Blame view
kernel/sched_fair.c
110 KB
bf0f6f24a sched: cfs core, ... |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 |
/* * Completely Fair Scheduling (CFS) Class (SCHED_NORMAL/SCHED_BATCH) * * Copyright (C) 2007 Red Hat, Inc., Ingo Molnar <mingo@redhat.com> * * Interactivity improvements by Mike Galbraith * (C) 2007 Mike Galbraith <efault@gmx.de> * * Various enhancements by Dmitry Adamushko. * (C) 2007 Dmitry Adamushko <dmitry.adamushko@gmail.com> * * Group scheduling enhancements by Srivatsa Vaddagiri * Copyright IBM Corporation, 2007 * Author: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com> * * Scaled math optimizations by Thomas Gleixner * Copyright (C) 2007, Thomas Gleixner <tglx@linutronix.de> |
218050855 sched: adaptive s... |
18 19 20 |
* * Adaptive scheduling granularity, math enhancements by Peter Zijlstra * Copyright (C) 2007 Red Hat, Inc., Peter Zijlstra <pzijlstr@redhat.com> |
bf0f6f24a sched: cfs core, ... |
21 |
*/ |
9745512ce sched: latencytop... |
22 |
#include <linux/latencytop.h> |
1983a922a sched: Make tunab... |
23 |
#include <linux/sched.h> |
3436ae129 sched: Fix rebala... |
24 |
#include <linux/cpumask.h> |
9745512ce sched: latencytop... |
25 |
|
bf0f6f24a sched: cfs core, ... |
26 |
/* |
218050855 sched: adaptive s... |
27 |
* Targeted preemption latency for CPU-bound tasks: |
864616ee6 sched: Comment up... |
28 |
* (default: 6ms * (1 + ilog(ncpus)), units: nanoseconds) |
bf0f6f24a sched: cfs core, ... |
29 |
* |
218050855 sched: adaptive s... |
30 |
* NOTE: this latency value is not the same as the concept of |
d274a4cee sched: update com... |
31 32 33 |
* 'timeslice length' - timeslices in CFS are of variable length * and have no persistent notion like in traditional, time-slice * based scheduling concepts. |
bf0f6f24a sched: cfs core, ... |
34 |
* |
d274a4cee sched: update com... |
35 36 |
* (to see the precise effective timeslice length of your workload, * run vmstat and monitor the context-switches (cs) field) |
bf0f6f24a sched: cfs core, ... |
37 |
*/ |
21406928a sched: Tweak sche... |
38 39 |
unsigned int sysctl_sched_latency = 6000000ULL; unsigned int normalized_sysctl_sched_latency = 6000000ULL; |
2bd8e6d42 sched: use consta... |
40 41 |
/* |
1983a922a sched: Make tunab... |
42 43 44 45 46 47 48 49 50 51 52 53 |
* The initial- and re-scaling of tunables is configurable * (default SCHED_TUNABLESCALING_LOG = *(1+ilog(ncpus)) * * Options are: * SCHED_TUNABLESCALING_NONE - unscaled, always *1 * SCHED_TUNABLESCALING_LOG - scaled logarithmical, *1+ilog(ncpus) * SCHED_TUNABLESCALING_LINEAR - scaled linear, *ncpus */ enum sched_tunable_scaling sysctl_sched_tunable_scaling = SCHED_TUNABLESCALING_LOG; /* |
b2be5e96d sched: reintroduc... |
54 |
* Minimal preemption granularity for CPU-bound tasks: |
864616ee6 sched: Comment up... |
55 |
* (default: 0.75 msec * (1 + ilog(ncpus)), units: nanoseconds) |
2bd8e6d42 sched: use consta... |
56 |
*/ |
0bf377bbb sched: Improve la... |
57 58 |
unsigned int sysctl_sched_min_granularity = 750000ULL; unsigned int normalized_sysctl_sched_min_granularity = 750000ULL; |
218050855 sched: adaptive s... |
59 60 |
/* |
b2be5e96d sched: reintroduc... |
61 62 |
* is kept at sysctl_sched_latency / sysctl_sched_min_granularity */ |
0bf377bbb sched: Improve la... |
63 |
static unsigned int sched_nr_latency = 8; |
b2be5e96d sched: reintroduc... |
64 65 |
/* |
2bba22c50 sched: Turn off c... |
66 |
* After fork, child runs first. If set to 0 (default) then |
b2be5e96d sched: reintroduc... |
67 |
* parent will (try to) run first. |
218050855 sched: adaptive s... |
