | /* | |
2195 | * We continue with the search, just in | |
2196 | * case there's an even higher prio task | |
25985edc | 2197 | * in another runqueue. (low likelihood |
f65eda4f | 2198 | * but possible) |
f65eda4f | 2199 | */ |
f65eda4f | 2200 | } |
49246274 | 2201 | skip: |
1b12bbc7 | 2202 | double_unlock_balance(this_rq, src_rq); |
f65eda4f SR |
2203 | } |
2204 | ||
8046d680 PZ |
2205 | if (resched) |
2206 | resched_curr(this_rq); | |
f65eda4f SR |
2207 | } |
2208 | ||
8ae121ac GH |
2209 | /* |
2210 | * If we are not running and we are not going to reschedule soon, we should | |
2211 | * try to push tasks away now | |
2212 | */ | |
efbbd05a | 2213 | static void task_woken_rt(struct rq *rq, struct task_struct *p) |
4642dafd | 2214 | { |
9a897c5a | 2215 | if (!task_running(rq, p) && |
8ae121ac | 2216 | !test_tsk_need_resched(rq->curr) && |
4b53a341 | 2217 | p->nr_cpus_allowed > 1 && |
1baca4ce | 2218 | (dl_task(rq->curr) || rt_task(rq->curr)) && |
4b53a341 | 2219 | (rq->curr->nr_cpus_allowed < 2 || |
3be209a8 | 2220 | rq->curr->prio <= p->prio)) |
4642dafd SR |
2221 | push_rt_tasks(rq); |
2222 | } | |
2223 | ||
bdd7c81b | 2224 | /* Assumes rq->lock is held */ |
1f11eb6a | 2225 | static void rq_online_rt(struct rq *rq) |
bdd7c81b IM |
2226 | { |
2227 | if (rq->rt.overloaded) | |
2228 | rt_set_overload(rq); | |
6e0534f2 | 2229 | |
7def2be1 PZ |
2230 | __enable_runtime(rq); |
2231 | ||
e864c499 | 2232 | cpupri_set(&rq->rd->cpupri, rq->cpu, rq->rt.highest_prio.curr); |
bdd7c81b IM |
2233 | } |
2234 | ||
2235 | /* Assumes rq->lock is held */ | |
1f11eb6a | 2236 | static void rq_offline_rt(struct rq *rq) |
bdd7c81b IM |
2237 | { |
2238 | if (rq->rt.overloaded) | |
2239 | rt_clear_overload(rq); | |
6e0534f2 | 2240 | |
7def2be1 PZ |
2241 | __disable_runtime(rq); |
2242 | ||
6e0534f2 | 2243 | cpupri_set(&rq->rd->cpupri, rq->cpu, CPUPRI_INVALID); |
bdd7c81b | 2244 | } |
cb469845 SR |
2245 | |
2246 | /* | |
2247 | * When switch from the rt queue, we bring ourselves to a position | |
2248 | * that we might want to pull RT tasks from other runqueues. | |
2249 | */ | |
da7a735e | 2250 | static void switched_from_rt(struct rq *rq, struct task_struct *p) |
cb469845 SR |
2251 | { |
2252 | /* | |
2253 | * If there are other RT tasks then we will reschedule | |
2254 | * and the scheduling of the other RT tasks will handle | |
2255 | * the balancing. But if we are the last RT task | |
2256 | * we may need to handle the pulling of RT tasks | |
2257 | * now. | |
2258 | */ | |
da0c1e65 | 2259 | if (!task_on_rq_queued(p) || rq->rt.rt_nr_running) |
1158ddb5 KT |
2260 | return; |
2261 | ||
fd7a4bed | 2262 | queue_pull_task(rq); |
cb469845 | 2263 | } |
3d8cbdf8 | 2264 | |
11c785b7 | 2265 | void __init init_sched_rt_class(void) |
3d8cbdf8 RR |
2266 | { |
2267 | unsigned int i; | |
2268 | ||
029632fb | 2269 | for_each_possible_cpu(i) { |
