Blame view
kernel/sched_rt.c
42 KB
bb44e5d1c
|
1 2 3 4 |
/* * Real-Time Scheduling Class (mapped to the SCHED_FIFO and SCHED_RR * policies) */ |
8f48894fc
|
5 6 7 |
#ifdef CONFIG_RT_GROUP_SCHED #define rt_entity_is_task(rt_se) (!(rt_se)->my_q) |
398a153b1
|
8 9 |
static inline struct task_struct *rt_task_of(struct sched_rt_entity *rt_se) { |
8f48894fc
|
10 11 12 |
#ifdef CONFIG_SCHED_DEBUG WARN_ON_ONCE(!rt_entity_is_task(rt_se)); #endif |
398a153b1
|
13 14 |
return container_of(rt_se, struct task_struct, rt); } |
398a153b1
|
15 16 17 18 19 20 21 22 23 24 25 |
static inline struct rq *rq_of_rt_rq(struct rt_rq *rt_rq) { return rt_rq->rq; } static inline struct rt_rq *rt_rq_of_se(struct sched_rt_entity *rt_se) { return rt_se->rt_rq; } #else /* CONFIG_RT_GROUP_SCHED */ |
a1ba4d8ba
|
26 |
#define rt_entity_is_task(rt_se) (1) |
8f48894fc
|
27 28 29 30 |
static inline struct task_struct *rt_task_of(struct sched_rt_entity *rt_se) { return container_of(rt_se, struct task_struct, rt); } |
398a153b1
|
31 32 33 34 35 36 37 38 39 40 41 42 43 44 |
static inline struct rq *rq_of_rt_rq(struct rt_rq *rt_rq) { return container_of(rt_rq, struct rq, rt); } static inline struct rt_rq *rt_rq_of_se(struct sched_rt_entity *rt_se) { struct task_struct *p = rt_task_of(rt_se); struct rq *rq = task_rq(p); return &rq->rt; } #endif /* CONFIG_RT_GROUP_SCHED */ |
4fd29176b
|
45 |
#ifdef CONFIG_SMP |
84de42748
|
46 |
|
637f50851
|
47 |
static inline int rt_overloaded(struct rq *rq) |
4fd29176b
|
48 |
{ |
637f50851
|
49 |
return atomic_read(&rq->rd->rto_count); |
4fd29176b
|
50 |
} |
84de42748
|
51 |
|
4fd29176b
|
52 53 |
static inline void rt_set_overload(struct rq *rq) { |
1f11eb6a8
|
54 55 |
if (!rq->online) return; |
c6c4927b2
|
56 |
cpumask_set_cpu(rq->cpu, rq->rd->rto_mask); |
4fd29176b
|
57 58 59 60 61 62 63 64 |
/* * Make sure the mask is visible before we set * the overload count. That is checked to determine * if we should look at the mask. It would be a shame * if we looked at the mask, but the mask was not * updated yet. */ wmb(); |
637f50851
|
65 |
atomic_inc(&rq->rd->rto_count); |
4fd29176b
|
66 |
} |
84de42748
|
67 |
|
4fd29176b
|
68 69 |
static inline void rt_clear_overload(struct rq *rq) { |
1f11eb6a8
|
70 71 |
if (!rq->online) return; |
4fd29176b
|
72 |
/* the order here really doesn't matter */ |
637f50851
|
73 |
atomic_dec(&rq->rd->rto_count); |
c6c4927b2
|
74 |
cpumask_clear_cpu(rq->cpu, rq->rd->rto_mask); |
4fd29176b
|
75 |
} |
73fe6aae8
|
76 |
|
398a153b1
|
77 |
static void update_rt_migration(struct rt_rq *rt_rq) |
73fe6aae8
|
78 |
{ |
a1ba4d8ba
|
79 |
if (rt_rq->rt_nr_migratory && rt_rq->rt_nr_total > 1) { |
398a153b1
|
80 81 82 |
if (!rt_rq->overloaded) { rt_set_overload(rq_of_rt_rq(rt_rq)); rt_rq->overloaded = 1; |
cdc8eb984
|
83 |
} |
398a153b1
|
84 85 86 |
} else if (rt_rq->overloaded) { rt_clear_overload(rq_of_rt_rq(rt_rq)); rt_rq->overloaded = 0; |
637f50851
|
87 |
} |
73fe6aae8
|
88 |
} |
4fd29176b
|
89 |
|
398a153b1
|
90 91 |
static void inc_rt_migration(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq) { |
a1ba4d8ba
|
92 93 94 95 96 97 |
if (!rt_entity_is_task(rt_se)) return; rt_rq = &rq_of_rt_rq(rt_rq)->rt; rt_rq->rt_nr_total++; |
398a153b1
|
98 99 100 101 102 103 104 105 |
if (rt_se->nr_cpus_allowed > 1) rt_rq->rt_nr_migratory++; update_rt_migration(rt_rq); } static void dec_rt_migration(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq) { |
a1ba4d8ba
|
106 107 108 109 110 111 |
if (!rt_entity_is_task(rt_se)) return; rt_rq = &rq_of_rt_rq(rt_rq)->rt; rt_rq->rt_nr_total--; |
398a153b1
|
112 113 114 115 116 |
if (rt_se->nr_cpus_allowed > 1) rt_rq->rt_nr_migratory--; update_rt_migration(rt_rq); } |
917b627d4
|
117 118 119 120 121 122 123 124 125 126 127 |
static void enqueue_pushable_task(struct rq *rq, struct task_struct *p) { plist_del(&p->pushable_tasks, &rq->rt.pushable_tasks); plist_node_init(&p->pushable_tasks, p->prio); plist_add(&p->pushable_tasks, &rq->rt.pushable_tasks); } static void dequeue_pushable_task(struct rq *rq, struct task_struct *p) { plist_del(&p->pushable_tasks, &rq->rt.pushable_tasks); } |
bcf08df3b
|
128 129 130 131 |
static inline int has_pushable_tasks(struct rq *rq) { return !plist_head_empty(&rq->rt.pushable_tasks); } |
917b627d4
|
132 |
#else |
ceacc2c1c
|
133 |
static inline void enqueue_pushable_task(struct rq *rq, struct task_struct *p) |
fa85ae241
|
134 |
{ |
6f505b164
|
135 |
} |
ceacc2c1c
|
136 137 138 |
static inline void dequeue_pushable_task(struct rq *rq, struct task_struct *p) { } |
b07430ac3
|
139 |
static inline |
ceacc2c1c
|
140 141 142 |
void inc_rt_migration(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq) { } |
398a153b1
|
143 |
static inline |
ceacc2c1c
|
144 145 146 |
void dec_rt_migration(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq) { } |
917b627d4
|
147 |
|
4fd29176b
|
148 |
#endif /* CONFIG_SMP */ |
6f505b164
|
149 150 151 152 |
static inline int on_rt_rq(struct sched_rt_entity *rt_se) { return !list_empty(&rt_se->run_list); } |
052f1dc7e
|
153 |
#ifdef CONFIG_RT_GROUP_SCHED |
6f505b164
|
154 |
|
9f0c1e560
|
155 |
static inline u64 sched_rt_runtime(struct rt_rq *rt_rq) |
6f505b164
|
156 157 |
{ if (!rt_rq->tg) |
9f0c1e560
|
158 |
return RUNTIME_INF; |
6f505b164
|
159 |
|
ac086bc22
|
160 161 162 163 164 165 |
return rt_rq->rt_runtime; } static inline u64 sched_rt_period(struct rt_rq *rt_rq) { return ktime_to_ns(rt_rq->tg->rt_bandwidth.rt_period); |
6f505b164
|
166 |
} |
ec514c487
|
167 168 169 170 171 172 173 |
typedef struct task_group *rt_rq_iter_t; #define for_each_rt_rq(rt_rq, iter, rq) \ for (iter = list_entry_rcu(task_groups.next, typeof(*iter), list); \ (&iter->list != &task_groups) && \ (rt_rq = iter->rt_rq[cpu_of(rq)]); \ iter = list_entry_rcu(iter->list.next, typeof(*iter), list)) |
3d4b47b4b
|
174 175 176 177 178 179 180 181 182 183 |
static inline void list_add_leaf_rt_rq(struct rt_rq *rt_rq) { list_add_rcu(&rt_rq->leaf_rt_rq_list, &rq_of_rt_rq(rt_rq)->leaf_rt_rq_list); } static inline void list_del_leaf_rt_rq(struct rt_rq *rt_rq) { list_del_rcu(&rt_rq->leaf_rt_rq_list); } |
6f505b164
|
184 |
#define for_each_leaf_rt_rq(rt_rq, rq) \ |
80f40ee4a
|
185 |
list_for_each_entry_rcu(rt_rq, &rq->leaf_rt_rq_list, leaf_rt_rq_list) |
6f505b164
|
186 |
|
6f505b164
|
187 188 189 190 191 192 193 |
#define for_each_sched_rt_entity(rt_se) \ for (; rt_se; rt_se = rt_se->parent) static inline struct rt_rq *group_rt_rq(struct sched_rt_entity *rt_se) { return rt_se->my_q; } |
37dad3fce
|
194 |
static void enqueue_rt_entity(struct sched_rt_entity *rt_se, bool head); |
6f505b164
|
195 |
static void dequeue_rt_entity(struct sched_rt_entity *rt_se); |
9f0c1e560
|
196 |
static void sched_rt_rq_enqueue(struct rt_rq *rt_rq) |
6f505b164
|
197 |
{ |
f6121f4f8
|
198 |
struct task_struct *curr = rq_of_rt_rq(rt_rq)->curr; |
74b7eb588
|
199 |
struct sched_rt_entity *rt_se; |
0c3b91680
|
200 201 202 |
int cpu = cpu_of(rq_of_rt_rq(rt_rq)); rt_se = rt_rq->tg->rt_se[cpu]; |
6f505b164
|
203 |
|
f6121f4f8
|
204 205 |
if (rt_rq->rt_nr_running) { if (rt_se && !on_rt_rq(rt_se)) |
37dad3fce
|
206 |
enqueue_rt_entity(rt_se, false); |
e864c499d
|
207 |
if (rt_rq->highest_prio.curr < curr->prio) |
1020387f5
|
208 |
resched_task(curr); |
6f505b164
|
209 210 |
} } |
9f0c1e560
|
211 |
static void sched_rt_rq_dequeue(struct rt_rq *rt_rq) |
6f505b164
|
212 |
{ |
74b7eb588
|
213 |
struct sched_rt_entity *rt_se; |
0c3b91680
|
214 |
int cpu = cpu_of(rq_of_rt_rq(rt_rq)); |
74b7eb588
|
215 |