68 |
*/ |
2bba22c50 sched: Turn off c... |
69 |
unsigned int sysctl_sched_child_runs_first __read_mostly; |
bf0f6f24a sched: cfs core, ... |
70 71 |
/* |
bf0f6f24a sched: cfs core, ... |
72 |
* SCHED_OTHER wake-up granularity. |
172e082a9 sched: Re-tune th... |
73 |
* (default: 1 msec * (1 + ilog(ncpus)), units: nanoseconds) |
bf0f6f24a sched: cfs core, ... |
74 75 76 77 78 |
* * This option delays the preemption effects of decoupled workloads * and reduces their over-scheduling. Synchronous workloads will still * have immediate wakeup/sleep latencies. */ |
172e082a9 sched: Re-tune th... |
79 |
unsigned int sysctl_sched_wakeup_granularity = 1000000UL; |
0bcdcf28c sched: Fix missin... |
80 |
unsigned int normalized_sysctl_sched_wakeup_granularity = 1000000UL; |
bf0f6f24a sched: cfs core, ... |
81 |
|
da84d9617 sched: reintroduc... |
82 |
const_debug unsigned int sysctl_sched_migration_cost = 500000UL; |
a7a4f8a75 sched: Add sysctl... |
83 84 85 86 87 88 |
/* * The exponential sliding window over which load is averaged for shares * distribution. * (default: 10msec) */ unsigned int __read_mostly sysctl_sched_shares_window = 10000000UL; |
a4c2f00f5 sched: fair sched... |
89 |
static const struct sched_class fair_sched_class; |
bf0f6f24a sched: cfs core, ... |
90 91 92 |
/************************************************************** * CFS operations on generic schedulable entities: */ |
62160e3f4 sched: track cfs_... |
93 |
#ifdef CONFIG_FAIR_GROUP_SCHED |
bf0f6f24a sched: cfs core, ... |
94 |
|
62160e3f4 sched: track cfs_... |
95 |
/* cpu runqueue to which this cfs_rq is attached */ |
bf0f6f24a sched: cfs core, ... |
96 97 |
static inline struct rq *rq_of(struct cfs_rq *cfs_rq) { |
62160e3f4 sched: track cfs_... |
98 |
return cfs_rq->rq; |
bf0f6f24a sched: cfs core, ... |
99 |
} |
62160e3f4 sched: track cfs_... |
100 101 |
/* An entity is a task if it doesn't "own" a runqueue */ #define entity_is_task(se) (!se->my_q) |
bf0f6f24a sched: cfs core, ... |
102 |
|
8f48894fc sched: Add debug ... |
103 104 105 106 107 108 109 |
static inline struct task_struct *task_of(struct sched_entity *se) { #ifdef CONFIG_SCHED_DEBUG WARN_ON_ONCE(!entity_is_task(se)); #endif return container_of(se, struct task_struct, se); } |
b758149c0 sched: prepatory ... |
110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 |
/* Walk up scheduling entities hierarchy */ #define for_each_sched_entity(se) \ for (; se; se = se->parent) static inline struct cfs_rq *task_cfs_rq(struct task_struct *p) { return p->se.cfs_rq; } /* runqueue on which this entity is (to be) queued */ static inline struct cfs_rq *cfs_rq_of(struct sched_entity *se) { return se->cfs_rq; } /* runqueue "owned" by this group */ static inline struct cfs_rq *group_cfs_rq(struct sched_entity *grp) { return grp->my_q; } |
3d4b47b4b sched: Implement ... |
130 131 132 |
static inline void list_add_leaf_cfs_rq(struct cfs_rq *cfs_rq) { if (!cfs_rq->on_list) { |
67e86250f sched: Introduce ... |
133 134 135 136 137 138 139 140 141 142 143 144 |
/* * Ensure we either appear before our parent (if already * enqueued) or force our parent to appear after us when it is * enqueued. The fact that we always enqueue bottom-up * reduces this to two cases. */ if (cfs_rq->tg->parent && cfs_rq->tg->parent->cfs_rq[cpu_of(rq_of(cfs_rq))]->on_list) { list_add_rcu(&cfs_rq->leaf_cfs_rq_list, &rq_of(cfs_rq)->leaf_cfs_rq_list); } else { list_add_tail_rcu(&cfs_rq->leaf_cfs_rq_list, |
3d4b47b4b sched: Implement ... |
145 |
&rq_of(cfs_rq)->leaf_cfs_rq_list); |
67e86250f sched: Introduce ... |
146 |
} |