eaa95840 | 2270 | zalloc_cpumask_var_node(&per_cpu(local_cpu_mask, i), |
6ca09dfc | 2271 | GFP_KERNEL, cpu_to_node(i)); |
029632fb | 2272 | } |
3d8cbdf8 | 2273 | } |
cb469845 SR |
2274 | #endif /* CONFIG_SMP */ |
2275 | ||
2276 | /* | |
2277 | * When switching a task to RT, we may overload the runqueue | |
2278 | * with RT tasks. In this case we try to push them off to | |
2279 | * other runqueues. | |
2280 | */ | |
da7a735e | 2281 | static void switched_to_rt(struct rq *rq, struct task_struct *p) |
cb469845 | 2282 | { |
cb469845 SR |
2283 | /* |
2284 | * If we are already running, then there's nothing | |
2285 | * that needs to be done. But if we are not running | |
2286 | * we may need to preempt the current running task. | |
2287 | * If that current running task is also an RT task | |
2288 | * then see if we can move to another run queue. | |
2289 | */ | |
da0c1e65 | 2290 | if (task_on_rq_queued(p) && rq->curr != p) { |
cb469845 | 2291 | #ifdef CONFIG_SMP |
4b53a341 | 2292 | if (p->nr_cpus_allowed > 1 && rq->rt.overloaded) |
fd7a4bed | 2293 | queue_push_tasks(rq); |
619bd4a7 | 2294 | #endif /* CONFIG_SMP */ |
fd7a4bed | 2295 | if (p->prio < rq->curr->prio) |
8875125e | 2296 | resched_curr(rq); |
cb469845 SR |
2297 | } |
2298 | } | |
2299 | ||
2300 | /* | |
2301 | * Priority of the task has changed. This may cause | |
2302 | * us to initiate a push or pull. | |
2303 | */ | |
da7a735e PZ |
2304 | static void |
2305 | prio_changed_rt(struct rq *rq, struct task_struct *p, int oldprio) | |
cb469845 | 2306 | { |
da0c1e65 | 2307 | if (!task_on_rq_queued(p)) |
da7a735e PZ |
2308 | return; |
2309 | ||
2310 | if (rq->curr == p) { | |
cb469845 SR |
2311 | #ifdef CONFIG_SMP |
2312 | /* | |
2313 | * If our priority decreases while running, we | |
2314 | * may need to pull tasks to this runqueue. | |
2315 | */ | |
2316 | if (oldprio < p->prio) | |
fd7a4bed PZ |
2317 | queue_pull_task(rq); |
2318 | ||
cb469845 SR |
2319 | /* |
2320 | * If there's a higher priority task waiting to run | |
fd7a4bed | 2321 | * then reschedule. |
cb469845 | 2322 | */ |
fd7a4bed | 2323 | if (p->prio > rq->rt.highest_prio.curr) |
8875125e | 2324 | resched_curr(rq); |
cb469845 SR |
2325 | #else |
2326 | /* For UP simply resched on drop of prio */ | |
2327 | if (oldprio < p->prio) | |
8875125e | 2328 | resched_curr(rq); |
e8fa1362 | 2329 | #endif /* CONFIG_SMP */ |
cb469845 SR |
2330 | } else { |
2331 | /* | |
2332 | * This task is not running, but if it is | |
2333 | * greater than the current running task | |
2334 | * then reschedule. | |
2335 | */ | |
2336 | if (p->prio < rq->curr->prio) | |
8875125e | 2337 | resched_curr(rq); |
cb469845 SR |
2338 | } |
2339 | } | |
2340 | ||
b18b6a9c | 2341 | #ifdef CONFIG_POSIX_TIMERS |
78f2c7db PZ |
2342 | static void watchdog(struct rq *rq, struct task_struct *p) |
2343 | { | |
2344 | unsigned long soft, hard; | |
