Commit | Line | Data |
---|---|---|
bb44e5d1 IM |
1 | /* |
2 | * Real-Time Scheduling Class (mapped to the SCHED_FIFO and SCHED_RR | |
3 | * policies) | |
4 | */ | |
5 | ||
4fd29176 | 6 | #ifdef CONFIG_SMP |
84de4274 | 7 | |
637f5085 | 8 | static inline int rt_overloaded(struct rq *rq) |
4fd29176 | 9 | { |
637f5085 | 10 | return atomic_read(&rq->rd->rto_count); |
4fd29176 | 11 | } |
84de4274 | 12 | |
4fd29176 SR |
13 | static inline void rt_set_overload(struct rq *rq) |
14 | { | |
637f5085 | 15 | cpu_set(rq->cpu, rq->rd->rto_mask); |
4fd29176 SR |
16 | /* |
17 | * Make sure the mask is visible before we set | |
18 | * the overload count. That is checked to determine | |
19 | * if we should look at the mask. It would be a shame | |
20 | * if we looked at the mask, but the mask was not | |
21 | * updated yet. | |
22 | */ | |
23 | wmb(); | |
637f5085 | 24 | atomic_inc(&rq->rd->rto_count); |
4fd29176 | 25 | } |
84de4274 | 26 | |
4fd29176 SR |
27 | static inline void rt_clear_overload(struct rq *rq) |
28 | { | |
29 | /* the order here really doesn't matter */ | |
637f5085 GH |
30 | atomic_dec(&rq->rd->rto_count); |
31 | cpu_clear(rq->cpu, rq->rd->rto_mask); | |
4fd29176 | 32 | } |
73fe6aae GH |
33 | |
34 | static void update_rt_migration(struct rq *rq) | |
35 | { | |
637f5085 | 36 | if (rq->rt.rt_nr_migratory && (rq->rt.rt_nr_running > 1)) { |
cdc8eb98 GH |
37 | if (!rq->rt.overloaded) { |
38 | rt_set_overload(rq); | |
39 | rq->rt.overloaded = 1; | |
40 | } | |
41 | } else if (rq->rt.overloaded) { | |
73fe6aae | 42 | rt_clear_overload(rq); |
637f5085 GH |
43 | rq->rt.overloaded = 0; |
44 | } | |
73fe6aae | 45 | } |
4fd29176 SR |
46 | #endif /* CONFIG_SMP */ |
47 | ||
6f505b16 | 48 | static inline struct task_struct *rt_task_of(struct sched_rt_entity *rt_se) |
fa85ae24 | 49 | { |
6f505b16 PZ |
50 | return container_of(rt_se, struct task_struct, rt); |
51 | } | |
52 | ||
53 | static inline int on_rt_rq(struct sched_rt_entity *rt_se) | |
54 | { | |
55 | return !list_empty(&rt_se->run_list); | |
56 | } | |
57 | ||
052f1dc7 | 58 | #ifdef CONFIG_RT_GROUP_SCHED |
6f505b16 | 59 | |
9f0c1e56 | 60 | static inline u64 sched_rt_runtime(struct rt_rq *rt_rq) |
6f505b16 PZ |
61 | { |
62 | if (!rt_rq->tg) | |
9f0c1e56 | 63 | return RUNTIME_INF; |
6f505b16 | 64 | |
9f0c1e56 | 65 | return rt_rq->tg->rt_runtime; |
6f505b16 PZ |
66 | } |
67 | ||
68 | #define for_each_leaf_rt_rq(rt_rq, rq) \ | |
69 | list_for_each_entry(rt_rq, &rq->leaf_rt_rq_list, leaf_rt_rq_list) | |
70 | ||
71 | static inline struct rq *rq_of_rt_rq(struct rt_rq *rt_rq) | |
72 | { | |
73 | return rt_rq->rq; | |
74 | } | |
75 | ||
76 | static inline struct rt_rq *rt_rq_of_se(struct sched_rt_entity *rt_se) | |
77 | { | |
78 | return rt_se->rt_rq; | |
79 | } | |
80 | ||
81 | #define for_each_sched_rt_entity(rt_se) \ | |
82 | for (; rt_se; rt_se = rt_se->parent) | |
83 | ||
84 | static inline struct rt_rq *group_rt_rq(struct sched_rt_entity *rt_se) | |
85 | { | |
86 | return rt_se->my_q; | |
87 | } | |
88 | ||
89 | static void enqueue_rt_entity(struct sched_rt_entity *rt_se); | |
90 | static void dequeue_rt_entity(struct sched_rt_entity *rt_se); | |
91 | ||
9f0c1e56 | 92 | static void sched_rt_rq_enqueue(struct rt_rq *rt_rq) |
6f505b16 PZ |
93 | { |
94 | struct sched_rt_entity *rt_se = rt_rq->rt_se; | |
95 | ||
96 | if (rt_se && !on_rt_rq(rt_se) && rt_rq->rt_nr_running) { | |
1020387f PZ |
97 | struct task_struct *curr = rq_of_rt_rq(rt_rq)->curr; |
98 | ||
6f505b16 | 99 | enqueue_rt_entity(rt_se); |
1020387f PZ |
100 | if (rt_rq->highest_prio < curr->prio) |
101 | resched_task(curr); | |
6f505b16 PZ |
102 | } |
103 | } | |
104 | ||
9f0c1e56 | 105 | static void sched_rt_rq_dequeue(struct rt_rq *rt_rq) |
6f505b16 PZ |
106 | { |
107 | struct sched_rt_entity *rt_se = rt_rq->rt_se; | |
108 | ||
109 | if (rt_se && on_rt_rq(rt_se)) | |
110 | dequeue_rt_entity(rt_se); | |
111 | } | |
112 | ||
23b0fdfc PZ |
113 | static inline int rt_rq_throttled(struct rt_rq *rt_rq) |
114 | { | |
115 | return rt_rq->rt_throttled && !rt_rq->rt_nr_boosted; | |
116 | } | |
117 | ||
118 | static int rt_se_boosted(struct sched_rt_entity *rt_se) | |
119 | { | |
120 | struct rt_rq *rt_rq = group_rt_rq(rt_se); | |
121 | struct task_struct *p; | |
122 | ||
123 | if (rt_rq) | |
124 | return !!rt_rq->rt_nr_boosted; | |
125 | ||
126 | p = rt_task_of(rt_se); | |
127 | return p->prio != p->normal_prio; | |
128 | } | |
129 | ||
6f505b16 PZ |
130 | #else |
131 | ||
9f0c1e56 | 132 | static inline u64 sched_rt_runtime(struct rt_rq *rt_rq) |
6f505b16 | 133 | { |
9f0c1e56 PZ |
134 | if (sysctl_sched_rt_runtime == -1) |
135 | return RUNTIME_INF; | |
136 | ||
137 | return (u64)sysctl_sched_rt_runtime * NSEC_PER_USEC; | |
6f505b16 PZ |
138 | } |
139 | ||
140 | #define for_each_leaf_rt_rq(rt_rq, rq) \ | |
141 | for (rt_rq = &rq->rt; rt_rq; rt_rq = NULL) | |
142 | ||
143 | static inline struct rq *rq_of_rt_rq(struct rt_rq *rt_rq) | |
144 | { | |
145 | return container_of(rt_rq, struct rq, rt); | |
146 | } | |
147 | ||
148 | static inline struct rt_rq *rt_rq_of_se(struct sched_rt_entity *rt_se) | |
149 | { | |
150 | struct task_struct *p = rt_task_of(rt_se); | |
151 | struct rq *rq = task_rq(p); | |
152 | ||
153 | return &rq->rt; | |
154 | } | |
155 | ||
156 | #define for_each_sched_rt_entity(rt_se) \ | |
157 | for (; rt_se; rt_se = NULL) | |
158 | ||