3d4b47b4b sched: Implement ... |
147 148 149 150 151 152 153 154 155 156 157 158 |
cfs_rq->on_list = 1; } } static inline void list_del_leaf_cfs_rq(struct cfs_rq *cfs_rq) { if (cfs_rq->on_list) { list_del_rcu(&cfs_rq->leaf_cfs_rq_list); cfs_rq->on_list = 0; } } |
b758149c0 sched: prepatory ... |
159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 |
/* Iterate thr' all leaf cfs_rq's on a runqueue */ #define for_each_leaf_cfs_rq(rq, cfs_rq) \ list_for_each_entry_rcu(cfs_rq, &rq->leaf_cfs_rq_list, leaf_cfs_rq_list) /* Do the two (enqueued) entities belong to the same group ? */ static inline int is_same_group(struct sched_entity *se, struct sched_entity *pse) { if (se->cfs_rq == pse->cfs_rq) return 1; return 0; } static inline struct sched_entity *parent_entity(struct sched_entity *se) { return se->parent; } |
464b75273 sched: re-instate... |
177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 |
/* return depth at which a sched entity is present in the hierarchy */ static inline int depth_se(struct sched_entity *se) { int depth = 0; for_each_sched_entity(se) depth++; return depth; } static void find_matching_se(struct sched_entity **se, struct sched_entity **pse) { int se_depth, pse_depth; /* * preemption test can be made between sibling entities who are in the * same cfs_rq i.e who have a common parent. Walk up the hierarchy of * both tasks until we find their ancestors who are siblings of common * parent. */ /* First walk up until both entities are at same depth */ se_depth = depth_se(*se); pse_depth = depth_se(*pse); while (se_depth > pse_depth) { se_depth--; *se = parent_entity(*se); } while (pse_depth > se_depth) { pse_depth--; *pse = parent_entity(*pse); } while (!is_same_group(*se, *pse)) { *se = parent_entity(*se); *pse = parent_entity(*pse); } } |
8f48894fc sched: Add debug ... |
219 220 221 222 223 224 |
#else /* !CONFIG_FAIR_GROUP_SCHED */ static inline struct task_struct *task_of(struct sched_entity *se) { return container_of(se, struct task_struct, se); } |
bf0f6f24a sched: cfs core, ... |
225 |
|
62160e3f4 sched: track cfs_... |
226 227 228 |
static inline struct rq *rq_of(struct cfs_rq *cfs_rq) { return container_of(cfs_rq, struct rq, cfs); |
bf0f6f24a sched: cfs core, ... |
229 230 231 |
} #define entity_is_task(se) 1 |
b758149c0 sched: prepatory ... |
232 233 |
#define for_each_sched_entity(se) \ for (; se; se = NULL) |
bf0f6f24a sched: cfs core, ... |
234 |
|
b758149c0 sched: prepatory ... |
235 |
static inline struct cfs_rq *task_cfs_rq(struct task_struct *p) |
bf0f6f24a sched: cfs core, ... |
236 |
{ |
b758149c0 sched: prepatory ... |
237 |
return &task_rq(p)->cfs; |
bf0f6f24a sched: cfs core, ... |
238 |
} |
b758149c0 sched: prepatory ... |
239 240 241 242 243 244 245 246 247 248 249 250 251 |
static inline struct cfs_rq *cfs_rq_of(struct sched_entity *se) { struct task_struct *p = task_of(se); struct rq *rq = task_rq(p); return &rq->cfs; } /* runqueue "owned" by this group */ static inline struct cfs_rq *group_cfs_rq(struct sched_entity *grp) { return NULL; } |
3d4b47b4b sched: Implement ... |
252 253 254 255 256 257 258 |
static inline void list_add_leaf_cfs_rq(struct cfs_rq *cfs_rq) { } static inline void list_del_leaf_cfs_rq(struct cfs_rq *cfs_rq) { } |
b758149c0 sched: prepatory ... |
259 260 261 262 263 264 265 266 267 268 269 270 271 |
#define for_each_leaf_cfs_rq(rq, cfs_rq) \ for (cfs_rq = &rq->cfs; cfs_rq; cfs_rq = NULL) static inline int is_same_group(struct sched_entity *se, struct sched_entity *pse) { return 1; } static inline struct sched_entity *parent_entity(struct sched_entity *se) { return NULL; } |