2345 | ||
78d7d407 JS |
2346 | /* max may change after cur was read, this will be fixed next tick */ |
2347 | soft = task_rlimit(p, RLIMIT_RTTIME); | |
2348 | hard = task_rlimit_max(p, RLIMIT_RTTIME); | |
78f2c7db PZ |
2349 | |
2350 | if (soft != RLIM_INFINITY) { | |
2351 | unsigned long next; | |
2352 | ||
57d2aa00 YX |
2353 | if (p->rt.watchdog_stamp != jiffies) { |
2354 | p->rt.timeout++; | |
2355 | p->rt.watchdog_stamp = jiffies; | |
2356 | } | |
2357 | ||
78f2c7db | 2358 | next = DIV_ROUND_UP(min(soft, hard), USEC_PER_SEC/HZ); |
5a52dd50 | 2359 | if (p->rt.timeout > next) |
f06febc9 | 2360 | p->cputime_expires.sched_exp = p->se.sum_exec_runtime; |
78f2c7db PZ |
2361 | } |
2362 | } | |
b18b6a9c NP |
2363 | #else |
2364 | static inline void watchdog(struct rq *rq, struct task_struct *p) { } | |
2365 | #endif | |
bb44e5d1 | 2366 | |
8f4d37ec | 2367 | static void task_tick_rt(struct rq *rq, struct task_struct *p, int queued) |
bb44e5d1 | 2368 | { |
454c7999 CC |
2369 | struct sched_rt_entity *rt_se = &p->rt; |
2370 | ||
67e2be02 PZ |
2371 | update_curr_rt(rq); |
2372 | ||
78f2c7db PZ |
2373 | watchdog(rq, p); |
2374 | ||
bb44e5d1 IM |
2375 | /* |
2376 | * RR tasks need a special form of timeslice management. | |
2377 | * FIFO tasks have no timeslices. | |
2378 | */ | |
2379 | if (p->policy != SCHED_RR) | |
2380 | return; | |
2381 | ||
fa717060 | 2382 | if (--p->rt.time_slice) |
bb44e5d1 IM |
2383 | return; |
2384 | ||
ce0dbbbb | 2385 | p->rt.time_slice = sched_rr_timeslice; |
bb44e5d1 | 2386 | |
98fbc798 | 2387 | /* |
e9aa39bb LB |
2388 | * Requeue to the end of queue if we (and all of our ancestors) are not |
2389 | * the only element on the queue | |
98fbc798 | 2390 | */ |
454c7999 CC |
2391 | for_each_sched_rt_entity(rt_se) { |
2392 | if (rt_se->run_list.prev != rt_se->run_list.next) { | |
2393 | requeue_task_rt(rq, p, 0); | |
8aa6f0eb | 2394 | resched_curr(rq); |
454c7999 CC |
2395 | return; |
2396 | } | |
98fbc798 | 2397 | } |
bb44e5d1 IM |
2398 | } |
2399 | ||
83b699ed SV |
2400 | static void set_curr_task_rt(struct rq *rq) |
2401 | { | |
2402 | struct task_struct *p = rq->curr; | |
2403 | ||
78becc27 | 2404 | p->se.exec_start = rq_clock_task(rq); |
917b627d GH |
2405 | |
2406 | /* The running task is never eligible for pushing */ | |
2407 | dequeue_pushable_task(rq, p); | |
83b699ed SV |
2408 | } |
2409 | ||
6d686f45 | 2410 | static unsigned int get_rr_interval_rt(struct rq *rq, struct task_struct *task) |
0d721cea PW |
2411 | { |
2412 | /* | |
2413 | * Time slice is 0 for SCHED_FIFO tasks | |
2414 | */ | |
2415 | if (task->policy == SCHED_RR) | |
ce0dbbbb | 2416 | return sched_rr_timeslice; |
0d721cea PW |
2417 | else |
2418 | return 0; | |
2419 | } | |
2420 | ||
029632fb | 2421 | const struct sched_class rt_sched_class = { |
5522d5d5 | 2422 | .next = &fair_sched_class, |
bb44e5d1 IM |