159 | static inline struct rt_rq *group_rt_rq(struct sched_rt_entity *rt_se) | |
160 | { | |
161 | return NULL; | |
162 | } | |
163 | ||
9f0c1e56 | 164 | static inline void sched_rt_rq_enqueue(struct rt_rq *rt_rq) |
6f505b16 PZ |
165 | { |
166 | } | |
167 | ||
9f0c1e56 | 168 | static inline void sched_rt_rq_dequeue(struct rt_rq *rt_rq) |
6f505b16 PZ |
169 | { |
170 | } | |
171 | ||
23b0fdfc PZ |
172 | static inline int rt_rq_throttled(struct rt_rq *rt_rq) |
173 | { | |
174 | return rt_rq->rt_throttled; | |
175 | } | |
6f505b16 PZ |
176 | #endif |
177 | ||
178 | static inline int rt_se_prio(struct sched_rt_entity *rt_se) | |
179 | { | |
052f1dc7 | 180 | #ifdef CONFIG_RT_GROUP_SCHED |
6f505b16 PZ |
181 | struct rt_rq *rt_rq = group_rt_rq(rt_se); |
182 | ||
183 | if (rt_rq) | |
184 | return rt_rq->highest_prio; | |
185 | #endif | |
186 | ||
187 | return rt_task_of(rt_se)->prio; | |
188 | } | |
189 | ||
9f0c1e56 | 190 | static int sched_rt_runtime_exceeded(struct rt_rq *rt_rq) |
6f505b16 | 191 | { |
9f0c1e56 | 192 | u64 runtime = sched_rt_runtime(rt_rq); |
fa85ae24 | 193 | |
9f0c1e56 | 194 | if (runtime == RUNTIME_INF) |
fa85ae24 PZ |
195 | return 0; |
196 | ||
197 | if (rt_rq->rt_throttled) | |
23b0fdfc | 198 | return rt_rq_throttled(rt_rq); |
fa85ae24 | 199 | |
9f0c1e56 | 200 | if (rt_rq->rt_time > runtime) { |
48d5e258 PZ |
201 | struct rq *rq = rq_of_rt_rq(rt_rq); |
202 | ||
203 | rq->rt_throttled = 1; | |
6f505b16 | 204 | rt_rq->rt_throttled = 1; |
48d5e258 | 205 | |
23b0fdfc | 206 | if (rt_rq_throttled(rt_rq)) { |
9f0c1e56 | 207 | sched_rt_rq_dequeue(rt_rq); |
23b0fdfc PZ |
208 | return 1; |
209 | } | |
fa85ae24 PZ |
210 | } |
211 | ||
212 | return 0; | |
213 | } | |
214 | ||
215 | static void update_sched_rt_period(struct rq *rq) | |
216 | { | |
6f505b16 PZ |
217 | struct rt_rq *rt_rq; |
218 | u64 period; | |
fa85ae24 | 219 | |
6f505b16 | 220 | while (rq->clock > rq->rt_period_expire) { |
9f0c1e56 | 221 | period = (u64)sysctl_sched_rt_period * NSEC_PER_USEC; |
fa85ae24 | 222 | rq->rt_period_expire += period; |
fa85ae24 | 223 | |
48d5e258 | 224 | for_each_leaf_rt_rq(rt_rq, rq) { |
9f0c1e56 | 225 | u64 runtime = sched_rt_runtime(rt_rq); |
48d5e258 | 226 | |
9f0c1e56 PZ |
227 | rt_rq->rt_time -= min(rt_rq->rt_time, runtime); |
228 | if (rt_rq->rt_throttled && rt_rq->rt_time < runtime) { | |
48d5e258 | 229 | rt_rq->rt_throttled = 0; |
9f0c1e56 | 230 | sched_rt_rq_enqueue(rt_rq); |
48d5e258 PZ |
231 | } |
232 | } | |
233 | ||
234 | rq->rt_throttled = 0; | |
fa85ae24 PZ |
235 | } |
236 | } | |
237 | ||
bb44e5d1 IM |
238 | /* |
239 | * Update the current task's runtime statistics. Skip current tasks that | |
240 | * are not in our scheduling class. | |
241 | */ | |
a9957449 | 242 | static void update_curr_rt(struct rq *rq) |
bb44e5d1 IM |
243 | { |
244 | struct task_struct *curr = rq->curr; | |
6f505b16 PZ |
245 | struct sched_rt_entity *rt_se = &curr->rt; |
246 | struct rt_rq *rt_rq = rt_rq_of_se(rt_se); | |
bb44e5d1 IM |
247 | u64 delta_exec; |
248 | ||
249 | if (!task_has_rt_policy(curr)) | |
250 | return; | |
251 | ||
d281918d | 252 | delta_exec = rq->clock - curr->se.exec_start; |
bb44e5d1 IM |
253 | if (unlikely((s64)delta_exec < 0)) |
254 | delta_exec = 0; | |
6cfb0d5d IM |
255 | |
256 | schedstat_set(curr->se.exec_max, max(curr->se.exec_max, delta_exec)); | |
bb44e5d1 IM |
257 | |
258 | curr->se.sum_exec_runtime += delta_exec; | |
d281918d | 259 | curr->se.exec_start = rq->clock; |
d842de87 | 260 | cpuacct_charge(curr, delta_exec); |
fa85ae24 | 261 | |
6f505b16 | 262 | rt_rq->rt_time += delta_exec; |
9f0c1e56 | 263 | if (sched_rt_runtime_exceeded(rt_rq)) |
fa85ae24 | 264 | resched_task(curr); |
bb44e5d1 IM |
265 | } |
266 | ||
6f505b16 PZ |
267 | static inline |
268 | void inc_rt_tasks(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq) | |
63489e45 | 269 | { |
6f505b16 PZ |
270 | WARN_ON(!rt_prio(rt_se_prio(rt_se))); |
271 | rt_rq->rt_nr_running++; | |
052f1dc7 | 272 | #if defined CONFIG_SMP || defined CONFIG_RT_GROUP_SCHED |
6f505b16 PZ |
273 | if (rt_se_prio(rt_se) < rt_rq->highest_prio) |
274 | rt_rq->highest_prio = rt_se_prio(rt_se); | |
275 | #endif | |
764a9d6f | 276 | #ifdef CONFIG_SMP |
6f505b16 PZ |
277 | if (rt_se->nr_cpus_allowed > 1) { |
278 | struct rq *rq = rq_of_rt_rq(rt_rq); | |
73fe6aae | 279 | rq->rt.rt_nr_migratory++; |
6f505b16 | 280 | } |
73fe6aae | 281 | |
6f505b16 PZ |
282 | update_rt_migration(rq_of_rt_rq(rt_rq)); |
283 | #endif | |
052f1dc7 | 284 | #ifdef CONFIG_RT_GROUP_SCHED |
23b0fdfc PZ |
285 | if (rt_se_boosted(rt_se)) |
286 | rt_rq->rt_nr_boosted++; | |
287 | #endif | |
63489e45 SR |
288 | } |
289 | ||
6f505b16 PZ |
290 | static inline |
291 | void dec_rt_tasks(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq) | |
63489e45 | 292 | { |
6f505b16 PZ |
293 | WARN_ON(!rt_prio(rt_se_prio(rt_se))); |
294 | WARN_ON(!rt_rq->rt_nr_running); | |
295 | rt_rq->rt_nr_running--; | |
052f1dc7 | 296 | #if defined CONFIG_SMP || defined CONFIG_RT_GROUP_SCHED |
6f505b16 | 297 | if (rt_rq->rt_nr_running) { |
764a9d6f SR |
298 | struct rt_prio_array *array; |
299 | ||
6f505b16 PZ |
300 | WARN_ON(rt_se_prio(rt_se) < rt_rq->highest_prio); |
301 | if (rt_se_prio(rt_se) == rt_rq->highest_prio) { | |
764a9d6f | 302 | /* recalculate */ |
6f505b16 PZ |
303 | array = &rt_rq->active; |
304 | rt_rq->highest_prio = | |
764a9d6f SR |
305 | sched_find_first_bit(array->bitmap); |