464b75273 sched: re-instate... |
272 273 274 275 |
static inline void find_matching_se(struct sched_entity **se, struct sched_entity **pse) { } |
b758149c0 sched: prepatory ... |
276 |
#endif /* CONFIG_FAIR_GROUP_SCHED */ |
bf0f6f24a sched: cfs core, ... |
277 278 279 280 |
/************************************************************** * Scheduling class tree data structure manipulation methods: */ |
0702e3ebc sched: cleanup: f... |
281 |
static inline u64 max_vruntime(u64 min_vruntime, u64 vruntime) |
02e0431a3 sched: better min... |
282 |
{ |
368059a97 sched: max_vrunti... |
283 284 |
s64 delta = (s64)(vruntime - min_vruntime); if (delta > 0) |
02e0431a3 sched: better min... |
285 286 287 288 |
min_vruntime = vruntime; return min_vruntime; } |
0702e3ebc sched: cleanup: f... |
289 |
static inline u64 min_vruntime(u64 min_vruntime, u64 vruntime) |
b0ffd246e sched: clean up m... |
290 291 292 293 294 295 296 |
{ s64 delta = (s64)(vruntime - min_vruntime); if (delta < 0) min_vruntime = vruntime; return min_vruntime; } |
54fdc5816 sched: Account fo... |
297 298 299 300 301 |
static inline int entity_before(struct sched_entity *a, struct sched_entity *b) { return (s64)(a->vruntime - b->vruntime) < 0; } |
1af5f730f sched: more accur... |
302 303 304 305 306 307 308 309 310 311 312 |
static void update_min_vruntime(struct cfs_rq *cfs_rq) { u64 vruntime = cfs_rq->min_vruntime; if (cfs_rq->curr) vruntime = cfs_rq->curr->vruntime; if (cfs_rq->rb_leftmost) { struct sched_entity *se = rb_entry(cfs_rq->rb_leftmost, struct sched_entity, run_node); |
e17036dac sched: fix update... |
313 |
if (!cfs_rq->curr) |
1af5f730f sched: more accur... |
314 315 316 317 318 319 |
vruntime = se->vruntime; else vruntime = min_vruntime(vruntime, se->vruntime); } cfs_rq->min_vruntime = max_vruntime(cfs_rq->min_vruntime, vruntime); |
3fe1698b7 sched: Deal with ... |
320 321 322 323 |
#ifndef CONFIG_64BIT smp_wmb(); cfs_rq->min_vruntime_copy = cfs_rq->min_vruntime; #endif |
1af5f730f sched: more accur... |
324 |
} |
bf0f6f24a sched: cfs core, ... |
325 326 327 |
/* * Enqueue an entity into the rb-tree: */ |
0702e3ebc sched: cleanup: f... |
328 |
static void __enqueue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se) |
bf0f6f24a sched: cfs core, ... |
329 330 331 332 |
{ struct rb_node **link = &cfs_rq->tasks_timeline.rb_node; struct rb_node *parent = NULL; struct sched_entity *entry; |
bf0f6f24a sched: cfs core, ... |
333 334 335 336 337 338 339 340 341 342 343 344 |
int leftmost = 1; /* * Find the right place in the rbtree: */ while (*link) { parent = *link; entry = rb_entry(parent, struct sched_entity, run_node); /* * We dont care about collisions. Nodes with * the same key stay together. */ |
2bd2d6f2d sched: Replace us... |
345 |
if (entity_before(se, entry)) { |
bf0f6f24a sched: cfs core, ... |
346 347 348 349 350 351 352 353 354 355 356 |
link = &parent->rb_left; } else { link = &parent->rb_right; leftmost = 0; } } /* * Maintain a cache of leftmost tree entries (it is frequently * used): */ |
1af5f730f sched: more accur... |
357 |
if (leftmost) |
57cb499df sched: remove set... |
358 |
cfs_rq->rb_leftmost = &se->run_node; |
bf0f6f24a sched: cfs core, ... |
359 360 361 |
rb_link_node(&se->run_node, parent, link); rb_insert_color(&se->run_node, &cfs_rq->tasks_timeline); |
bf0f6f24a sched: cfs core, ... |
362 |
} |
0702e3ebc sched: cleanup: f... |
363 |
static void __dequeue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se) |
bf0f6f24a sched: cfs core, ... |
364 |
{ |
3fe69747d sched: min_vrunti... |