|
0c3b91680
|
216 |
rt_se = rt_rq->tg->rt_se[cpu]; |
6f505b164
|
217 218 219 220 |
if (rt_se && on_rt_rq(rt_se)) dequeue_rt_entity(rt_se); } |
23b0fdfc9
|
221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 |
static inline int rt_rq_throttled(struct rt_rq *rt_rq) { return rt_rq->rt_throttled && !rt_rq->rt_nr_boosted; } static int rt_se_boosted(struct sched_rt_entity *rt_se) { struct rt_rq *rt_rq = group_rt_rq(rt_se); struct task_struct *p; if (rt_rq) return !!rt_rq->rt_nr_boosted; p = rt_task_of(rt_se); return p->prio != p->normal_prio; } |
d0b27fa77
|
237 |
#ifdef CONFIG_SMP |
c6c4927b2
|
238 |
static inline const struct cpumask *sched_rt_period_mask(void) |
d0b27fa77
|
239 240 241 |
{ return cpu_rq(smp_processor_id())->rd->span; } |
6f505b164
|
242 |
#else |
c6c4927b2
|
243 |
static inline const struct cpumask *sched_rt_period_mask(void) |
d0b27fa77
|
244 |
{ |
c6c4927b2
|
245 |
return cpu_online_mask; |
d0b27fa77
|
246 247 |
} #endif |
6f505b164
|
248 |
|
d0b27fa77
|
249 250 |
static inline struct rt_rq *sched_rt_period_rt_rq(struct rt_bandwidth *rt_b, int cpu) |
6f505b164
|
251 |
{ |
d0b27fa77
|
252 253 |
return container_of(rt_b, struct task_group, rt_bandwidth)->rt_rq[cpu]; } |
9f0c1e560
|
254 |
|
ac086bc22
|
255 256 257 258 |
static inline struct rt_bandwidth *sched_rt_bandwidth(struct rt_rq *rt_rq) { return &rt_rq->tg->rt_bandwidth; } |
55e12e5e7
|
259 |
#else /* !CONFIG_RT_GROUP_SCHED */ |
d0b27fa77
|
260 261 262 |
static inline u64 sched_rt_runtime(struct rt_rq *rt_rq) { |
ac086bc22
|
263 264 265 266 267 268 |
return rt_rq->rt_runtime; } static inline u64 sched_rt_period(struct rt_rq *rt_rq) { return ktime_to_ns(def_rt_bandwidth.rt_period); |
6f505b164
|
269 |
} |
ec514c487
|
270 271 272 273 |
typedef struct rt_rq *rt_rq_iter_t; #define for_each_rt_rq(rt_rq, iter, rq) \ for ((void) iter, rt_rq = &rq->rt; rt_rq; rt_rq = NULL) |
3d4b47b4b
|
274 275 276 277 278 279 280 |
static inline void list_add_leaf_rt_rq(struct rt_rq *rt_rq) { } static inline void list_del_leaf_rt_rq(struct rt_rq *rt_rq) { } |
6f505b164
|
281 282 |
#define for_each_leaf_rt_rq(rt_rq, rq) \ for (rt_rq = &rq->rt; rt_rq; rt_rq = NULL) |
6f505b164
|
283 284 285 286 287 288 289 |
#define for_each_sched_rt_entity(rt_se) \ for (; rt_se; rt_se = NULL) static inline struct rt_rq *group_rt_rq(struct sched_rt_entity *rt_se) { return NULL; } |
9f0c1e560
|
290 |
static inline void sched_rt_rq_enqueue(struct rt_rq *rt_rq) |
6f505b164
|
291 |
{ |
f3ade8378
|
292 293 |
if (rt_rq->rt_nr_running) resched_task(rq_of_rt_rq(rt_rq)->curr); |
6f505b164
|
294 |
} |
9f0c1e560
|
295 |
static inline void sched_rt_rq_dequeue(struct rt_rq *rt_rq) |
6f505b164
|
296 297 |
{ } |
23b0fdfc9
|
298 299 300 301 |
static inline int rt_rq_throttled(struct rt_rq *rt_rq) { return rt_rq->rt_throttled; } |
d0b27fa77
|
302 |
|
c6c4927b2
|
303 |
static inline const struct cpumask *sched_rt_period_mask(void) |
d0b27fa77
|
304 |
{ |
c6c4927b2
|
305 |
return cpu_online_mask; |
d0b27fa77
|
306 307 308 309 310 311 312 |
} static inline struct rt_rq *sched_rt_period_rt_rq(struct rt_bandwidth *rt_b, int cpu) { return &cpu_rq(cpu)->rt; } |
ac086bc22
|
313 314 315 316 |
static inline struct rt_bandwidth *sched_rt_bandwidth(struct rt_rq *rt_rq) { return &def_rt_bandwidth; } |
55e12e5e7
|
317 |
#endif /* CONFIG_RT_GROUP_SCHED */ |
d0b27fa77
|
318 |
|
ac086bc22
|
319 |
#ifdef CONFIG_SMP |
78333cdd0
|
320 321 322 |
/* * We ran out of runtime, see if we can borrow some from our neighbours. */ |
b79f3833d
|
323 |
static int do_balance_runtime(struct rt_rq *rt_rq) |
ac086bc22
|
324 325 326 327 328 |
{ struct rt_bandwidth *rt_b = sched_rt_bandwidth(rt_rq); struct root_domain *rd = cpu_rq(smp_processor_id())->rd; int i, weight, more = 0; u64 rt_period; |
c6c4927b2
|
329 |
weight = cpumask_weight(rd->span); |
ac086bc22
|
330 |
|
0986b11b1
|
331 |
raw_spin_lock(&rt_b->rt_runtime_lock); |
ac086bc22
|
332 |
rt_period = ktime_to_ns(rt_b->rt_period); |
c6c4927b2
|
333 |
for_each_cpu(i, rd->span) { |
ac086bc22
|
334 335 336 337 338 |
struct rt_rq *iter = sched_rt_period_rt_rq(rt_b, i); s64 diff; if (iter == rt_rq) continue; |
0986b11b1
|
339 |
raw_spin_lock(&iter->rt_runtime_lock); |
78333cdd0
|
340 341 342 343 344 |
/* * Either all rqs have inf runtime and there's nothing to steal * or __disable_runtime() below sets a specific rq to inf to * indicate its been disabled and disalow stealing. */ |
7def2be1d
|
345 346 |
if (iter->rt_runtime == RUNTIME_INF) goto next; |
78333cdd0
|
347 348 349 350 |
/* * From runqueues with spare time, take 1/n part of their * spare time, but no more than our period. */ |
ac086bc22
|
351 352 |
diff = iter->rt_runtime - iter->rt_time; if (diff > 0) { |
58838cf3c
|
353 |
diff = div_u64((u64)diff, weight); |
ac086bc22
|
354 355 356 357 358 359 |
if (rt_rq->rt_runtime + diff > rt_period) diff = rt_period - rt_rq->rt_runtime; iter->rt_runtime -= diff; rt_rq->rt_runtime += diff; more = 1; if (rt_rq->rt_runtime == rt_period) { |
0986b11b1
|
360 |
raw_spin_unlock(&iter->rt_runtime_lock); |
ac086bc22
|
361 362 363 |
break; } } |
7def2be1d
|
364 |
next: |
0986b11b1
|
365 |
raw_spin_unlock(&iter->rt_runtime_lock); |
ac086bc22
|
366 |
} |
0986b11b1
|
367 |
raw_spin_unlock(&rt_b->rt_runtime_lock); |
ac086bc22
|
368 369 370 |
return more; } |
7def2be1d
|
371 |
|
78333cdd0
|
372 373 374 |
/* * Ensure this RQ takes back all the runtime it lend to its neighbours. */ |
7def2be1d
|
375 376 377 |
static void __disable_runtime(struct rq *rq) { struct root_domain *rd = rq->rd; |
ec514c487
|
378 |
rt_rq_iter_t iter; |
7def2be1d
|
379 380 381 382 |
struct rt_rq *rt_rq; if (unlikely(!scheduler_running)) return; |
ec514c487
|
383 |
for_each_rt_rq(rt_rq, iter, rq) { |
7def2be1d
|
384 385 386 |
struct rt_bandwidth *rt_b = sched_rt_bandwidth(rt_rq); s64 want; int i; |
0986b11b1
|
387 388 |
raw_spin_lock(&rt_b->rt_runtime_lock); raw_spin_lock(&rt_rq->rt_runtime_lock); |
78333cdd0
|
389 390 391 392 393 |
/* * Either we're all inf and nobody needs to borrow, or we're * already disabled and thus have nothing to do, or we have * exactly the right amount of runtime to take out. */ |
7def2be1d
|
394 395 396 |
if (rt_rq->rt_runtime == RUNTIME_INF || rt_rq->rt_runtime == rt_b->rt_runtime) goto balanced; |
0986b11b1
|
397 |
raw_spin_unlock(&rt_rq->rt_runtime_lock); |
7def2be1d
|
398 |
|
78333cdd0
|
399 400 401 402 403 |
/* * Calculate the difference between what we started out with * and what we current have, that's the amount of runtime * we lend and now have to reclaim. */ |
7def2be1d
|
404 |
want = rt_b->rt_runtime - rt_rq->rt_runtime; |
78333cdd0
|
405 406 407 |
/* * Greedy reclaim, take back as much as we can. */ |
c6c4927b2
|
408 |
for_each_cpu(i, rd->span) { |
7def2be1d
|
409 410 |
struct rt_rq *iter = sched_rt_period_rt_rq(rt_b, i); s64 diff; |
78333cdd0
|
411 412 413 |
/* * Can't reclaim from ourselves or disabled runqueues. */ |
f1679d084
|
414 |
if (iter == rt_rq || iter->rt_runtime == RUNTIME_INF) |
7def2be1d
|
415 |
continue; |
0986b11b1
|
416 |
raw_spin_lock(&iter->rt_runtime_lock); |
7def2be1d
|
417 418 419 420 421 422 423 424 |
if (want > 0) { diff = min_t(s64, iter->rt_runtime, want); iter->rt_runtime -= diff; want -= diff; } else { iter->rt_runtime -= want; want -= want; } |
0986b11b1
|
425 |
raw_spin_unlock(&iter->rt_runtime_lock); |
7def2be1d
|
426 427 428 429 |
if (!want) break; } |
0986b11b1
|
430 |
raw_spin_lock(&rt_rq->rt_runtime_lock); |
78333cdd0
|
431 432 433 434 |