2423 | .enqueue_task = enqueue_task_rt, |
2424 | .dequeue_task = dequeue_task_rt, | |
2425 | .yield_task = yield_task_rt, | |
2426 | ||
2427 | .check_preempt_curr = check_preempt_curr_rt, | |
2428 | ||
2429 | .pick_next_task = pick_next_task_rt, | |
2430 | .put_prev_task = put_prev_task_rt, | |
2431 | ||
681f3e68 | 2432 | #ifdef CONFIG_SMP |
4ce72a2c LZ |
2433 | .select_task_rq = select_task_rq_rt, |
2434 | ||
6c37067e | 2435 | .set_cpus_allowed = set_cpus_allowed_common, |
1f11eb6a GH |
2436 | .rq_online = rq_online_rt, |
2437 | .rq_offline = rq_offline_rt, | |
efbbd05a | 2438 | .task_woken = task_woken_rt, |
cb469845 | 2439 | .switched_from = switched_from_rt, |
681f3e68 | 2440 | #endif |
bb44e5d1 | 2441 | |
83b699ed | 2442 | .set_curr_task = set_curr_task_rt, |
bb44e5d1 | 2443 | .task_tick = task_tick_rt, |
cb469845 | 2444 | |
0d721cea PW |
2445 | .get_rr_interval = get_rr_interval_rt, |
2446 | ||
cb469845 SR |
2447 | .prio_changed = prio_changed_rt, |
2448 | .switched_to = switched_to_rt, | |
6e998916 SG |
2449 | |
2450 | .update_curr = update_curr_rt, | |
bb44e5d1 | 2451 | }; |
ada18de2 | 2452 | |
8887cd99 NP |
2453 | #ifdef CONFIG_RT_GROUP_SCHED |
2454 | /* | |
2455 | * Ensure that the real time constraints are schedulable. | |
2456 | */ | |
2457 | static DEFINE_MUTEX(rt_constraints_mutex); | |
2458 | ||
2459 | /* Must be called with tasklist_lock held */ | |
2460 | static inline int tg_has_rt_tasks(struct task_group *tg) | |
2461 | { | |
2462 | struct task_struct *g, *p; | |
2463 | ||
2464 | /* | |
2465 | * Autogroups do not have RT tasks; see autogroup_create(). | |
2466 | */ | |
2467 | if (task_group_is_autogroup(tg)) | |
2468 | return 0; | |
2469 | ||
2470 | for_each_process_thread(g, p) { | |
2471 | if (rt_task(p) && task_group(p) == tg) | |
2472 | return 1; | |
2473 | } | |
2474 | ||
2475 | return 0; | |
2476 | } | |
2477 | ||
2478 | struct rt_schedulable_data { | |
2479 | struct task_group *tg; | |
2480 | u64 rt_period; | |
2481 | u64 rt_runtime; | |
2482 | }; | |
2483 | ||
2484 | static int tg_rt_schedulable(struct task_group *tg, void *data) | |
2485 | { | |
2486 | struct rt_schedulable_data *d = data; | |
2487 | struct task_group *child; | |
2488 | unsigned long total, sum = 0; | |
2489 | u64 period, runtime; | |
2490 | ||
2491 | period = ktime_to_ns(tg->rt_bandwidth.rt_period); | |
2492 | runtime = tg->rt_bandwidth.rt_runtime; | |
2493 | ||
2494 | if (tg == d->tg) { | |
2495 | period = d->rt_period; | |
2496 | runtime = d->rt_runtime; | |
2497 | } | |
2498 | ||
2499 | /* | |
2500 | * Cannot have more runtime than the period. | |
2501 | */ | |
2502 | if (runtime > period && runtime != RUNTIME_INF) | |
2503 | return -EINVAL; | |
2504 | ||
2505 | /* | |
2506 | * Ensure we don't starve existing RT tasks. | |
2507 | */ | |
2508 | if (rt_bandwidth_enabled() && !runtime && tg_has_rt_tasks(tg)) | |