306 | } /* otherwise leave rq->highest prio alone */ | |
307 | } else | |
6f505b16 PZ |
308 | rt_rq->highest_prio = MAX_RT_PRIO; |
309 | #endif | |
310 | #ifdef CONFIG_SMP | |
311 | if (rt_se->nr_cpus_allowed > 1) { | |
312 | struct rq *rq = rq_of_rt_rq(rt_rq); | |
73fe6aae | 313 | rq->rt.rt_nr_migratory--; |
6f505b16 | 314 | } |
73fe6aae | 315 | |
6f505b16 | 316 | update_rt_migration(rq_of_rt_rq(rt_rq)); |
764a9d6f | 317 | #endif /* CONFIG_SMP */ |
052f1dc7 | 318 | #ifdef CONFIG_RT_GROUP_SCHED |
23b0fdfc PZ |
319 | if (rt_se_boosted(rt_se)) |
320 | rt_rq->rt_nr_boosted--; | |
321 | ||
322 | WARN_ON(!rt_rq->rt_nr_running && rt_rq->rt_nr_boosted); | |
323 | #endif | |
63489e45 SR |
324 | } |
325 | ||
6f505b16 | 326 | static void enqueue_rt_entity(struct sched_rt_entity *rt_se) |
bb44e5d1 | 327 | { |
6f505b16 PZ |
328 | struct rt_rq *rt_rq = rt_rq_of_se(rt_se); |
329 | struct rt_prio_array *array = &rt_rq->active; | |
330 | struct rt_rq *group_rq = group_rt_rq(rt_se); | |
bb44e5d1 | 331 | |
23b0fdfc | 332 | if (group_rq && rt_rq_throttled(group_rq)) |
6f505b16 | 333 | return; |
63489e45 | 334 | |
6f505b16 PZ |
335 | list_add_tail(&rt_se->run_list, array->queue + rt_se_prio(rt_se)); |
336 | __set_bit(rt_se_prio(rt_se), array->bitmap); | |
78f2c7db | 337 | |
6f505b16 PZ |
338 | inc_rt_tasks(rt_se, rt_rq); |
339 | } | |
340 | ||
341 | static void dequeue_rt_entity(struct sched_rt_entity *rt_se) | |
342 | { | |
343 | struct rt_rq *rt_rq = rt_rq_of_se(rt_se); | |
344 | struct rt_prio_array *array = &rt_rq->active; | |
345 | ||
346 | list_del_init(&rt_se->run_list); | |
347 | if (list_empty(array->queue + rt_se_prio(rt_se))) | |
348 | __clear_bit(rt_se_prio(rt_se), array->bitmap); | |
349 | ||
350 | dec_rt_tasks(rt_se, rt_rq); | |
351 | } | |
352 | ||
353 | /* | |
354 | * Because the prio of an upper entry depends on the lower | |
355 | * entries, we must remove entries top - down. | |
356 | * | |
357 | * XXX: O(1/2 h^2) because we can only walk up, not down the chain. | |
358 | * doesn't matter much for now, as h=2 for GROUP_SCHED. | |
359 | */ | |
360 | static void dequeue_rt_stack(struct task_struct *p) | |
361 | { | |
362 | struct sched_rt_entity *rt_se, *top_se; | |
363 | ||
364 | /* | |
365 | * dequeue all, top - down. | |
366 | */ | |
367 | do { | |
368 | rt_se = &p->rt; | |
369 | top_se = NULL; | |
370 | for_each_sched_rt_entity(rt_se) { | |
371 | if (on_rt_rq(rt_se)) | |
372 | top_se = rt_se; | |
373 | } | |
374 | if (top_se) | |
375 | dequeue_rt_entity(top_se); | |
376 | } while (top_se); | |
bb44e5d1 IM |
377 | } |
378 | ||
379 | /* | |
380 | * Adding/removing a task to/from a priority array: | |
381 | */ | |
6f505b16 PZ |
382 | static void enqueue_task_rt(struct rq *rq, struct task_struct *p, int wakeup) |
383 | { | |
384 | struct sched_rt_entity *rt_se = &p->rt; | |
385 | ||
386 | if (wakeup) | |
387 | rt_se->timeout = 0; | |
388 | ||
389 | dequeue_rt_stack(p); | |
390 | ||
391 | /* | |
392 | * enqueue everybody, bottom - up. | |
393 | */ | |
394 | for_each_sched_rt_entity(rt_se) | |
395 | enqueue_rt_entity(rt_se); | |
396 | ||
397 | inc_cpu_load(rq, p->se.load.weight); | |
398 | } | |
399 | ||
f02231e5 | 400 | static void dequeue_task_rt(struct rq *rq, struct task_struct *p, int sleep) |
bb44e5d1 | 401 | { |
6f505b16 PZ |
402 | struct sched_rt_entity *rt_se = &p->rt; |
403 | struct rt_rq *rt_rq; | |
bb44e5d1 | 404 | |
f1e14ef6 | 405 | update_curr_rt(rq); |
bb44e5d1 | 406 | |
6f505b16 PZ |
407 | dequeue_rt_stack(p); |
408 | ||
409 | /* | |
410 | * re-enqueue all non-empty rt_rq entities. | |
411 | */ | |
412 | for_each_sched_rt_entity(rt_se) { | |
413 | rt_rq = group_rt_rq(rt_se); | |
414 | if (rt_rq && rt_rq->rt_nr_running) | |
415 | enqueue_rt_entity(rt_se); | |
416 | } | |
63489e45 | 417 | |
6f505b16 | 418 | dec_cpu_load(rq, p->se.load.weight); |
bb44e5d1 IM |
419 | } |
420 | ||
421 | /* | |
422 | * Put task to the end of the run list without the overhead of dequeue | |
423 | * followed by enqueue. | |
424 | */ | |
6f505b16 PZ |
425 | static |
426 | void requeue_rt_entity(struct rt_rq *rt_rq, struct sched_rt_entity *rt_se) | |
427 | { | |
428 | struct rt_prio_array *array = &rt_rq->active; | |
429 | ||
430 | list_move_tail(&rt_se->run_list, array->queue + rt_se_prio(rt_se)); | |
431 | } | |
432 | ||
bb44e5d1 IM |
433 | static void requeue_task_rt(struct rq *rq, struct task_struct *p) |
434 | { | |
6f505b16 PZ |
435 | struct sched_rt_entity *rt_se = &p->rt; |
436 | struct rt_rq *rt_rq; | |
bb44e5d1 | 437 | |
6f505b16 PZ |
438 | for_each_sched_rt_entity(rt_se) { |
439 | rt_rq = rt_rq_of_se(rt_se); | |
440 | requeue_rt_entity(rt_rq, rt_se); | |
441 | } | |
bb44e5d1 IM |
442 | } |
443 | ||
6f505b16 | 444 | static void yield_task_rt(struct rq *rq) |
bb44e5d1 | 445 | { |
4530d7ab | 446 | requeue_task_rt(rq, rq->curr); |
bb44e5d1 IM |
447 | } |
448 | ||
e7693a36 | 449 | #ifdef CONFIG_SMP |
318e0893 GH |
450 | static int find_lowest_rq(struct task_struct *task); |
451 | ||
e7693a36 GH |
452 | static int select_task_rq_rt(struct task_struct *p, int sync) |
453 | { | |
318e0893 GH |
454 | struct rq *rq = task_rq(p); |
455 | ||
456 | /* | |
e1f47d89 SR |
457 | * If the current task is an RT task, then |
458 | * try to see if we can wake this RT task up on another | |
459 | * runqueue. Otherwise simply start this RT task | |
460 | * on its current runqueue. | |
461 | * | |