/* * We cannot be left wanting - that would mean some runtime * leaked out of the system. */ |
7def2be1d
|
435 436 |
BUG_ON(want); balanced: |
78333cdd0
|
437 438 439 440 |
/* * Disable all the borrow logic by pretending we have inf * runtime - in which case borrowing doesn't make sense. */ |
7def2be1d
|
441 |
rt_rq->rt_runtime = RUNTIME_INF; |
0986b11b1
|
442 443 |
raw_spin_unlock(&rt_rq->rt_runtime_lock); raw_spin_unlock(&rt_b->rt_runtime_lock); |
7def2be1d
|
444 445 446 447 448 449 |
} } static void disable_runtime(struct rq *rq) { unsigned long flags; |
05fa785cf
|
450 |
raw_spin_lock_irqsave(&rq->lock, flags); |
7def2be1d
|
451 |
__disable_runtime(rq); |
05fa785cf
|
452 |
raw_spin_unlock_irqrestore(&rq->lock, flags); |
7def2be1d
|
453 454 455 456 |
} static void __enable_runtime(struct rq *rq) { |
ec514c487
|
457 |
rt_rq_iter_t iter; |
7def2be1d
|
458 459 460 461 |
struct rt_rq *rt_rq; if (unlikely(!scheduler_running)) return; |
78333cdd0
|
462 463 464 |
/* * Reset each runqueue's bandwidth settings */ |
ec514c487
|
465 |
for_each_rt_rq(rt_rq, iter, rq) { |
7def2be1d
|
466 |
struct rt_bandwidth *rt_b = sched_rt_bandwidth(rt_rq); |
0986b11b1
|
467 468 |
raw_spin_lock(&rt_b->rt_runtime_lock); raw_spin_lock(&rt_rq->rt_runtime_lock); |
7def2be1d
|
469 470 |
rt_rq->rt_runtime = rt_b->rt_runtime; rt_rq->rt_time = 0; |
baf25731e
|
471 |
rt_rq->rt_throttled = 0; |
0986b11b1
|
472 473 |
raw_spin_unlock(&rt_rq->rt_runtime_lock); raw_spin_unlock(&rt_b->rt_runtime_lock); |
7def2be1d
|
474 475 476 477 478 479 |
} } static void enable_runtime(struct rq *rq) { unsigned long flags; |
05fa785cf
|
480 |
raw_spin_lock_irqsave(&rq->lock, flags); |
7def2be1d
|
481 |
__enable_runtime(rq); |
05fa785cf
|
482 |
raw_spin_unlock_irqrestore(&rq->lock, flags); |
7def2be1d
|
483 |
} |
eff6549b9
|
484 485 486 487 488 |
static int balance_runtime(struct rt_rq *rt_rq) { int more = 0; if (rt_rq->rt_time > rt_rq->rt_runtime) { |
0986b11b1
|
489 |
raw_spin_unlock(&rt_rq->rt_runtime_lock); |
eff6549b9
|
490 |
more = do_balance_runtime(rt_rq); |
0986b11b1
|
491 |
raw_spin_lock(&rt_rq->rt_runtime_lock); |
eff6549b9
|
492 493 494 495 |
} return more; } |
55e12e5e7
|
496 |
#else /* !CONFIG_SMP */ |
eff6549b9
|
497 498 499 500 |
static inline int balance_runtime(struct rt_rq *rt_rq) { return 0; } |
55e12e5e7
|
501 |
#endif /* CONFIG_SMP */ |
ac086bc22
|
502 |
|
eff6549b9
|
503 504 505 |
static int do_sched_rt_period_timer(struct rt_bandwidth *rt_b, int overrun) { int i, idle = 1; |
c6c4927b2
|
506 |
const struct cpumask *span; |
eff6549b9
|
507 |
|
0b148fa04
|
508 |
if (!rt_bandwidth_enabled() || rt_b->rt_runtime == RUNTIME_INF) |
eff6549b9
|
509 510 511 |
return 1; span = sched_rt_period_mask(); |
c6c4927b2
|
512 |
for_each_cpu(i, span) { |
eff6549b9
|
513 514 515 |
int enqueue = 0; struct rt_rq *rt_rq = sched_rt_period_rt_rq(rt_b, i); struct rq *rq = rq_of_rt_rq(rt_rq); |
05fa785cf
|
516 |
raw_spin_lock(&rq->lock); |
eff6549b9
|
517 518 |
if (rt_rq->rt_time) { u64 runtime; |
0986b11b1
|
519 |
raw_spin_lock(&rt_rq->rt_runtime_lock); |
eff6549b9
|
520 521 522 523 524 525 526 |
if (rt_rq->rt_throttled) balance_runtime(rt_rq); runtime = rt_rq->rt_runtime; rt_rq->rt_time -= min(rt_rq->rt_time, overrun*runtime); if (rt_rq->rt_throttled && rt_rq->rt_time < runtime) { rt_rq->rt_throttled = 0; enqueue = 1; |
61eadef6a
|
527 528 529 530 531 532 533 |
/* * Force a clock update if the CPU was idle, * lest wakeup -> unthrottle time accumulate. */ if (rt_rq->rt_nr_running && rq->curr == rq->idle) rq->skip_clock_update = -1; |
eff6549b9
|
534 535 536 |
} if (rt_rq->rt_time || rt_rq->rt_nr_running) idle = 0; |
0986b11b1
|
537 |
raw_spin_unlock(&rt_rq->rt_runtime_lock); |
0c3b91680
|
538 |
} else if (rt_rq->rt_nr_running) { |
6c3df2551
|
539 |
idle = 0; |
0c3b91680
|
540 541 542 |
if (!rt_rq_throttled(rt_rq)) enqueue = 1; } |
eff6549b9
|
543 544 545 |
if (enqueue) sched_rt_rq_enqueue(rt_rq); |
05fa785cf
|
546 |
raw_spin_unlock(&rq->lock); |
eff6549b9
|
547 548 549 550 |
} return idle; } |
ac086bc22
|
551 |
|
6f505b164
|
552 553 |
static inline int rt_se_prio(struct sched_rt_entity *rt_se) { |
052f1dc7e
|
554 |
#ifdef CONFIG_RT_GROUP_SCHED |
6f505b164
|
555 556 557 |
struct rt_rq *rt_rq = group_rt_rq(rt_se); if (rt_rq) |
e864c499d
|
558 |
return rt_rq->highest_prio.curr; |
6f505b164
|
559 560 561 562 |
#endif return rt_task_of(rt_se)->prio; } |
9f0c1e560
|
563 |
static int sched_rt_runtime_exceeded(struct rt_rq *rt_rq) |
6f505b164
|
564 |
{ |
9f0c1e560
|
565 |
u64 runtime = sched_rt_runtime(rt_rq); |
fa85ae241
|
566 |
|
fa85ae241
|
567 |
if (rt_rq->rt_throttled) |
23b0fdfc9
|
568 |
return rt_rq_throttled(rt_rq); |
fa85ae241
|
569 |
|
ac086bc22
|
570 571 |
if (sched_rt_runtime(rt_rq) >= sched_rt_period(rt_rq)) return 0; |
b79f3833d
|
572 573 574 575 |
balance_runtime(rt_rq); runtime = sched_rt_runtime(rt_rq); if (runtime == RUNTIME_INF) return 0; |
ac086bc22
|
576 |
|
9f0c1e560
|
577 |
if (rt_rq->rt_time > runtime) { |
6f505b164
|
578 |
rt_rq->rt_throttled = 1; |
23b0fdfc9
|
579 |
if (rt_rq_throttled(rt_rq)) { |
9f0c1e560
|
580 |
sched_rt_rq_dequeue(rt_rq); |
23b0fdfc9
|
581 582 |
return 1; } |
fa85ae241
|
583 584 585 586 |
} return 0; } |
bb44e5d1c
|
587 588 589 590 |
/* * Update the current task's runtime statistics. Skip current tasks that * are not in our scheduling class. */ |
a9957449b
|
591 |
static void update_curr_rt(struct rq *rq) |
bb44e5d1c
|
592 593 |
{ struct task_struct *curr = rq->curr; |
6f505b164
|
594 595 |
struct sched_rt_entity *rt_se = &curr->rt; struct rt_rq *rt_rq = rt_rq_of_se(rt_se); |
bb44e5d1c
|
596 |
u64 delta_exec; |
06c3bc655
|
597 |
if (curr->sched_class != &rt_sched_class) |
bb44e5d1c
|
598 |
return; |
305e6835e
|
599 |
delta_exec = rq->clock_task - curr->se.exec_start; |
bb44e5d1c
|
600 601 |
if (unlikely((s64)delta_exec < 0)) delta_exec = 0; |
6cfb0d5d0
|
602 |
|
41acab885
|
603 |
schedstat_set(curr->se.statistics.exec_max, max(curr->se.statistics.exec_max, delta_exec)); |
bb44e5d1c
|
604 605 |
curr->se.sum_exec_runtime += delta_exec; |
f06febc96
|
606 |
account_group_exec_runtime(curr, delta_exec); |
305e6835e
|
607 |
curr->se.exec_start = rq->clock_task; |
d842de871
|
608 |
cpuacct_charge(curr, delta_exec); |
fa85ae241
|
609 |
|
e9e9250bc
|
610 |
sched_rt_avg_update(rq, delta_exec); |
0b148fa04
|
611 612 |
if (!rt_bandwidth_enabled()) return; |
354d60c2f
|
613 614 |
for_each_sched_rt_entity(rt_se) { rt_rq = rt_rq_of_se(rt_se); |
cc2991cf1
|
615 |
if (sched_rt_runtime(rt_rq) != RUNTIME_INF) { |
0986b11b1
|
616 |
raw_spin_lock(&rt_rq->rt_runtime_lock); |
cc2991cf1
|
617 618 619 |
rt_rq->rt_time += delta_exec; if (sched_rt_runtime_exceeded(rt_rq)) resched_task(curr); |
0986b11b1
|
620 |
raw_spin_unlock(&rt_rq->rt_runtime_lock); |
cc2991cf1
|
621 |
} |
354d60c2f
|
622 |
} |
bb44e5d1c
|
623 |
} |
398a153b1
|
624 |
#if defined CONFIG_SMP |
e864c499d
|
625 626 627 628 |
static struct task_struct *pick_next_highest_task_rt(struct rq *rq, int cpu); static inline int next_prio(struct rq *rq) |
63489e45e
|
629 |
{ |
e864c499d
|
630 631 632 633 634 635 636 |
struct task_struct *next = pick_next_highest_task_rt(rq, rq->cpu); if (next && rt_prio(next->prio)) return next->prio; else return MAX_RT_PRIO; } |
e864c499d
|
637 |
|
398a153b1
|
638 639 |
static void inc_rt_prio_smp(struct rt_rq *rt_rq, int prio, int prev_prio) |
63489e45e
|
640 |
{ |
4d9842776
|
641 |
struct rq *rq = rq_of_rt_rq(rt_rq); |
1f11eb6a8
|
642 |
|
398a153b1
|
643 |
if (prio < prev_prio) { |
4d9842776
|
644 |