2509 | return -EBUSY; | |
2510 | ||
2511 | total = to_ratio(period, runtime); | |
2512 | ||
2513 | /* | |
2514 | * Nobody can have more than the global setting allows. | |
2515 | */ | |
2516 | if (total > to_ratio(global_rt_period(), global_rt_runtime())) | |
2517 | return -EINVAL; | |
2518 | ||
2519 | /* | |
2520 | * The sum of our children's runtime should not exceed our own. | |
2521 | */ | |
2522 | list_for_each_entry_rcu(child, &tg->children, siblings) { | |
2523 | period = ktime_to_ns(child->rt_bandwidth.rt_period); | |
2524 | runtime = child->rt_bandwidth.rt_runtime; | |
2525 | ||
2526 | if (child == d->tg) { | |
2527 | period = d->rt_period; | |
2528 | runtime = d->rt_runtime; | |
2529 | } | |
2530 | ||
2531 | sum += to_ratio(period, runtime); | |
2532 | } | |
2533 | ||
2534 | if (sum > total) | |
2535 | return -EINVAL; | |
2536 | ||
2537 | return 0; | |
2538 | } | |
2539 | ||
2540 | static int __rt_schedulable(struct task_group *tg, u64 period, u64 runtime) | |
2541 | { | |
2542 | int ret; | |
2543 | ||
2544 | struct rt_schedulable_data data = { | |
2545 | .tg = tg, | |
2546 | .rt_period = period, | |
2547 | .rt_runtime = runtime, | |
2548 | }; | |
2549 | ||
2550 | rcu_read_lock(); | |
2551 | ret = walk_tg_tree(tg_rt_schedulable, tg_nop, &data); | |
2552 | rcu_read_unlock(); | |
2553 | ||
2554 | return ret; | |
2555 | } | |
2556 | ||
2557 | static int tg_set_rt_bandwidth(struct task_group *tg, | |
2558 | u64 rt_period, u64 rt_runtime) | |
2559 | { | |
2560 | int i, err = 0; | |
2561 | ||
2562 | /* | |
2563 | * Disallowing the root group RT runtime is BAD, it would disallow the | |
2564 | * kernel creating (and or operating) RT threads. | |
2565 | */ | |
2566 | if (tg == &root_task_group && rt_runtime == 0) | |
2567 | return -EINVAL; | |
2568 | ||
2569 | /* No period doesn't make any sense. */ | |
2570 | if (rt_period == 0) | |
2571 | return -EINVAL; | |
2572 | ||
2573 | mutex_lock(&rt_constraints_mutex); | |
2574 | read_lock(&tasklist_lock); | |
2575 | err = __rt_schedulable(tg, rt_period, rt_runtime); | |
2576 | if (err) | |
2577 | goto unlock; | |
2578 | ||
2579 | raw_spin_lock_irq(&tg->rt_bandwidth.rt_runtime_lock); | |
2580 | tg->rt_bandwidth.rt_period = ns_to_ktime(rt_period); | |
2581 | tg->rt_bandwidth.rt_runtime = rt_runtime; | |
2582 | ||
2583 | for_each_possible_cpu(i) { | |
2584 | struct rt_rq *rt_rq = tg->rt_rq[i]; | |
2585 | ||
2586 | raw_spin_lock(&rt_rq->rt_runtime_lock); | |
2587 | rt_rq->rt_runtime = rt_runtime; | |
2588 | raw_spin_unlock(&rt_rq->rt_runtime_lock); | |
2589 | } | |
2590 | raw_spin_unlock_irq(&tg->rt_bandwidth.rt_runtime_lock); | |
2591 | unlock: | |
2592 | read_unlock(&tasklist_lock); | |
2593 | mutex_unlock(&rt_constraints_mutex); | |
2594 | ||
2595 | return err; | |
2596 | } | |
2597 | ||
2598 | int sched_group_set_rt_runtime(struct task_group *tg, long rt_runtime_us) | |