462 | * We want to avoid overloading runqueues. Even if | |
463 | * the RT task is of higher priority than the current RT task. | |
464 | * RT tasks behave differently than other tasks. If | |
465 | * one gets preempted, we try to push it off to another queue. | |
466 | * So trying to keep a preempting RT task on the same | |
467 | * cache hot CPU will force the running RT task to | |
468 | * a cold CPU. So we waste all the cache for the lower | |
469 | * RT task in hopes of saving some of a RT task | |
470 | * that is just being woken and probably will have | |
471 | * cold cache anyway. | |
318e0893 | 472 | */ |
17b3279b | 473 | if (unlikely(rt_task(rq->curr)) && |
6f505b16 | 474 | (p->rt.nr_cpus_allowed > 1)) { |
318e0893 GH |
475 | int cpu = find_lowest_rq(p); |
476 | ||
477 | return (cpu == -1) ? task_cpu(p) : cpu; | |
478 | } | |
479 | ||
480 | /* | |
481 | * Otherwise, just let it ride on the affined RQ and the | |
482 | * post-schedule router will push the preempted task away | |
483 | */ | |
e7693a36 GH |
484 | return task_cpu(p); |
485 | } | |
486 | #endif /* CONFIG_SMP */ | |
487 | ||
bb44e5d1 IM |
488 | /* |
489 | * Preempt the current task with a newly woken task if needed: | |
490 | */ | |
491 | static void check_preempt_curr_rt(struct rq *rq, struct task_struct *p) | |
492 | { | |
493 | if (p->prio < rq->curr->prio) | |
494 | resched_task(rq->curr); | |
495 | } | |
496 | ||
6f505b16 PZ |
497 | static struct sched_rt_entity *pick_next_rt_entity(struct rq *rq, |
498 | struct rt_rq *rt_rq) | |
bb44e5d1 | 499 | { |
6f505b16 PZ |
500 | struct rt_prio_array *array = &rt_rq->active; |
501 | struct sched_rt_entity *next = NULL; | |
bb44e5d1 IM |
502 | struct list_head *queue; |
503 | int idx; | |
504 | ||
505 | idx = sched_find_first_bit(array->bitmap); | |
6f505b16 | 506 | BUG_ON(idx >= MAX_RT_PRIO); |
bb44e5d1 IM |
507 | |
508 | queue = array->queue + idx; | |
6f505b16 | 509 | next = list_entry(queue->next, struct sched_rt_entity, run_list); |
326587b8 | 510 | |
6f505b16 PZ |
511 | return next; |
512 | } | |
bb44e5d1 | 513 | |
6f505b16 PZ |
514 | static struct task_struct *pick_next_task_rt(struct rq *rq) |
515 | { | |
516 | struct sched_rt_entity *rt_se; | |
517 | struct task_struct *p; | |
518 | struct rt_rq *rt_rq; | |
bb44e5d1 | 519 | |
6f505b16 PZ |
520 | rt_rq = &rq->rt; |
521 | ||
522 | if (unlikely(!rt_rq->rt_nr_running)) | |
523 | return NULL; | |
524 | ||
23b0fdfc | 525 | if (rt_rq_throttled(rt_rq)) |
6f505b16 PZ |
526 | return NULL; |
527 | ||
528 | do { | |
529 | rt_se = pick_next_rt_entity(rq, rt_rq); | |
326587b8 | 530 | BUG_ON(!rt_se); |
6f505b16 PZ |
531 | rt_rq = group_rt_rq(rt_se); |
532 | } while (rt_rq); | |
533 | ||
534 | p = rt_task_of(rt_se); | |
535 | p->se.exec_start = rq->clock; | |
536 | return p; | |
bb44e5d1 IM |
537 | } |
538 | ||
31ee529c | 539 | static void put_prev_task_rt(struct rq *rq, struct task_struct *p) |
bb44e5d1 | 540 | { |
f1e14ef6 | 541 | update_curr_rt(rq); |
bb44e5d1 IM |
542 | p->se.exec_start = 0; |
543 | } | |
544 | ||
681f3e68 | 545 | #ifdef CONFIG_SMP |
6f505b16 | 546 | |
e8fa1362 SR |
547 | /* Only try algorithms three times */ |
548 | #define RT_MAX_TRIES 3 | |
549 | ||
550 | static int double_lock_balance(struct rq *this_rq, struct rq *busiest); | |
551 | static void deactivate_task(struct rq *rq, struct task_struct *p, int sleep); | |
552 | ||
f65eda4f SR |
553 | static int pick_rt_task(struct rq *rq, struct task_struct *p, int cpu) |
554 | { | |
555 | if (!task_running(rq, p) && | |
73fe6aae | 556 | (cpu < 0 || cpu_isset(cpu, p->cpus_allowed)) && |
6f505b16 | 557 | (p->rt.nr_cpus_allowed > 1)) |
f65eda4f SR |
558 | return 1; |
559 | return 0; | |
560 | } | |
561 | ||
e8fa1362 | 562 | /* Return the second highest RT task, NULL otherwise */ |
79064fbf | 563 | static struct task_struct *pick_next_highest_task_rt(struct rq *rq, int cpu) |
e8fa1362 | 564 | { |
6f505b16 PZ |
565 | struct task_struct *next = NULL; |
566 | struct sched_rt_entity *rt_se; | |
567 | struct rt_prio_array *array; | |
568 | struct rt_rq *rt_rq; | |
e8fa1362 SR |
569 | int idx; |
570 | ||
6f505b16 PZ |
571 | for_each_leaf_rt_rq(rt_rq, rq) { |
572 | array = &rt_rq->active; | |
573 | idx = sched_find_first_bit(array->bitmap); | |
574 | next_idx: | |
575 | if (idx >= MAX_RT_PRIO) | |
576 | continue; | |
577 | if (next && next->prio < idx) | |
578 | continue; | |
579 | list_for_each_entry(rt_se, array->queue + idx, run_list) { | |
580 | struct task_struct *p = rt_task_of(rt_se); | |
581 | if (pick_rt_task(rq, p, cpu)) { | |
582 | next = p; | |
583 | break; | |
584 | } | |
585 | } | |
586 | if (!next) { | |
587 | idx = find_next_bit(array->bitmap, MAX_RT_PRIO, idx+1); | |
588 | goto next_idx; | |
589 | } | |
f65eda4f SR |
590 | } |
591 | ||
e8fa1362 SR |
592 | return next; |
593 | } | |
594 | ||
595 | static DEFINE_PER_CPU(cpumask_t, local_cpu_mask); | |
596 | ||
6e1254d2 | 597 | static int find_lowest_cpus(struct task_struct *task, cpumask_t *lowest_mask) |
e8fa1362 | 598 | { |
6e1254d2 | 599 | int lowest_prio = -1; |
610bf056 | 600 | int lowest_cpu = -1; |
06f90dbd | 601 | int count = 0; |
610bf056 | 602 | int cpu; |
e8fa1362 | 603 | |
637f5085 | 604 | cpus_and(*lowest_mask, task_rq(task)->rd->online, task->cpus_allowed); |
e8fa1362 | 605 | |
07b4032c GH |
606 | /* |
607 | * Scan each rq for the lowest prio. | |
608 | */ | |