|
e864c499d
|
645 646 |
/* * If the new task is higher in priority than anything on the |
398a153b1
|
647 648 |
* run-queue, we know that the previous high becomes our * next-highest. |
e864c499d
|
649 |
*/ |
398a153b1
|
650 |
rt_rq->highest_prio.next = prev_prio; |
1f11eb6a8
|
651 652 |
if (rq->online) |
4d9842776
|
653 |
cpupri_set(&rq->rd->cpupri, rq->cpu, prio); |
1100ac91b
|
654 |
|
e864c499d
|
655 656 657 658 659 660 661 662 663 664 665 666 |
} else if (prio == rt_rq->highest_prio.curr) /* * If the next task is equal in priority to the highest on * the run-queue, then we implicitly know that the next highest * task cannot be any lower than current */ rt_rq->highest_prio.next = prio; else if (prio < rt_rq->highest_prio.next) /* * Otherwise, we need to recompute next-highest */ rt_rq->highest_prio.next = next_prio(rq); |
398a153b1
|
667 |
} |
73fe6aae8
|
668 |
|
398a153b1
|
669 670 671 672 |
static void dec_rt_prio_smp(struct rt_rq *rt_rq, int prio, int prev_prio) { struct rq *rq = rq_of_rt_rq(rt_rq); |
d0b27fa77
|
673 |
|
398a153b1
|
674 675 676 677 678 |
if (rt_rq->rt_nr_running && (prio <= rt_rq->highest_prio.next)) rt_rq->highest_prio.next = next_prio(rq); if (rq->online && rt_rq->highest_prio.curr != prev_prio) cpupri_set(&rq->rd->cpupri, rq->cpu, rt_rq->highest_prio.curr); |
63489e45e
|
679 |
} |
398a153b1
|
680 |
#else /* CONFIG_SMP */ |
6f505b164
|
681 |
static inline |
398a153b1
|
682 683 684 685 686 |
void inc_rt_prio_smp(struct rt_rq *rt_rq, int prio, int prev_prio) {} static inline void dec_rt_prio_smp(struct rt_rq *rt_rq, int prio, int prev_prio) {} #endif /* CONFIG_SMP */ |
6e0534f27
|
687 |
|
052f1dc7e
|
688 |
#if defined CONFIG_SMP || defined CONFIG_RT_GROUP_SCHED |
398a153b1
|
689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 |
static void inc_rt_prio(struct rt_rq *rt_rq, int prio) { int prev_prio = rt_rq->highest_prio.curr; if (prio < prev_prio) rt_rq->highest_prio.curr = prio; inc_rt_prio_smp(rt_rq, prio, prev_prio); } static void dec_rt_prio(struct rt_rq *rt_rq, int prio) { int prev_prio = rt_rq->highest_prio.curr; |
6f505b164
|
704 |
if (rt_rq->rt_nr_running) { |
764a9d6fe
|
705 |
|
398a153b1
|
706 |
WARN_ON(prio < prev_prio); |
764a9d6fe
|
707 |
|
e864c499d
|
708 |
/* |
398a153b1
|
709 710 |
* This may have been our highest task, and therefore * we may have some recomputation to do |
e864c499d
|
711 |
*/ |
398a153b1
|
712 |
if (prio == prev_prio) { |
e864c499d
|
713 714 715 |
struct rt_prio_array *array = &rt_rq->active; rt_rq->highest_prio.curr = |
764a9d6fe
|
716 |
sched_find_first_bit(array->bitmap); |
e864c499d
|
717 |
} |
764a9d6fe
|
718 |
} else |
e864c499d
|
719 |
rt_rq->highest_prio.curr = MAX_RT_PRIO; |
73fe6aae8
|
720 |
|
398a153b1
|
721 722 |
dec_rt_prio_smp(rt_rq, prio, prev_prio); } |
1f11eb6a8
|
723 |
|
398a153b1
|
724 725 726 727 728 729 |
#else static inline void inc_rt_prio(struct rt_rq *rt_rq, int prio) {} static inline void dec_rt_prio(struct rt_rq *rt_rq, int prio) {} #endif /* CONFIG_SMP || CONFIG_RT_GROUP_SCHED */ |
6e0534f27
|
730 |
|
052f1dc7e
|
731 |
#ifdef CONFIG_RT_GROUP_SCHED |
398a153b1
|
732 733 734 735 736 737 738 739 740 741 742 743 744 745 |
static void inc_rt_group(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq) { if (rt_se_boosted(rt_se)) rt_rq->rt_nr_boosted++; if (rt_rq->tg) start_rt_bandwidth(&rt_rq->tg->rt_bandwidth); } static void dec_rt_group(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq) { |
23b0fdfc9
|
746 747 748 749 |
if (rt_se_boosted(rt_se)) rt_rq->rt_nr_boosted--; WARN_ON(!rt_rq->rt_nr_running && rt_rq->rt_nr_boosted); |
398a153b1
|
750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 |
} #else /* CONFIG_RT_GROUP_SCHED */ static void inc_rt_group(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq) { start_rt_bandwidth(&def_rt_bandwidth); } static inline void dec_rt_group(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq) {} #endif /* CONFIG_RT_GROUP_SCHED */ static inline void inc_rt_tasks(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq) { int prio = rt_se_prio(rt_se); WARN_ON(!rt_prio(prio)); rt_rq->rt_nr_running++; inc_rt_prio(rt_rq, prio); inc_rt_migration(rt_se, rt_rq); inc_rt_group(rt_se, rt_rq); } static inline void dec_rt_tasks(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq) { WARN_ON(!rt_prio(rt_se_prio(rt_se))); WARN_ON(!rt_rq->rt_nr_running); rt_rq->rt_nr_running--; dec_rt_prio(rt_rq, rt_se_prio(rt_se)); dec_rt_migration(rt_se, rt_rq); dec_rt_group(rt_se, rt_rq); |
63489e45e
|
788 |
} |
37dad3fce
|
789 |
static void __enqueue_rt_entity(struct sched_rt_entity *rt_se, bool head) |
bb44e5d1c
|
790 |
{ |
6f505b164
|
791 792 793 |
struct rt_rq *rt_rq = rt_rq_of_se(rt_se); struct rt_prio_array *array = &rt_rq->active; struct rt_rq *group_rq = group_rt_rq(rt_se); |
20b6331bf
|
794 |
struct list_head *queue = array->queue + rt_se_prio(rt_se); |
bb44e5d1c
|
795 |
|
ad2a3f13b
|
796 797 798 799 800 801 802 |
/* * Don't enqueue the group if its throttled, or when empty. * The latter is a consequence of the former when a child group * get throttled and the current group doesn't have any other * active members. */ if (group_rq && (rt_rq_throttled(group_rq) || !group_rq->rt_nr_running)) |
6f505b164
|
803 |
return; |
63489e45e
|
804 |
|
3d4b47b4b
|
805 806 |
if (!rt_rq->rt_nr_running) list_add_leaf_rt_rq(rt_rq); |
37dad3fce
|
807 808 809 810 |
if (head) list_add(&rt_se->run_list, queue); else list_add_tail(&rt_se->run_list, queue); |
6f505b164
|
811 |
__set_bit(rt_se_prio(rt_se), array->bitmap); |
78f2c7db6
|
812 |
|
6f505b164
|
813 814 |
inc_rt_tasks(rt_se, rt_rq); } |
ad2a3f13b
|
815 |
static void __dequeue_rt_entity(struct sched_rt_entity *rt_se) |
6f505b164
|
816 817 818 819 820 821 822 823 824 |
{ struct rt_rq *rt_rq = rt_rq_of_se(rt_se); struct rt_prio_array *array = &rt_rq->active; list_del_init(&rt_se->run_list); if (list_empty(array->queue + rt_se_prio(rt_se))) __clear_bit(rt_se_prio(rt_se), array->bitmap); dec_rt_tasks(rt_se, rt_rq); |
3d4b47b4b
|
825 826 |
if (!rt_rq->rt_nr_running) list_del_leaf_rt_rq(rt_rq); |
6f505b164
|
827 828 829 830 831 |
} /* * Because the prio of an upper entry depends on the lower * entries, we must remove entries top - down. |
6f505b164
|
832 |
*/ |
ad2a3f13b
|
833 |
static void dequeue_rt_stack(struct sched_rt_entity *rt_se) |
6f505b164
|
834 |
{ |
ad2a3f13b
|
835 |
struct sched_rt_entity *back = NULL; |
6f505b164
|
836 |
|
58d6c2d72
|
837 838 839 840 841 842 843 |
for_each_sched_rt_entity(rt_se) { rt_se->back = back; back = rt_se; } for (rt_se = back; rt_se; rt_se = rt_se->back) { if (on_rt_rq(rt_se)) |
ad2a3f13b
|
844 845 846 |
__dequeue_rt_entity(rt_se); } } |
37dad3fce
|
847 |
static void enqueue_rt_entity(struct sched_rt_entity *rt_se, bool head) |
ad2a3f13b
|
848 849 850 |
{ dequeue_rt_stack(rt_se); for_each_sched_rt_entity(rt_se) |
37dad3fce
|
851 |
__enqueue_rt_entity(rt_se, head); |
ad2a3f13b
|
852 853 854 855 856 857 858 859 860 861 |
} static void dequeue_rt_entity(struct sched_rt_entity *rt_se) { dequeue_rt_stack(rt_se); for_each_sched_rt_entity(rt_se) { struct rt_rq *rt_rq = group_rt_rq(rt_se); if (rt_rq && rt_rq->rt_nr_running) |
37dad3fce
|
862 |
__enqueue_rt_entity(rt_se, false); |
58d6c2d72
|
863 |
} |
bb44e5d1c
|
864 865 866 867 868 |
} /* * Adding/removing a task to/from a priority array: */ |
ea87bb785
|
869 |
static void |
371fd7e7a
|
870 |
enqueue_task_rt(struct rq *rq, struct task_struct *p, int flags) |
6f505b164
|
871 872 |