2599 | { | |
2600 | u64 rt_runtime, rt_period; | |
2601 | ||
2602 | rt_period = ktime_to_ns(tg->rt_bandwidth.rt_period); | |
2603 | rt_runtime = (u64)rt_runtime_us * NSEC_PER_USEC; | |
2604 | if (rt_runtime_us < 0) | |
2605 | rt_runtime = RUNTIME_INF; | |
2606 | ||
2607 | return tg_set_rt_bandwidth(tg, rt_period, rt_runtime); | |
2608 | } | |
2609 | ||
2610 | long sched_group_rt_runtime(struct task_group *tg) | |
2611 | { | |
2612 | u64 rt_runtime_us; | |
2613 | ||
2614 | if (tg->rt_bandwidth.rt_runtime == RUNTIME_INF) | |
2615 | return -1; | |
2616 | ||
2617 | rt_runtime_us = tg->rt_bandwidth.rt_runtime; | |
2618 | do_div(rt_runtime_us, NSEC_PER_USEC); | |
2619 | return rt_runtime_us; | |
2620 | } | |
2621 | ||
2622 | int sched_group_set_rt_period(struct task_group *tg, u64 rt_period_us) | |
2623 | { | |
2624 | u64 rt_runtime, rt_period; | |
2625 | ||
2626 | rt_period = rt_period_us * NSEC_PER_USEC; | |
2627 | rt_runtime = tg->rt_bandwidth.rt_runtime; | |
2628 | ||
2629 | return tg_set_rt_bandwidth(tg, rt_period, rt_runtime); | |
2630 | } | |
2631 | ||
2632 | long sched_group_rt_period(struct task_group *tg) | |
2633 | { | |
2634 | u64 rt_period_us; | |
2635 | ||
2636 | rt_period_us = ktime_to_ns(tg->rt_bandwidth.rt_period); | |
2637 | do_div(rt_period_us, NSEC_PER_USEC); | |
2638 | return rt_period_us; | |
2639 | } | |
2640 | ||
2641 | static int sched_rt_global_constraints(void) | |
2642 | { | |
2643 | int ret = 0; | |
2644 | ||
2645 | mutex_lock(&rt_constraints_mutex); | |
2646 | read_lock(&tasklist_lock); | |
2647 | ret = __rt_schedulable(NULL, 0, 0); | |
2648 | read_unlock(&tasklist_lock); | |
2649 | mutex_unlock(&rt_constraints_mutex); | |
2650 | ||
2651 | return ret; | |
2652 | } | |
2653 | ||
2654 | int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk) | |
2655 | { | |
2656 | /* Don't accept realtime tasks when there is no way for them to run */ | |
2657 | if (rt_task(tsk) && tg->rt_bandwidth.rt_runtime == 0) | |
2658 | return 0; | |
2659 | ||
2660 | return 1; | |
2661 | } | |
2662 | ||
2663 | #else /* !CONFIG_RT_GROUP_SCHED */ | |
2664 | static int sched_rt_global_constraints(void) | |
2665 | { | |
2666 | unsigned long flags; | |
2667 | int i; | |
2668 | ||
2669 | raw_spin_lock_irqsave(&def_rt_bandwidth.rt_runtime_lock, flags); | |
2670 | for_each_possible_cpu(i) { | |
2671 | struct rt_rq *rt_rq = &cpu_rq(i)->rt; | |
2672 | ||
2673 | raw_spin_lock(&rt_rq->rt_runtime_lock); | |
2674 | rt_rq->rt_runtime = global_rt_runtime(); | |
2675 | raw_spin_unlock(&rt_rq->rt_runtime_lock); | |
2676 | } | |
2677 | raw_spin_unlock_irqrestore(&def_rt_bandwidth.rt_runtime_lock, flags); | |
2678 | ||
2679 | return 0; | |
2680 | } | |
2681 | #endif /* CONFIG_RT_GROUP_SCHED */ | |
2682 | ||
2683 | static int sched_rt_global_validate(void) | |
2684 | { | |
2685 | if (sysctl_sched_rt_period <= 0) | |
2686 | return -EINVAL; | |