610bf056 | 609 | for_each_cpu_mask(cpu, *lowest_mask) { |
07b4032c | 610 | struct rq *rq = cpu_rq(cpu); |
e8fa1362 | 611 | |
07b4032c GH |
612 | /* We look for lowest RT prio or non-rt CPU */ |
613 | if (rq->rt.highest_prio >= MAX_RT_PRIO) { | |
610bf056 SR |
614 | /* |
615 | * if we already found a low RT queue | |
616 | * and now we found this non-rt queue | |
617 | * clear the mask and set our bit. | |
618 | * Otherwise just return the queue as is | |
619 | * and the count==1 will cause the algorithm | |
620 | * to use the first bit found. | |
621 | */ | |
622 | if (lowest_cpu != -1) { | |
6e1254d2 | 623 | cpus_clear(*lowest_mask); |
610bf056 SR |
624 | cpu_set(rq->cpu, *lowest_mask); |
625 | } | |
6e1254d2 | 626 | return 1; |
07b4032c GH |
627 | } |
628 | ||
629 | /* no locking for now */ | |
6e1254d2 GH |
630 | if ((rq->rt.highest_prio > task->prio) |
631 | && (rq->rt.highest_prio >= lowest_prio)) { | |
632 | if (rq->rt.highest_prio > lowest_prio) { | |
633 | /* new low - clear old data */ | |
634 | lowest_prio = rq->rt.highest_prio; | |
610bf056 SR |
635 | lowest_cpu = cpu; |
636 | count = 0; | |
6e1254d2 | 637 | } |
06f90dbd | 638 | count++; |
610bf056 SR |
639 | } else |
640 | cpu_clear(cpu, *lowest_mask); | |
641 | } | |
642 | ||
643 | /* | |
644 | * Clear out all the set bits that represent | |
645 | * runqueues that were of higher prio than | |
646 | * the lowest_prio. | |
647 | */ | |
648 | if (lowest_cpu > 0) { | |
649 | /* | |
650 | * Perhaps we could add another cpumask op to | |
651 | * zero out bits. Like cpu_zero_bits(cpumask, nrbits); | |
652 | * Then that could be optimized to use memset and such. | |
653 | */ | |
654 | for_each_cpu_mask(cpu, *lowest_mask) { | |
655 | if (cpu >= lowest_cpu) | |
656 | break; | |
657 | cpu_clear(cpu, *lowest_mask); | |
e8fa1362 | 658 | } |
07b4032c GH |
659 | } |
660 | ||
06f90dbd | 661 | return count; |
6e1254d2 GH |
662 | } |
663 | ||
664 | static inline int pick_optimal_cpu(int this_cpu, cpumask_t *mask) | |
665 | { | |
666 | int first; | |
667 | ||
668 | /* "this_cpu" is cheaper to preempt than a remote processor */ | |
669 | if ((this_cpu != -1) && cpu_isset(this_cpu, *mask)) | |
670 | return this_cpu; | |
671 | ||
672 | first = first_cpu(*mask); | |
673 | if (first != NR_CPUS) | |
674 | return first; | |
675 | ||
676 | return -1; | |
677 | } | |
678 | ||
679 | static int find_lowest_rq(struct task_struct *task) | |
680 | { | |
681 | struct sched_domain *sd; | |
682 | cpumask_t *lowest_mask = &__get_cpu_var(local_cpu_mask); | |
683 | int this_cpu = smp_processor_id(); | |
684 | int cpu = task_cpu(task); | |
06f90dbd GH |
685 | int count = find_lowest_cpus(task, lowest_mask); |
686 | ||
687 | if (!count) | |
688 | return -1; /* No targets found */ | |
6e1254d2 | 689 | |
06f90dbd GH |
690 | /* |
691 | * There is no sense in performing an optimal search if only one | |
692 | * target is found. | |
693 | */ | |
694 | if (count == 1) | |
695 | return first_cpu(*lowest_mask); | |
6e1254d2 GH |
696 | |
697 | /* | |
698 | * At this point we have built a mask of cpus representing the | |
699 | * lowest priority tasks in the system. Now we want to elect | |
700 | * the best one based on our affinity and topology. | |
701 | * | |
702 | * We prioritize the last cpu that the task executed on since | |
703 | * it is most likely cache-hot in that location. | |
704 | */ | |
705 | if (cpu_isset(cpu, *lowest_mask)) | |
706 | return cpu; | |
707 | ||
708 | /* | |
709 | * Otherwise, we consult the sched_domains span maps to figure | |
710 | * out which cpu is logically closest to our hot cache data. | |
711 | */ | |
712 | if (this_cpu == cpu) | |
713 | this_cpu = -1; /* Skip this_cpu opt if the same */ | |
714 | ||
715 | for_each_domain(cpu, sd) { | |
716 | if (sd->flags & SD_WAKE_AFFINE) { | |
717 | cpumask_t domain_mask; | |
718 | int best_cpu; | |
719 | ||
720 | cpus_and(domain_mask, sd->span, *lowest_mask); | |
721 | ||
722 | best_cpu = pick_optimal_cpu(this_cpu, | |
723 | &domain_mask); | |
724 | if (best_cpu != -1) | |
725 | return best_cpu; | |
726 | } | |
727 | } | |
728 | ||
729 | /* | |
730 | * And finally, if there were no matches within the domains | |
731 | * just give the caller *something* to work with from the compatible | |
732 | * locations. | |
733 | */ | |
734 | return pick_optimal_cpu(this_cpu, lowest_mask); | |
07b4032c GH |
735 | } |
736 | ||
737 | /* Will lock the rq it finds */ | |
4df64c0b | 738 | static struct rq *find_lock_lowest_rq(struct task_struct *task, struct rq *rq) |
07b4032c GH |
739 | { |
740 | struct rq *lowest_rq = NULL; | |
07b4032c | 741 | int tries; |
4df64c0b | 742 | int cpu; |
e8fa1362 | 743 | |
07b4032c GH |
744 | for (tries = 0; tries < RT_MAX_TRIES; tries++) { |
745 | cpu = find_lowest_rq(task); | |
746 | ||
2de0b463 | 747 | if ((cpu == -1) || (cpu == rq->cpu)) |
e8fa1362 SR |
748 | break; |
749 | ||
07b4032c GH |
750 | lowest_rq = cpu_rq(cpu); |
751 | ||
e8fa1362 | 752 | /* if the prio of this runqueue changed, try again */ |
07b4032c | 753 | if (double_lock_balance(rq, lowest_rq)) { |
e8fa1362 SR |
754 | /* |
755 | * We had to unlock the run queue. In | |
756 | * the mean time, task could have | |
757 | * migrated already or had its affinity changed. | |
758 | * Also make sure that it wasn't scheduled on its rq. | |
759 | */ | |
07b4032c | 760 | if (unlikely(task_rq(task) != rq || |
4df64c0b IM |