{ struct sched_rt_entity *rt_se = &p->rt; |
371fd7e7a
|
873 |
if (flags & ENQUEUE_WAKEUP) |
6f505b164
|
874 |
rt_se->timeout = 0; |
371fd7e7a
|
875 |
enqueue_rt_entity(rt_se, flags & ENQUEUE_HEAD); |
c09595f63
|
876 |
|
917b627d4
|
877 878 |
if (!task_current(rq, p) && p->rt.nr_cpus_allowed > 1) enqueue_pushable_task(rq, p); |
6f505b164
|
879 |
} |
371fd7e7a
|
880 |
static void dequeue_task_rt(struct rq *rq, struct task_struct *p, int flags) |
bb44e5d1c
|
881 |
{ |
6f505b164
|
882 |
struct sched_rt_entity *rt_se = &p->rt; |
bb44e5d1c
|
883 |
|
f1e14ef64
|
884 |
update_curr_rt(rq); |
ad2a3f13b
|
885 |
dequeue_rt_entity(rt_se); |
c09595f63
|
886 |
|
917b627d4
|
887 |
dequeue_pushable_task(rq, p); |
bb44e5d1c
|
888 889 890 891 892 893 |
} /* * Put task to the end of the run list without the overhead of dequeue * followed by enqueue. */ |
7ebefa8ce
|
894 895 |
static void requeue_rt_entity(struct rt_rq *rt_rq, struct sched_rt_entity *rt_se, int head) |
6f505b164
|
896 |
{ |
1cdad7153
|
897 |
if (on_rt_rq(rt_se)) { |
7ebefa8ce
|
898 899 900 901 902 903 904 |
struct rt_prio_array *array = &rt_rq->active; struct list_head *queue = array->queue + rt_se_prio(rt_se); if (head) list_move(&rt_se->run_list, queue); else list_move_tail(&rt_se->run_list, queue); |
1cdad7153
|
905 |
} |
6f505b164
|
906 |
} |
7ebefa8ce
|
907 |
static void requeue_task_rt(struct rq *rq, struct task_struct *p, int head) |
bb44e5d1c
|
908 |
{ |
6f505b164
|
909 910 |
struct sched_rt_entity *rt_se = &p->rt; struct rt_rq *rt_rq; |
bb44e5d1c
|
911 |
|
6f505b164
|
912 913 |
for_each_sched_rt_entity(rt_se) { rt_rq = rt_rq_of_se(rt_se); |
7ebefa8ce
|
914 |
requeue_rt_entity(rt_rq, rt_se, head); |
6f505b164
|
915 |
} |
bb44e5d1c
|
916 |
} |
6f505b164
|
917 |
static void yield_task_rt(struct rq *rq) |
bb44e5d1c
|
918 |
{ |
7ebefa8ce
|
919 |
requeue_task_rt(rq, rq->curr, 0); |
bb44e5d1c
|
920 |
} |
e7693a362
|
921 |
#ifdef CONFIG_SMP |
318e0893c
|
922 |
static int find_lowest_rq(struct task_struct *task); |
0017d7350
|
923 |
static int |
7608dec2c
|
924 |
select_task_rq_rt(struct task_struct *p, int sd_flag, int flags) |
e7693a362
|
925 |
{ |
7608dec2c
|
926 927 928 |
struct task_struct *curr; struct rq *rq; int cpu; |
0763a660a
|
929 |
if (sd_flag != SD_BALANCE_WAKE) |
5f3edc1b1
|
930 |
return smp_processor_id(); |
7608dec2c
|
931 932 933 934 935 |
cpu = task_cpu(p); rq = cpu_rq(cpu); rcu_read_lock(); curr = ACCESS_ONCE(rq->curr); /* unlocked access */ |
318e0893c
|
936 |
/* |
7608dec2c
|
937 |
* If the current task on @p's runqueue is an RT task, then |
e1f47d891
|
938 939 940 941 |
* try to see if we can wake this RT task up on another * runqueue. Otherwise simply start this RT task * on its current runqueue. * |
43fa5460f
|
942 943 944 945 946 947 948 949 950 |
* We want to avoid overloading runqueues. If the woken * task is a higher priority, then it will stay on this CPU * and the lower prio task should be moved to another CPU. * Even though this will probably make the lower prio task * lose its cache, we do not want to bounce a higher task * around just because it gave up its CPU, perhaps for a * lock? * * For equal prio tasks, we just let the scheduler sort it out. |
7608dec2c
|
951 952 953 954 955 956 |
* * Otherwise, just let it ride on the affined RQ and the * post-schedule router will push the preempted task away * * This test is optimistic, if we get it wrong the load-balancer * will have to sort it out. |
318e0893c
|
957 |
*/ |
7608dec2c
|
958 959 960 |
if (curr && unlikely(rt_task(curr)) && (curr->rt.nr_cpus_allowed < 2 || curr->prio < p->prio) && |
6f505b164
|
961 |
(p->rt.nr_cpus_allowed > 1)) { |
7608dec2c
|
962 |
int target = find_lowest_rq(p); |
318e0893c
|
963 |
|
7608dec2c
|
964 965 |
if (target != -1) cpu = target; |
318e0893c
|
966 |
} |
7608dec2c
|
967 |
rcu_read_unlock(); |
318e0893c
|
968 |
|
7608dec2c
|
969 |
return cpu; |
e7693a362
|
970 |
} |
7ebefa8ce
|
971 972 973 |
static void check_preempt_equal_prio(struct rq *rq, struct task_struct *p) { |
7ebefa8ce
|
974 975 |
if (rq->curr->rt.nr_cpus_allowed == 1) return; |
24600ce89
|
976 |
if (p->rt.nr_cpus_allowed != 1 |
13b8bd0a5
|
977 978 |
&& cpupri_find(&rq->rd->cpupri, p, NULL)) return; |
24600ce89
|
979 |
|
13b8bd0a5
|
980 981 |
if (!cpupri_find(&rq->rd->cpupri, rq->curr, NULL)) return; |
7ebefa8ce
|
982 983 984 985 986 987 988 989 990 |
/* * There appears to be other cpus that can accept * current and none to run 'p', so lets reschedule * to try and push current away: */ requeue_task_rt(rq, p, 1); resched_task(rq->curr); } |
e7693a362
|
991 |
#endif /* CONFIG_SMP */ |
bb44e5d1c
|
992 993 994 |
/* * Preempt the current task with a newly woken task if needed: */ |
7d4787214
|
995 |
static void check_preempt_curr_rt(struct rq *rq, struct task_struct *p, int flags) |
bb44e5d1c
|
996 |
{ |
45c01e824
|
997 |
if (p->prio < rq->curr->prio) { |
bb44e5d1c
|
998 |
resched_task(rq->curr); |
45c01e824
|
999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 |
return; } #ifdef CONFIG_SMP /* * If: * * - the newly woken task is of equal priority to the current task * - the newly woken task is non-migratable while current is migratable * - current will be preempted on the next reschedule * * we should check to see if current can readily move to a different * cpu. If so, we will reschedule to allow the push logic to try * to move current somewhere else, making room for our non-migratable * task. */ |
7ebefa8ce
|
1015 1016 |
if (p->prio == rq->curr->prio && !need_resched()) check_preempt_equal_prio(rq, p); |
45c01e824
|
1017 |
#endif |
bb44e5d1c
|
1018 |
} |
6f505b164
|
1019 1020 |
static struct sched_rt_entity *pick_next_rt_entity(struct rq *rq, struct rt_rq *rt_rq) |
bb44e5d1c
|
1021 |
{ |
6f505b164
|
1022 1023 |
struct rt_prio_array *array = &rt_rq->active; struct sched_rt_entity *next = NULL; |
bb44e5d1c
|
1024 1025 1026 1027 |
struct list_head *queue; int idx; idx = sched_find_first_bit(array->bitmap); |
6f505b164
|
1028 |
BUG_ON(idx >= MAX_RT_PRIO); |
bb44e5d1c
|
1029 1030 |
queue = array->queue + idx; |
6f505b164
|
1031 |
next = list_entry(queue->next, struct sched_rt_entity, run_list); |
326587b84
|
1032 |
|
6f505b164
|
1033 1034 |
return next; } |
bb44e5d1c
|
1035 |
|
917b627d4
|
1036 |
static struct task_struct *_pick_next_task_rt(struct rq *rq) |
6f505b164
|
1037 1038 1039 1040 |
{ struct sched_rt_entity *rt_se; struct task_struct *p; struct rt_rq *rt_rq; |
bb44e5d1c
|
1041 |
|
6f505b164
|
1042 1043 1044 1045 |
rt_rq = &rq->rt; if (unlikely(!rt_rq->rt_nr_running)) return NULL; |
23b0fdfc9
|
1046 |
if (rt_rq_throttled(rt_rq)) |
6f505b164
|
1047 1048 1049 1050 |
return NULL; do { rt_se = pick_next_rt_entity(rq, rt_rq); |
326587b84
|
1051 |
BUG_ON(!rt_se); |
6f505b164
|
1052 1053 1054 1055 |
rt_rq = group_rt_rq(rt_se); } while (rt_rq); p = rt_task_of(rt_se); |
305e6835e
|
1056 |
p->se.exec_start = rq->clock_task; |
917b627d4
|
1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 |
return p; } static struct task_struct *pick_next_task_rt(struct rq *rq) { struct task_struct *p = _pick_next_task_rt(rq); /* The running task is never eligible for pushing */ if (p) dequeue_pushable_task(rq, p); |
bcf08df3b
|
1068 |
#ifdef CONFIG_SMP |
3f029d3c6
|
1069 1070 1071 1072 1073 |
/* * We detect this state here so that we can avoid taking the RQ * lock again later if there is no need to push */ rq->post_schedule = has_pushable_tasks(rq); |
bcf08df3b
|
1074 |
#endif |
3f029d3c6
|
1075 |
|
6f505b164
|
1076 |
return p; |
bb44e5d1c