2687 | ||
2688 | if ((sysctl_sched_rt_runtime != RUNTIME_INF) && | |
2689 | (sysctl_sched_rt_runtime > sysctl_sched_rt_period)) | |
2690 | return -EINVAL; | |
2691 | ||
2692 | return 0; | |
2693 | } | |
2694 | ||
2695 | static void sched_rt_do_global(void) | |
2696 | { | |
2697 | def_rt_bandwidth.rt_runtime = global_rt_runtime(); | |
2698 | def_rt_bandwidth.rt_period = ns_to_ktime(global_rt_period()); | |
2699 | } | |
2700 | ||
2701 | int sched_rt_handler(struct ctl_table *table, int write, | |
2702 | void __user *buffer, size_t *lenp, | |
2703 | loff_t *ppos) | |
2704 | { | |
2705 | int old_period, old_runtime; | |
2706 | static DEFINE_MUTEX(mutex); | |
2707 | int ret; | |
2708 | ||
2709 | mutex_lock(&mutex); | |
2710 | old_period = sysctl_sched_rt_period; | |
2711 | old_runtime = sysctl_sched_rt_runtime; | |
2712 | ||
2713 | ret = proc_dointvec(table, write, buffer, lenp, ppos); | |
2714 | ||
2715 | if (!ret && write) { | |
2716 | ret = sched_rt_global_validate(); | |
2717 | if (ret) | |
2718 | goto undo; | |
2719 | ||
2720 | ret = sched_dl_global_validate(); | |
2721 | if (ret) | |
2722 | goto undo; | |
2723 | ||
2724 | ret = sched_rt_global_constraints(); | |
2725 | if (ret) | |
2726 | goto undo; | |
2727 | ||
2728 | sched_rt_do_global(); | |
2729 | sched_dl_do_global(); | |
2730 | } | |
2731 | if (0) { | |
2732 | undo: | |
2733 | sysctl_sched_rt_period = old_period; | |
2734 | sysctl_sched_rt_runtime = old_runtime; | |
2735 | } | |
2736 | mutex_unlock(&mutex); | |
2737 | ||
2738 | return ret; | |
2739 | } | |
2740 | ||
2741 | int sched_rr_handler(struct ctl_table *table, int write, | |
2742 | void __user *buffer, size_t *lenp, | |
2743 | loff_t *ppos) | |
2744 | { | |
2745 | int ret; | |
2746 | static DEFINE_MUTEX(mutex); | |
2747 | ||
2748 | mutex_lock(&mutex); | |
2749 | ret = proc_dointvec(table, write, buffer, lenp, ppos); | |
2750 | /* | |
2751 | * Make sure that internally we keep jiffies. | |
2752 | * Also, writing zero resets the timeslice to default: | |
2753 | */ | |
2754 | if (!ret && write) { | |
2755 | sched_rr_timeslice = | |
2756 | sysctl_sched_rr_timeslice <= 0 ? RR_TIMESLICE : | |
2757 | msecs_to_jiffies(sysctl_sched_rr_timeslice); | |
2758 | } | |
2759 | mutex_unlock(&mutex); | |
2760 | return ret; | |
2761 | } | |
2762 | ||
ada18de2 PZ |
2763 | #ifdef CONFIG_SCHED_DEBUG |
2764 | extern void print_rt_rq(struct seq_file *m, int cpu, struct rt_rq *rt_rq); | |
2765 | ||
029632fb | 2766 | void print_rt_stats(struct seq_file *m, int cpu) |
ada18de2 | 2767 | { |
ec514c48 | 2768 | rt_rq_iter_t iter; |
ada18de2 PZ |
2769 | struct rt_rq *rt_rq; |
2770 | ||
2771 | rcu_read_lock(); | |
ec514c48 | 2772 | for_each_rt_rq(rt_rq, iter, cpu_rq(cpu)) |
ada18de2 PZ |
2773 | print_rt_rq(m, cpu, rt_rq); |
2774 | rcu_read_unlock(); | |
2775 | } | |
55e12e5e | 2776 | #endif /* CONFIG_SCHED_DEBUG */ |