761 | !cpu_isset(lowest_rq->cpu, |
762 | task->cpus_allowed) || | |
07b4032c | 763 | task_running(rq, task) || |
e8fa1362 | 764 | !task->se.on_rq)) { |
4df64c0b | 765 | |
e8fa1362 SR |
766 | spin_unlock(&lowest_rq->lock); |
767 | lowest_rq = NULL; | |
768 | break; | |
769 | } | |
770 | } | |
771 | ||
772 | /* If this rq is still suitable use it. */ | |
773 | if (lowest_rq->rt.highest_prio > task->prio) | |
774 | break; | |
775 | ||
776 | /* try again */ | |
777 | spin_unlock(&lowest_rq->lock); | |
778 | lowest_rq = NULL; | |
779 | } | |
780 | ||
781 | return lowest_rq; | |
782 | } | |
783 | ||
784 | /* | |
785 | * If the current CPU has more than one RT task, see if the non | |
786 | * running task can migrate over to a CPU that is running a task | |
787 | * of lesser priority. | |
788 | */ | |
697f0a48 | 789 | static int push_rt_task(struct rq *rq) |
e8fa1362 SR |
790 | { |
791 | struct task_struct *next_task; | |
792 | struct rq *lowest_rq; | |
793 | int ret = 0; | |
794 | int paranoid = RT_MAX_TRIES; | |
795 | ||
a22d7fc1 GH |
796 | if (!rq->rt.overloaded) |
797 | return 0; | |
798 | ||
697f0a48 | 799 | next_task = pick_next_highest_task_rt(rq, -1); |
e8fa1362 SR |
800 | if (!next_task) |
801 | return 0; | |
802 | ||
803 | retry: | |
697f0a48 | 804 | if (unlikely(next_task == rq->curr)) { |
f65eda4f | 805 | WARN_ON(1); |
e8fa1362 | 806 | return 0; |
f65eda4f | 807 | } |
e8fa1362 SR |
808 | |
809 | /* | |
810 | * It's possible that the next_task slipped in of | |
811 | * higher priority than current. If that's the case | |
812 | * just reschedule current. | |
813 | */ | |
697f0a48 GH |
814 | if (unlikely(next_task->prio < rq->curr->prio)) { |
815 | resched_task(rq->curr); | |
e8fa1362 SR |
816 | return 0; |
817 | } | |
818 | ||
697f0a48 | 819 | /* We might release rq lock */ |
e8fa1362 SR |
820 | get_task_struct(next_task); |
821 | ||
822 | /* find_lock_lowest_rq locks the rq if found */ | |
697f0a48 | 823 | lowest_rq = find_lock_lowest_rq(next_task, rq); |
e8fa1362 SR |
824 | if (!lowest_rq) { |
825 | struct task_struct *task; | |
826 | /* | |
697f0a48 | 827 | * find lock_lowest_rq releases rq->lock |
e8fa1362 SR |
828 | * so it is possible that next_task has changed. |
829 | * If it has, then try again. | |
830 | */ | |
697f0a48 | 831 | task = pick_next_highest_task_rt(rq, -1); |
e8fa1362 SR |
832 | if (unlikely(task != next_task) && task && paranoid--) { |
833 | put_task_struct(next_task); | |
834 | next_task = task; | |
835 | goto retry; | |
836 | } | |
837 | goto out; | |
838 | } | |
839 | ||
697f0a48 | 840 | deactivate_task(rq, next_task, 0); |
e8fa1362 SR |
841 | set_task_cpu(next_task, lowest_rq->cpu); |
842 | activate_task(lowest_rq, next_task, 0); | |
843 | ||
844 | resched_task(lowest_rq->curr); | |
845 | ||
846 | spin_unlock(&lowest_rq->lock); | |
847 | ||
848 | ret = 1; | |
849 | out: | |
850 | put_task_struct(next_task); | |
851 | ||
852 | return ret; | |
853 | } | |
854 | ||
855 | /* | |
856 | * TODO: Currently we just use the second highest prio task on | |
857 | * the queue, and stop when it can't migrate (or there's | |
858 | * no more RT tasks). There may be a case where a lower | |
859 | * priority RT task has a different affinity than the | |
860 | * higher RT task. In this case the lower RT task could | |
861 | * possibly be able to migrate where as the higher priority | |
862 | * RT task could not. We currently ignore this issue. | |
863 | * Enhancements are welcome! | |
864 | */ | |
865 | static void push_rt_tasks(struct rq *rq) | |
866 | { | |
867 | /* push_rt_task will return true if it moved an RT */ | |
868 | while (push_rt_task(rq)) | |
869 | ; | |
870 | } | |
871 | ||
f65eda4f SR |
872 | static int pull_rt_task(struct rq *this_rq) |
873 | { | |
80bf3171 IM |
874 | int this_cpu = this_rq->cpu, ret = 0, cpu; |
875 | struct task_struct *p, *next; | |
f65eda4f | 876 | struct rq *src_rq; |
f65eda4f | 877 | |
637f5085 | 878 | if (likely(!rt_overloaded(this_rq))) |
f65eda4f SR |
879 | return 0; |
880 | ||
881 | next = pick_next_task_rt(this_rq); | |
882 | ||
637f5085 | 883 | for_each_cpu_mask(cpu, this_rq->rd->rto_mask) { |
f65eda4f SR |
884 | if (this_cpu == cpu) |
885 | continue; | |
886 | ||
887 | src_rq = cpu_rq(cpu); | |
f65eda4f SR |
888 | /* |
889 | * We can potentially drop this_rq's lock in | |
890 | * double_lock_balance, and another CPU could | |
891 | * steal our next task - hence we must cause | |
892 | * the caller to recalculate the next task | |
893 | * in that case: | |
894 | */ | |
895 | if (double_lock_balance(this_rq, src_rq)) { | |
896 | struct task_struct *old_next = next; | |
80bf3171 | 897 | |
f65eda4f SR |
898 | next = pick_next_task_rt(this_rq); |
899 | if (next != old_next) | |
900 | ret = 1; | |
901 | } | |
902 | ||
903 | /* | |
904 | * Are there still pullable RT tasks? | |
905 | */ | |
614ee1f6 MG |
906 | if (src_rq->rt.rt_nr_running <= 1) |
907 | goto skip; | |
f65eda4f | 908 | |
f65eda4f SR |
909 | p = pick_next_highest_task_rt(src_rq, this_cpu); |
910 | ||
911 | /* | |
912 | * Do we have an RT task that preempts | |
913 | * the to-be-scheduled task? | |
914 | */ | |
915 | if (p && (!next || (p->prio < next->prio))) { | |
916 | WARN_ON(p == src_rq->curr); | |
917 | WARN_ON(!p->se.on_rq); | |
918 | ||
919 | /* | |
920 | * There's a chance that p is higher in priority | |
921 | * than what's currently running on its cpu. | |