|
1077 |
} |
31ee529cc
|
1078 |
static void put_prev_task_rt(struct rq *rq, struct task_struct *p) |
bb44e5d1c
|
1079 |
{ |
f1e14ef64
|
1080 |
update_curr_rt(rq); |
bb44e5d1c
|
1081 |
p->se.exec_start = 0; |
917b627d4
|
1082 1083 1084 1085 1086 |
/* * The previous task needs to be made eligible for pushing * if it is still active */ |
fd2f4419b
|
1087 |
if (on_rt_rq(&p->rt) && p->rt.nr_cpus_allowed > 1) |
917b627d4
|
1088 |
enqueue_pushable_task(rq, p); |
bb44e5d1c
|
1089 |
} |
681f3e685
|
1090 |
#ifdef CONFIG_SMP |
6f505b164
|
1091 |
|
e8fa13626
|
1092 1093 |
/* Only try algorithms three times */ #define RT_MAX_TRIES 3 |
e8fa13626
|
1094 |
static void deactivate_task(struct rq *rq, struct task_struct *p, int sleep); |
f65eda4f7
|
1095 1096 1097 |
static int pick_rt_task(struct rq *rq, struct task_struct *p, int cpu) { if (!task_running(rq, p) && |
96f874e26
|
1098 |
(cpu < 0 || cpumask_test_cpu(cpu, &p->cpus_allowed)) && |
6f505b164
|
1099 |
(p->rt.nr_cpus_allowed > 1)) |
f65eda4f7
|
1100 1101 1102 |
return 1; return 0; } |
e8fa13626
|
1103 |
/* Return the second highest RT task, NULL otherwise */ |
79064fbf7
|
1104 |
static struct task_struct *pick_next_highest_task_rt(struct rq *rq, int cpu) |
e8fa13626
|
1105 |
{ |
6f505b164
|
1106 1107 1108 1109 |
struct task_struct *next = NULL; struct sched_rt_entity *rt_se; struct rt_prio_array *array; struct rt_rq *rt_rq; |
e8fa13626
|
1110 |
int idx; |
6f505b164
|
1111 1112 1113 |
for_each_leaf_rt_rq(rt_rq, rq) { array = &rt_rq->active; idx = sched_find_first_bit(array->bitmap); |
492462742
|
1114 |
next_idx: |
6f505b164
|
1115 1116 1117 1118 1119 |
if (idx >= MAX_RT_PRIO) continue; if (next && next->prio < idx) continue; list_for_each_entry(rt_se, array->queue + idx, run_list) { |
3d07467b7
|
1120 1121 1122 1123 1124 1125 |
struct task_struct *p; if (!rt_entity_is_task(rt_se)) continue; p = rt_task_of(rt_se); |
6f505b164
|
1126 1127 1128 1129 1130 1131 1132 1133 1134 |
if (pick_rt_task(rq, p, cpu)) { next = p; break; } } if (!next) { idx = find_next_bit(array->bitmap, MAX_RT_PRIO, idx+1); goto next_idx; } |
f65eda4f7
|
1135 |
} |
e8fa13626
|
1136 1137 |
return next; } |
0e3900e6d
|
1138 |
static DEFINE_PER_CPU(cpumask_var_t, local_cpu_mask); |
e8fa13626
|
1139 |
|
6e1254d2c
|
1140 1141 1142 |
static int find_lowest_rq(struct task_struct *task) { struct sched_domain *sd; |
96f874e26
|
1143 |
struct cpumask *lowest_mask = __get_cpu_var(local_cpu_mask); |
6e1254d2c
|
1144 1145 |
int this_cpu = smp_processor_id(); int cpu = task_cpu(task); |
06f90dbd7
|
1146 |
|
6e0534f27
|
1147 1148 |
if (task->rt.nr_cpus_allowed == 1) return -1; /* No other targets possible */ |
6e1254d2c
|
1149 |
|
6e0534f27
|
1150 1151 |
if (!cpupri_find(&task_rq(task)->rd->cpupri, task, lowest_mask)) return -1; /* No targets found */ |
6e1254d2c
|
1152 1153 1154 1155 1156 1157 1158 1159 1160 |
/* * At this point we have built a mask of cpus representing the * lowest priority tasks in the system. Now we want to elect * the best one based on our affinity and topology. * * We prioritize the last cpu that the task executed on since * it is most likely cache-hot in that location. */ |
96f874e26
|
1161 |
if (cpumask_test_cpu(cpu, lowest_mask)) |
6e1254d2c
|
1162 1163 1164 1165 1166 1167 |
return cpu; /* * Otherwise, we consult the sched_domains span maps to figure * out which cpu is logically closest to our hot cache data. */ |
e2c880630
|
1168 1169 |
if (!cpumask_test_cpu(this_cpu, lowest_mask)) this_cpu = -1; /* Skip this_cpu opt if not among lowest */ |
6e1254d2c
|
1170 |
|
e2c880630
|
1171 1172 1173 |
for_each_domain(cpu, sd) { if (sd->flags & SD_WAKE_AFFINE) { int best_cpu; |
6e1254d2c
|
1174 |
|
e2c880630
|
1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 |
/* * "this_cpu" is cheaper to preempt than a * remote processor. */ if (this_cpu != -1 && cpumask_test_cpu(this_cpu, sched_domain_span(sd))) return this_cpu; best_cpu = cpumask_first_and(lowest_mask, sched_domain_span(sd)); if (best_cpu < nr_cpu_ids) return best_cpu; |
6e1254d2c
|
1187 1188 1189 1190 1191 1192 1193 1194 |
} } /* * And finally, if there were no matches within the domains * just give the caller *something* to work with from the compatible * locations. */ |
e2c880630
|
1195 1196 1197 1198 1199 1200 1201 |
if (this_cpu != -1) return this_cpu; cpu = cpumask_any(lowest_mask); if (cpu < nr_cpu_ids) return cpu; return -1; |
07b4032c9
|
1202 1203 1204 |
} /* Will lock the rq it finds */ |
4df64c0bf
|
1205 |
static struct rq *find_lock_lowest_rq(struct task_struct *task, struct rq *rq) |
07b4032c9
|
1206 1207 |
{ struct rq *lowest_rq = NULL; |
07b4032c9
|
1208 |
int tries; |
4df64c0bf
|
1209 |
int cpu; |
e8fa13626
|
1210 |
|
07b4032c9
|
1211 1212 |
for (tries = 0; tries < RT_MAX_TRIES; tries++) { cpu = find_lowest_rq(task); |
2de0b4639
|
1213 |
if ((cpu == -1) || (cpu == rq->cpu)) |
e8fa13626
|
1214 |
break; |
07b4032c9
|
1215 |
lowest_rq = cpu_rq(cpu); |
e8fa13626
|
1216 |
/* if the prio of this runqueue changed, try again */ |
07b4032c9
|
1217 |
if (double_lock_balance(rq, lowest_rq)) { |
e8fa13626
|
1218 1219 1220 1221 1222 1223 |
/* * We had to unlock the run queue. In * the mean time, task could have * migrated already or had its affinity changed. * Also make sure that it wasn't scheduled on its rq. */ |
07b4032c9
|
1224 |
if (unlikely(task_rq(task) != rq || |
96f874e26
|
1225 1226 |
!cpumask_test_cpu(lowest_rq->cpu, &task->cpus_allowed) || |
07b4032c9
|
1227 |
task_running(rq, task) || |
fd2f4419b
|
1228 |
!task->on_rq)) { |
4df64c0bf
|
1229 |
|
05fa785cf
|
1230 |
raw_spin_unlock(&lowest_rq->lock); |
e8fa13626
|
1231 1232 1233 1234 1235 1236 |
lowest_rq = NULL; break; } } /* If this rq is still suitable use it. */ |
e864c499d
|
1237 |
if (lowest_rq->rt.highest_prio.curr > task->prio) |
e8fa13626
|
1238 1239 1240 |
break; /* try again */ |
1b12bbc74
|
1241 |
double_unlock_balance(rq, lowest_rq); |
e8fa13626
|
1242 1243 1244 1245 1246 |
lowest_rq = NULL; } return lowest_rq; } |
917b627d4
|
1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 |
static struct task_struct *pick_next_pushable_task(struct rq *rq) { struct task_struct *p; if (!has_pushable_tasks(rq)) return NULL; p = plist_first_entry(&rq->rt.pushable_tasks, struct task_struct, pushable_tasks); BUG_ON(rq->cpu != task_cpu(p)); BUG_ON(task_current(rq, p)); BUG_ON(p->rt.nr_cpus_allowed <= 1); |
fd2f4419b
|
1260 |
BUG_ON(!p->on_rq); |
917b627d4
|
1261 1262 1263 1264 |
BUG_ON(!rt_task(p)); return p; } |
e8fa13626
|
1265 1266 1267 1268 1269 |
/* * If the current CPU has more than one RT task, see if the non * running task can migrate over to a CPU that is running a task * of lesser priority. */ |
697f0a487
|
1270 |
static int push_rt_task(struct rq *rq) |
e8fa13626
|
1271 1272 1273 |
{ struct task_struct *next_task; struct rq *lowest_rq; |
e8fa13626
|
1274 |
|
a22d7fc18
|
1275 1276 |
if (!rq->rt.overloaded) return 0; |
917b627d4
|
1277 |
next_task = pick_next_pushable_task(rq); |
e8fa13626
|
1278 1279 |
if (!next_task) return 0; |
492462742
|
1280 |
retry: |
697f0a487
|
1281 |
if (unlikely(next_task == rq->curr)) { |
f65eda4f7
|
1282 |
WARN_ON(1); |
e8fa13626
|
1283 |
return 0; |
f65eda4f7
|
1284 |
} |
e8fa13626
|
1285 1286 1287 1288 1289 1290 |
/* * It's possible that the next_task slipped in of * higher priority than current. If that's the case * just reschedule current. */ |
697f0a487
|
1291 1292 |
if (unlikely(next_task->prio < rq->curr->prio)) { resched_task(rq->curr); |
e8fa13626