922 | * This is just that p is wakeing up and hasn't | |
923 | * had a chance to schedule. We only pull | |
924 | * p if it is lower in priority than the | |
925 | * current task on the run queue or | |
926 | * this_rq next task is lower in prio than | |
927 | * the current task on that rq. | |
928 | */ | |
929 | if (p->prio < src_rq->curr->prio || | |
930 | (next && next->prio < src_rq->curr->prio)) | |
614ee1f6 | 931 | goto skip; |
f65eda4f SR |
932 | |
933 | ret = 1; | |
934 | ||
935 | deactivate_task(src_rq, p, 0); | |
936 | set_task_cpu(p, this_cpu); | |
937 | activate_task(this_rq, p, 0); | |
938 | /* | |
939 | * We continue with the search, just in | |
940 | * case there's an even higher prio task | |
941 | * in another runqueue. (low likelyhood | |
942 | * but possible) | |
80bf3171 | 943 | * |
f65eda4f SR |
944 | * Update next so that we won't pick a task |
945 | * on another cpu with a priority lower (or equal) | |
946 | * than the one we just picked. | |
947 | */ | |
948 | next = p; | |
949 | ||
950 | } | |
614ee1f6 | 951 | skip: |
f65eda4f SR |
952 | spin_unlock(&src_rq->lock); |
953 | } | |
954 | ||
955 | return ret; | |
956 | } | |
957 | ||
9a897c5a | 958 | static void pre_schedule_rt(struct rq *rq, struct task_struct *prev) |
f65eda4f SR |
959 | { |
960 | /* Try to pull RT tasks here if we lower this rq's prio */ | |
7f51f298 | 961 | if (unlikely(rt_task(prev)) && rq->rt.highest_prio > prev->prio) |
f65eda4f SR |
962 | pull_rt_task(rq); |
963 | } | |
964 | ||
9a897c5a | 965 | static void post_schedule_rt(struct rq *rq) |
e8fa1362 SR |
966 | { |
967 | /* | |
968 | * If we have more than one rt_task queued, then | |
969 | * see if we can push the other rt_tasks off to other CPUS. | |
970 | * Note we may release the rq lock, and since | |
971 | * the lock was owned by prev, we need to release it | |
972 | * first via finish_lock_switch and then reaquire it here. | |
973 | */ | |
a22d7fc1 | 974 | if (unlikely(rq->rt.overloaded)) { |
e8fa1362 SR |
975 | spin_lock_irq(&rq->lock); |
976 | push_rt_tasks(rq); | |
977 | spin_unlock_irq(&rq->lock); | |
978 | } | |
979 | } | |
980 | ||
4642dafd | 981 | |
9a897c5a | 982 | static void task_wake_up_rt(struct rq *rq, struct task_struct *p) |
4642dafd | 983 | { |
9a897c5a | 984 | if (!task_running(rq, p) && |
a22d7fc1 GH |
985 | (p->prio >= rq->rt.highest_prio) && |
986 | rq->rt.overloaded) | |
4642dafd SR |
987 | push_rt_tasks(rq); |
988 | } | |
989 | ||
43010659 | 990 | static unsigned long |
bb44e5d1 | 991 | load_balance_rt(struct rq *this_rq, int this_cpu, struct rq *busiest, |
e1d1484f PW |
992 | unsigned long max_load_move, |
993 | struct sched_domain *sd, enum cpu_idle_type idle, | |
994 | int *all_pinned, int *this_best_prio) | |
bb44e5d1 | 995 | { |
c7a1e46a SR |
996 | /* don't touch RT tasks */ |
997 | return 0; | |
e1d1484f PW |
998 | } |
999 | ||
1000 | static int | |
1001 | move_one_task_rt(struct rq *this_rq, int this_cpu, struct rq *busiest, | |
1002 | struct sched_domain *sd, enum cpu_idle_type idle) | |
1003 | { | |
c7a1e46a SR |
1004 | /* don't touch RT tasks */ |
1005 | return 0; | |
bb44e5d1 | 1006 | } |
deeeccd4 | 1007 | |
73fe6aae GH |
1008 | static void set_cpus_allowed_rt(struct task_struct *p, cpumask_t *new_mask) |
1009 | { | |
1010 | int weight = cpus_weight(*new_mask); | |
1011 | ||
1012 | BUG_ON(!rt_task(p)); | |
1013 | ||
1014 | /* | |
1015 | * Update the migration status of the RQ if we have an RT task | |
1016 | * which is running AND changing its weight value. | |
1017 | */ | |
6f505b16 | 1018 | if (p->se.on_rq && (weight != p->rt.nr_cpus_allowed)) { |
73fe6aae GH |
1019 | struct rq *rq = task_rq(p); |
1020 | ||
6f505b16 | 1021 | if ((p->rt.nr_cpus_allowed <= 1) && (weight > 1)) { |
73fe6aae | 1022 | rq->rt.rt_nr_migratory++; |
6f505b16 | 1023 | } else if ((p->rt.nr_cpus_allowed > 1) && (weight <= 1)) { |
73fe6aae GH |
1024 | BUG_ON(!rq->rt.rt_nr_migratory); |
1025 | rq->rt.rt_nr_migratory--; | |
1026 | } | |
1027 | ||
1028 | update_rt_migration(rq); | |
1029 | } | |
1030 | ||
1031 | p->cpus_allowed = *new_mask; | |
6f505b16 | 1032 | p->rt.nr_cpus_allowed = weight; |
73fe6aae | 1033 | } |
deeeccd4 | 1034 | |
bdd7c81b IM |
1035 | /* Assumes rq->lock is held */ |
1036 | static void join_domain_rt(struct rq *rq) | |
1037 | { | |
1038 | if (rq->rt.overloaded) | |
1039 | rt_set_overload(rq); | |
1040 | } | |
1041 | ||
1042 | /* Assumes rq->lock is held */ | |
1043 | static void leave_domain_rt(struct rq *rq) | |
1044 | { | |
1045 | if (rq->rt.overloaded) | |
1046 | rt_clear_overload(rq); | |
1047 | } | |
cb469845 SR |
1048 | |
1049 | /* | |
1050 | * When switch from the rt queue, we bring ourselves to a position | |
1051 | * that we might want to pull RT tasks from other runqueues. | |
1052 | */ | |
1053 | static void switched_from_rt(struct rq *rq, struct task_struct *p, | |
1054 | int running) | |
1055 | { | |
1056 | /* | |
1057 | * If there are other RT tasks then we will reschedule | |
1058 | * and the scheduling of the other RT tasks will handle | |
1059 | * the balancing. But if we are the last RT task | |
1060 | * we may need to handle the pulling of RT tasks | |
1061 | * now. | |
1062 | */ | |
1063 | if (!rq->rt.rt_nr_running) | |
1064 | pull_rt_task(rq); | |
1065 | } | |
1066 | #endif /* CONFIG_SMP */ | |
1067 | ||
1068 | /* | |
1069 | * When switching a task to RT, we may overload the runqueue | |
1070 | * with RT tasks. In this case we try to push them off to | |
1071 | * other runqueues. | |
1072 | */ | |
1073 | static void switched_to_rt(struct rq *rq, struct task_struct *p, | |
1074 | int running) | |
1075 | { | |
1076 | int check_resched = 1; | |
1077 | ||
1078 | /* | |
1079 | * If we are already running, then there's nothing | |
1080 | * that needs to be done. But if we are not running | |
1081 | * we may need to preempt the current running task. | |
1082 | * If that current running task is also an RT task | |
1083 | * then see if we can move to another run queue. | |
1084 | */ | |
1085 | if (!running) { | |
1086 | #ifdef CONFIG_SMP | |
1087 | if (rq->rt.overloaded && push_rt_task(rq) && | |
1088 | /* Don't resched if we changed runqueues */ | |
1089 | rq != task_rq(p)) | |
1090 | check_resched = 0; | |
1091 | #endif /* CONFIG_SMP */ | |
1092 | if (check_resched && p->prio < rq->curr->prio) | |
1093 | resched_task(rq->curr); | |
1094 | } | |
1095 | } | |
1096 | ||
1097 | /* | |
1098 | * Priority of the task has changed. This may cause | |
1099 | * us to initiate a push or pull. | |
1100 | */ | |
1101 | static void prio_changed_rt(struct rq *rq, struct task_struct *p, | |
1102 | int oldprio, int running) | |
1103 | { | |
1104 | if (running) { | |
1105 | #ifdef CONFIG_SMP | |
1106 | /* | |
1107 | * If our priority decreases while running, we | |
1108 | * may need to pull tasks to this runqueue. | |
1109 | */ | |
1110 | if (oldprio < p->prio) | |
1111 | pull_rt_task(rq); | |
1112 | /* | |
1113 | * If there's a higher priority task waiting to run | |
1114 | * then reschedule. | |
1115 | */ | |
1116 | if (p->prio > rq->rt.highest_prio) | |
1117 | resched_task(p); | |
1118 | #else | |
1119 | /* For UP simply resched on drop of prio */ | |
1120 | if (oldprio < p->prio) | |
1121 | resched_task(p); | |
e8fa1362 | 1122 | #endif /* CONFIG_SMP */ |
cb469845 SR |
1123 | } else { |
1124 | /* | |
1125 | * This task is not running, but if it is | |
1126 | * greater than the current running task | |
1127 | * then reschedule. | |
1128 | */ | |
1129 | if (p->prio < rq->curr->prio) | |
1130 | resched_task(rq->curr); | |
1131 | } | |
1132 | } | |
1133 | ||
78f2c7db PZ |
1134 | static void watchdog(struct rq *rq, struct task_struct *p) |
1135 | { | |
1136 | unsigned long soft, hard; | |
1137 | ||
1138 | if (!p->signal) | |
1139 | return; | |
1140 | ||
1141 | soft = p->signal->rlim[RLIMIT_RTTIME].rlim_cur; | |
1142 | hard = p->signal->rlim[RLIMIT_RTTIME].rlim_max; | |
1143 | ||
1144 | if (soft != RLIM_INFINITY) { | |
1145 | unsigned long next; | |
1146 | ||
1147 | p->rt.timeout++; | |
1148 | next = DIV_ROUND_UP(min(soft, hard), USEC_PER_SEC/HZ); | |
5a52dd50 | 1149 | if (p->rt.timeout > next) |
78f2c7db PZ |
1150 | p->it_sched_expires = p->se.sum_exec_runtime; |
1151 | } | |
1152 | } | |
bb44e5d1 | 1153 | |
8f4d37ec | 1154 | static void task_tick_rt(struct rq *rq, struct task_struct *p, int queued) |
bb44e5d1 | 1155 | { |
67e2be02 PZ |
1156 | update_curr_rt(rq); |
1157 | ||
78f2c7db PZ |
1158 | watchdog(rq, p); |
1159 | ||
bb44e5d1 IM |
1160 | /* |
1161 | * RR tasks need a special form of timeslice management. | |
1162 | * FIFO tasks have no timeslices. | |
1163 | */ | |
1164 | if (p->policy != SCHED_RR) | |
1165 | return; | |
1166 | ||
fa717060 | 1167 | if (--p->rt.time_slice) |
bb44e5d1 IM |
1168 | return; |
1169 | ||
fa717060 | 1170 | p->rt.time_slice = DEF_TIMESLICE; |
bb44e5d1 | 1171 | |
98fbc798 DA |
1172 | /* |
1173 | * Requeue to the end of queue if we are not the only element | |
1174 | * on the queue: | |
1175 | */ | |
fa717060 | 1176 | if (p->rt.run_list.prev != p->rt.run_list.next) { |
98fbc798 DA |
1177 | requeue_task_rt(rq, p); |
1178 | set_tsk_need_resched(p); | |
1179 | } | |
bb44e5d1 IM |
1180 | } |
1181 | ||
83b699ed SV |
1182 | static void set_curr_task_rt(struct rq *rq) |
1183 | { | |
1184 | struct task_struct *p = rq->curr; | |
1185 | ||
1186 | p->se.exec_start = rq->clock; | |
1187 | } | |
1188 | ||
5522d5d5 IM |
1189 | const struct sched_class rt_sched_class = { |
1190 | .next = &fair_sched_class, | |
bb44e5d1 IM |
1191 | .enqueue_task = enqueue_task_rt, |
1192 | .dequeue_task = dequeue_task_rt, | |
1193 | .yield_task = yield_task_rt, | |
e7693a36 GH |
1194 | #ifdef CONFIG_SMP |
1195 | .select_task_rq = select_task_rq_rt, | |
1196 | #endif /* CONFIG_SMP */ | |
bb44e5d1 IM |
1197 | |
1198 | .check_preempt_curr = check_preempt_curr_rt, | |
1199 | ||
1200 | .pick_next_task = pick_next_task_rt, | |
1201 | .put_prev_task = put_prev_task_rt, | |
1202 | ||
681f3e68 | 1203 | #ifdef CONFIG_SMP |
bb44e5d1 | 1204 | .load_balance = load_balance_rt, |
e1d1484f | 1205 | .move_one_task = move_one_task_rt, |
73fe6aae | 1206 | .set_cpus_allowed = set_cpus_allowed_rt, |
bdd7c81b IM |
1207 | .join_domain = join_domain_rt, |
1208 | .leave_domain = leave_domain_rt, | |
9a897c5a SR |
1209 | .pre_schedule = pre_schedule_rt, |
1210 | .post_schedule = post_schedule_rt, | |
1211 | .task_wake_up = task_wake_up_rt, | |
cb469845 | 1212 | .switched_from = switched_from_rt, |
681f3e68 | 1213 | #endif |
bb44e5d1 | 1214 | |
83b699ed | 1215 | .set_curr_task = set_curr_task_rt, |
bb44e5d1 | 1216 | .task_tick = task_tick_rt, |
cb469845 SR |
1217 | |
1218 | .prio_changed = prio_changed_rt, | |
1219 | .switched_to = switched_to_rt, | |
bb44e5d1 | 1220 | }; |