|
1293 1294 |
return 0; } |
697f0a487
|
1295 |
/* We might release rq lock */ |
e8fa13626
|
1296 1297 1298 |
get_task_struct(next_task); /* find_lock_lowest_rq locks the rq if found */ |
697f0a487
|
1299 |
lowest_rq = find_lock_lowest_rq(next_task, rq); |
e8fa13626
|
1300 1301 1302 |
if (!lowest_rq) { struct task_struct *task; /* |
697f0a487
|
1303 |
* find lock_lowest_rq releases rq->lock |
1563513d3
|
1304 1305 1306 1307 1308 |
* so it is possible that next_task has migrated. * * We need to make sure that the task is still on the same * run-queue and is also still the next task eligible for * pushing. |
e8fa13626
|
1309 |
*/ |
917b627d4
|
1310 |
task = pick_next_pushable_task(rq); |
1563513d3
|
1311 1312 |
if (task_cpu(next_task) == rq->cpu && task == next_task) { /* |
25985edce
|
1313 |
* If we get here, the task hasn't moved at all, but |
1563513d3
|
1314 1315 1316 1317 1318 1319 |
* it has failed to push. We will not try again, * since the other cpus will pull from us when they * are ready. */ dequeue_pushable_task(rq, next_task); goto out; |
e8fa13626
|
1320 |
} |
917b627d4
|
1321 |
|
1563513d3
|
1322 1323 1324 |
if (!task) /* No more tasks, just exit */ goto out; |
917b627d4
|
1325 |
/* |
1563513d3
|
1326 |
* Something has shifted, try again. |
917b627d4
|
1327 |
*/ |
1563513d3
|
1328 1329 1330 |
put_task_struct(next_task); next_task = task; goto retry; |
e8fa13626
|
1331 |
} |
697f0a487
|
1332 |
deactivate_task(rq, next_task, 0); |
e8fa13626
|
1333 1334 1335 1336 |
set_task_cpu(next_task, lowest_rq->cpu); activate_task(lowest_rq, next_task, 0); resched_task(lowest_rq->curr); |
1b12bbc74
|
1337 |
double_unlock_balance(rq, lowest_rq); |
e8fa13626
|
1338 |
|
e8fa13626
|
1339 1340 |
out: put_task_struct(next_task); |
917b627d4
|
1341 |
return 1; |
e8fa13626
|
1342 |
} |
e8fa13626
|
1343 1344 1345 1346 1347 1348 |
static void push_rt_tasks(struct rq *rq) { /* push_rt_task will return true if it moved an RT */ while (push_rt_task(rq)) ; } |
f65eda4f7
|
1349 1350 |
static int pull_rt_task(struct rq *this_rq) { |
80bf3171d
|
1351 |
int this_cpu = this_rq->cpu, ret = 0, cpu; |
a8728944e
|
1352 |
struct task_struct *p; |
f65eda4f7
|
1353 |
struct rq *src_rq; |
f65eda4f7
|
1354 |
|
637f50851
|
1355 |
if (likely(!rt_overloaded(this_rq))) |
f65eda4f7
|
1356 |
return 0; |
c6c4927b2
|
1357 |
for_each_cpu(cpu, this_rq->rd->rto_mask) { |
f65eda4f7
|
1358 1359 1360 1361 |
if (this_cpu == cpu) continue; src_rq = cpu_rq(cpu); |
74ab8e4f6
|
1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 |
/* * Don't bother taking the src_rq->lock if the next highest * task is known to be lower-priority than our current task. * This may look racy, but if this value is about to go * logically higher, the src_rq will push this task away. * And if its going logically lower, we do not care */ if (src_rq->rt.highest_prio.next >= this_rq->rt.highest_prio.curr) continue; |
f65eda4f7
|
1373 1374 1375 |
/* * We can potentially drop this_rq's lock in * double_lock_balance, and another CPU could |
a8728944e
|
1376 |
* alter this_rq |
f65eda4f7
|
1377 |
*/ |
a8728944e
|
1378 |
double_lock_balance(this_rq, src_rq); |
f65eda4f7
|
1379 1380 1381 1382 |
/* * Are there still pullable RT tasks? */ |
614ee1f61
|
1383 1384 |
if (src_rq->rt.rt_nr_running <= 1) goto skip; |
f65eda4f7
|
1385 |
|
f65eda4f7
|
1386 1387 1388 1389 1390 1391 |
p = pick_next_highest_task_rt(src_rq, this_cpu); /* * Do we have an RT task that preempts * the to-be-scheduled task? */ |
a8728944e
|
1392 |
if (p && (p->prio < this_rq->rt.highest_prio.curr)) { |
f65eda4f7
|
1393 |
WARN_ON(p == src_rq->curr); |
fd2f4419b
|
1394 |
WARN_ON(!p->on_rq); |
f65eda4f7
|
1395 1396 1397 1398 1399 1400 1401 |
/* * There's a chance that p is higher in priority * than what's currently running on its cpu. * This is just that p is wakeing up and hasn't * had a chance to schedule. We only pull * p if it is lower in priority than the |
a8728944e
|
1402 |
* current task on the run queue |
f65eda4f7
|
1403 |
*/ |
a8728944e
|
1404 |
if (p->prio < src_rq->curr->prio) |
614ee1f61
|
1405 |
goto skip; |
f65eda4f7
|
1406 1407 1408 1409 1410 1411 1412 1413 1414 |
ret = 1; deactivate_task(src_rq, p, 0); set_task_cpu(p, this_cpu); activate_task(this_rq, p, 0); /* * We continue with the search, just in * case there's an even higher prio task |
25985edce
|
1415 |
* in another runqueue. (low likelihood |
f65eda4f7
|
1416 |
* but possible) |
f65eda4f7
|
1417 |
*/ |
f65eda4f7
|
1418 |
} |
492462742
|
1419 |
skip: |
1b12bbc74
|
1420 |
double_unlock_balance(this_rq, src_rq); |
f65eda4f7
|
1421 1422 1423 1424 |
} return ret; } |
9a897c5a6
|
1425 |
static void pre_schedule_rt(struct rq *rq, struct task_struct *prev) |
f65eda4f7
|
1426 1427 |
{ /* Try to pull RT tasks here if we lower this rq's prio */ |
e864c499d
|
1428 |
if (unlikely(rt_task(prev)) && rq->rt.highest_prio.curr > prev->prio) |
f65eda4f7
|
1429 1430 |
pull_rt_task(rq); } |
9a897c5a6
|
1431 |
static void post_schedule_rt(struct rq *rq) |
e8fa13626
|
1432 |
{ |
967fc0467
|
1433 |
push_rt_tasks(rq); |
e8fa13626
|
1434 |
} |
8ae121ac8
|
1435 1436 1437 1438 |
/* * If we are not running and we are not going to reschedule soon, we should * try to push tasks away now */ |
efbbd05a5
|
1439 |
static void task_woken_rt(struct rq *rq, struct task_struct *p) |
4642dafdf
|
1440 |
{ |
9a897c5a6
|
1441 |
if (!task_running(rq, p) && |
8ae121ac8
|
1442 |
!test_tsk_need_resched(rq->curr) && |
917b627d4
|
1443 |
has_pushable_tasks(rq) && |
b3bc211cf
|
1444 |
p->rt.nr_cpus_allowed > 1 && |
43fa5460f
|
1445 |
rt_task(rq->curr) && |
b3bc211cf
|
1446 1447 |
(rq->curr->rt.nr_cpus_allowed < 2 || rq->curr->prio < p->prio)) |
4642dafdf
|
1448 1449 |
push_rt_tasks(rq); } |
cd8ba7cd9
|
1450 |
static void set_cpus_allowed_rt(struct task_struct *p, |
96f874e26
|
1451 |
const struct cpumask *new_mask) |
73fe6aae8
|
1452 |
{ |
96f874e26
|
1453 |
int weight = cpumask_weight(new_mask); |
73fe6aae8
|
1454 1455 1456 1457 1458 1459 1460 |
BUG_ON(!rt_task(p)); /* * Update the migration status of the RQ if we have an RT task * which is running AND changing its weight value. */ |
fd2f4419b
|
1461 |
if (p->on_rq && (weight != p->rt.nr_cpus_allowed)) { |
73fe6aae8
|
1462 |
struct rq *rq = task_rq(p); |
917b627d4
|
1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 |
if (!task_current(rq, p)) { /* * Make sure we dequeue this task from the pushable list * before going further. It will either remain off of * the list because we are no longer pushable, or it * will be requeued. */ if (p->rt.nr_cpus_allowed > 1) dequeue_pushable_task(rq, p); /* * Requeue if our weight is changing and still > 1 */ if (weight > 1) enqueue_pushable_task(rq, p); } |
6f505b164
|
1480 |
if ((p->rt.nr_cpus_allowed <= 1) && (weight > 1)) { |
73fe6aae8
|
1481 |
rq->rt.rt_nr_migratory++; |
6f505b164
|
1482 |
} else if ((p->rt.nr_cpus_allowed > 1) && (weight <= 1)) { |
73fe6aae8
|
1483 1484 1485 |
BUG_ON(!rq->rt.rt_nr_migratory); rq->rt.rt_nr_migratory--; } |
398a153b1
|
1486 |
update_rt_migration(&rq->rt); |
73fe6aae8
|
1487 |
} |
96f874e26
|
1488 |
cpumask_copy(&p->cpus_allowed, new_mask); |
6f505b164
|
1489 |
p->rt.nr_cpus_allowed = weight; |
73fe6aae8
|
1490 |
} |
deeeccd41
|
1491 |
|
bdd7c81b4
|
1492 |
/* Assumes rq->lock is held */ |
1f11eb6a8
|
1493 |
static void rq_online_rt(struct rq *rq) |
bdd7c81b4
|
1494 1495 1496 |
{ if (rq->rt.overloaded) rt_set_overload(rq); |
6e0534f27
|
1497 |
|
7def2be1d
|
1498 |
__enable_runtime(rq); |
e864c499d
|
1499 |
cpupri_set(&rq->rd->cpupri, rq->cpu, rq->rt.highest_prio.curr); |
bdd7c81b4
|
1500 1501 1502 |
} /* Assumes rq->lock is held */ |
1f11eb6a8
|
1503 |
static void rq_offline_rt(struct rq *rq) |
bdd7c81b4
|
1504 1505 1506 |
{ if (rq->rt.overloaded) rt_clear_overload(rq); |
6e0534f27
|
1507 |
|
7def2be1d
|
1508 |
__disable_runtime(rq); |
6e0534f27
|
1509 |
cpupri_set(&rq->rd->cpupri, rq->cpu, CPUPRI_INVALID); |
bdd7c81b4
|
1510 |
} |
cb4698450
|
1511 1512 1513 1514 1515 |
/* * When switch from the rt queue, we bring ourselves to a position * that we might want to pull RT tasks from other runqueues. */ |
da7a735e5
|
1516 |
static void switched_from_rt(struct rq *rq, struct task_struct *p) |
cb4698450
|
1517 1518 1519 1520 1521 1522 1523 1524 |
{ /* * If there are other RT tasks then we will reschedule * and the scheduling of the other RT tasks will handle * the balancing. But if we are the last RT task * we may need to handle the pulling of RT tasks * now. */ |
fd2f4419b
|
1525 |
if (p->on_rq && !rq->rt.rt_nr_running) |
cb4698450
|
1526 1527 |
pull_rt_task(rq); } |
3d8cbdf86
|
1528 1529 1530 1531 1532 1533 |
static inline void init_sched_rt_class(void) { unsigned int i; for_each_possible_cpu(i) |
eaa958402
|
1534 |
zalloc_cpumask_var_node(&per_cpu(local_cpu_mask, i), |
6ca09dfc9
|
1535 |
GFP_KERNEL, cpu_to_node(i)); |
3d8cbdf86
|
1536 |
} |
cb4698450
|
1537 1538 1539 1540 1541 1542 1543 |
#endif /* CONFIG_SMP */ /* * When switching a task to RT, we may overload the runqueue * with RT tasks. In this case we try to push them off to * other runqueues. */ |
da7a735e5
|
1544 |
static void switched_to_rt(struct rq *rq, struct task_struct *p) |
cb4698450
|
1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 |
{ int check_resched = 1; /* * If we are already running, then there's nothing * that needs to be done. But if we are not running * we may need to preempt the current running task. * If that current running task is also an RT task * then see if we can move to another run queue. */ |
fd2f4419b
|
1555 |
if (p->on_rq && rq->curr != p) { |
cb4698450
|
1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 |
#ifdef CONFIG_SMP if (rq->rt.overloaded && push_rt_task(rq) && /* Don't resched if we changed runqueues */ rq != task_rq(p)) check_resched = 0; #endif /* CONFIG_SMP */ if (check_resched && p->prio < rq->curr->prio) resched_task(rq->curr); } } /* * Priority of the task has changed. This may cause * us to initiate a push or pull. */ |
da7a735e5
|
1571 1572 |
static void prio_changed_rt(struct rq *rq, struct task_struct *p, int oldprio) |
cb4698450
|
1573 |
{ |
fd2f4419b
|
1574 |
if (!p->on_rq) |
da7a735e5
|
1575 1576 1577 |
return; if (rq->curr == p) { |
cb4698450
|
1578 1579 1580 1581 1582 1583 1584 1585 1586 |
#ifdef CONFIG_SMP /* * If our priority decreases while running, we * may need to pull tasks to this runqueue. */ if (oldprio < p->prio) pull_rt_task(rq); /* * If there's a higher priority task waiting to run |
6fa46fa52
|
1587 1588 1589 |
* then reschedule. Note, the above pull_rt_task * can release the rq lock and p could migrate. * Only reschedule if p is still on the same runqueue. |
cb4698450
|
1590 |
*/ |
e864c499d
|
1591 |
if (p->prio > rq->rt.highest_prio.curr && rq->curr == p) |
cb4698450
|
1592 1593 1594 1595 1596 |
resched_task(p); #else /* For UP simply resched on drop of prio */ if (oldprio < p->prio) resched_task(p); |
e8fa13626
|
1597 |
#endif /* CONFIG_SMP */ |
cb4698450
|
1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 |
} else { /* * This task is not running, but if it is * greater than the current running task * then reschedule. */ if (p->prio < rq->curr->prio) resched_task(rq->curr); } } |
78f2c7db6
|
1608 1609 1610 |
static void watchdog(struct rq *rq, struct task_struct *p) { unsigned long soft, hard; |
78d7d407b
|
1611 1612 1613 |
/* max may change after cur was read, this will be fixed next tick */ soft = task_rlimit(p, RLIMIT_RTTIME); hard = task_rlimit_max(p, RLIMIT_RTTIME); |
78f2c7db6
|
1614 1615 1616 1617 1618 1619 |
if (soft != RLIM_INFINITY) { unsigned long next; p->rt.timeout++; next = DIV_ROUND_UP(min(soft, hard), USEC_PER_SEC/HZ); |
5a52dd500
|
1620 |
if (p->rt.timeout > next) |
f06febc96
|
1621 |
p->cputime_expires.sched_exp = p->se.sum_exec_runtime; |
78f2c7db6
|
1622 1623 |
} } |
bb44e5d1c
|
1624 |
|
8f4d37ec0
|
1625 |
static void task_tick_rt(struct rq *rq, struct task_struct *p, int queued) |
bb44e5d1c
|
1626 |
{ |
67e2be023
|
1627 |
update_curr_rt(rq); |
78f2c7db6
|
1628 |
watchdog(rq, p); |
bb44e5d1c
|
1629 1630 1631 1632 1633 1634 |
/* * RR tasks need a special form of timeslice management. * FIFO tasks have no timeslices. */ if (p->policy != SCHED_RR) return; |
fa717060f
|
1635 |
if (--p->rt.time_slice) |
bb44e5d1c
|
1636 |
return; |
fa717060f
|
1637 |
p->rt.time_slice = DEF_TIMESLICE; |
bb44e5d1c
|
1638 |
|
98fbc7985
|
1639 1640 1641 1642 |
/* * Requeue to the end of queue if we are not the only element * on the queue: */ |
fa717060f
|
1643 |
if (p->rt.run_list.prev != p->rt.run_list.next) { |
7ebefa8ce
|
1644 |
requeue_task_rt(rq, p, 0); |
98fbc7985
|
1645 1646 |
set_tsk_need_resched(p); } |
bb44e5d1c
|
1647 |
} |
83b699ed2
|
1648 1649 1650 |
static void set_curr_task_rt(struct rq *rq) { struct task_struct *p = rq->curr; |
305e6835e
|
1651 |
p->se.exec_start = rq->clock_task; |
917b627d4
|
1652 1653 1654 |
/* The running task is never eligible for pushing */ dequeue_pushable_task(rq, p); |
83b699ed2
|
1655 |
} |
6d686f456
|
1656 |
static unsigned int get_rr_interval_rt(struct rq *rq, struct task_struct *task) |
0d721cead
|
1657 1658 1659 1660 1661 1662 1663 1664 1665 |
{ /* * Time slice is 0 for SCHED_FIFO tasks */ if (task->policy == SCHED_RR) return DEF_TIMESLICE; else return 0; } |
2abdad0a4
|
1666 |
static const struct sched_class rt_sched_class = { |
5522d5d5f
|
1667 |
.next = &fair_sched_class, |
bb44e5d1c
|
1668 1669 1670 1671 1672 1673 1674 1675 |
.enqueue_task = enqueue_task_rt, .dequeue_task = dequeue_task_rt, .yield_task = yield_task_rt, .check_preempt_curr = check_preempt_curr_rt, .pick_next_task = pick_next_task_rt, .put_prev_task = put_prev_task_rt, |
681f3e685
|
1676 |
#ifdef CONFIG_SMP |
4ce72a2c0
|
1677 |
.select_task_rq = select_task_rq_rt, |
73fe6aae8
|
1678 |
.set_cpus_allowed = set_cpus_allowed_rt, |
1f11eb6a8
|
1679 1680 |
.rq_online = rq_online_rt, .rq_offline = rq_offline_rt, |
9a897c5a6
|
1681 1682 |
.pre_schedule = pre_schedule_rt, .post_schedule = post_schedule_rt, |
efbbd05a5
|
1683 |
.task_woken = task_woken_rt, |
cb4698450
|
1684 |
.switched_from = switched_from_rt, |
681f3e685
|
1685 |
#endif |
bb44e5d1c
|
1686 |
|
83b699ed2
|
1687 |
.set_curr_task = set_curr_task_rt, |
bb44e5d1c
|
1688 |
.task_tick = task_tick_rt, |
cb4698450
|
1689 |
|
0d721cead
|
1690 |
.get_rr_interval = get_rr_interval_rt, |
cb4698450
|
1691 1692 |
.prio_changed = prio_changed_rt, .switched_to = switched_to_rt, |
bb44e5d1c
|
1693 |
}; |
ada18de2e
|
1694 1695 1696 1697 1698 1699 |
#ifdef CONFIG_SCHED_DEBUG extern void print_rt_rq(struct seq_file *m, int cpu, struct rt_rq *rt_rq); static void print_rt_stats(struct seq_file *m, int cpu) { |
ec514c487
|
1700 |
rt_rq_iter_t iter; |
ada18de2e
|
1701 1702 1703 |
struct rt_rq *rt_rq; rcu_read_lock(); |
ec514c487
|
1704 |
for_each_rt_rq(rt_rq, iter, cpu_rq(cpu)) |
ada18de2e
|
1705 1706 1707 |
print_rt_rq(m, cpu, rt_rq); rcu_read_unlock(); } |
55e12e5e7
|
1708 |
#endif /* CONFIG_SCHED_DEBUG */ |
0e3900e6d
|
1709 |