365 366 |
if (cfs_rq->rb_leftmost == &se->run_node) { struct rb_node *next_node; |
3fe69747d sched: min_vrunti... |
367 368 369 |
next_node = rb_next(&se->run_node); cfs_rq->rb_leftmost = next_node; |
3fe69747d sched: min_vrunti... |
370 |
} |
e9acbff64 sched: introduce ... |
371 |
|
bf0f6f24a sched: cfs core, ... |
372 |
rb_erase(&se->run_node, &cfs_rq->tasks_timeline); |
bf0f6f24a sched: cfs core, ... |
373 |
} |
ac53db596 sched: Use a budd... |
374 |
static struct sched_entity *__pick_first_entity(struct cfs_rq *cfs_rq) |
bf0f6f24a sched: cfs core, ... |
375 |
{ |
f4b6755fb sched: cleanup fa... |
376 377 378 379 380 381 |
struct rb_node *left = cfs_rq->rb_leftmost; if (!left) return NULL; return rb_entry(left, struct sched_entity, run_node); |
bf0f6f24a sched: cfs core, ... |
382 |
} |
ac53db596 sched: Use a budd... |
383 384 385 386 387 388 389 390 391 392 393 |
static struct sched_entity *__pick_next_entity(struct sched_entity *se) { struct rb_node *next = rb_next(&se->run_node); if (!next) return NULL; return rb_entry(next, struct sched_entity, run_node); } #ifdef CONFIG_SCHED_DEBUG |
f4b6755fb sched: cleanup fa... |
394 |
static struct sched_entity *__pick_last_entity(struct cfs_rq *cfs_rq) |
aeb73b040 sched: clean up n... |
395 |
{ |
7eee3e677 sched: clean up _... |
396 |
struct rb_node *last = rb_last(&cfs_rq->tasks_timeline); |
aeb73b040 sched: clean up n... |
397 |
|
70eee74b7 sched: remove dup... |
398 399 |
if (!last) return NULL; |
7eee3e677 sched: clean up _... |
400 401 |
return rb_entry(last, struct sched_entity, run_node); |
aeb73b040 sched: clean up n... |
402 |
} |
bf0f6f24a sched: cfs core, ... |
403 404 405 |
/************************************************************** * Scheduling class statistics methods: */ |
acb4a848d sched: Update nor... |
406 |
int sched_proc_update_handler(struct ctl_table *table, int write, |
8d65af789 sysctl: remove "s... |
407 |
void __user *buffer, size_t *lenp, |
b2be5e96d sched: reintroduc... |
408 409 |
loff_t *ppos) { |
8d65af789 sysctl: remove "s... |
410 |
int ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos); |
acb4a848d sched: Update nor... |
411 |
int factor = get_update_sysctl_factor(); |
b2be5e96d sched: reintroduc... |
412 413 414 415 416 417 |
if (ret || !write) return ret; sched_nr_latency = DIV_ROUND_UP(sysctl_sched_latency, sysctl_sched_min_granularity); |
acb4a848d sched: Update nor... |
418 419 420 421 422 |
#define WRT_SYSCTL(name) \ (normalized_sysctl_##name = sysctl_##name / (factor)) WRT_SYSCTL(sched_min_granularity); WRT_SYSCTL(sched_latency); WRT_SYSCTL(sched_wakeup_granularity); |
acb4a848d sched: Update nor... |
423 |
#undef WRT_SYSCTL |
b2be5e96d sched: reintroduc... |
424 425 426 |
return 0; } #endif |
647e7cac2 sched: vslice fix... |
427 428 |
/* |
f9c0b0950 sched: revert bac... |
429 |
* delta /= w |
a7be37ac8 sched: revert the... |
430 431 432 433 |
*/ static inline unsigned long calc_delta_fair(unsigned long delta, struct sched_entity *se) { |
f9c0b0950 sched: revert bac... |
434 435 |
if (unlikely(se->load.weight != NICE_0_LOAD)) delta = calc_delta_mine(delta, NICE_0_LOAD, &se->load); |
a7be37ac8 sched: revert the... |
436 437 438 439 440 |
return delta; } /* |
647e7cac2 sched: vslice fix... |
441 442 443 444 445 446 447 |
* The idea is to set a period in which each task runs once. * * When there are too many tasks (sysctl_sched_nr_latency) we have to stretch * this period because otherwise the slices get too small. * * p = (nr <= nl) ? l : l*nr/nl */ |
4d78e7b65 sched: new task p... |
448 449 450 |
static u64 __sched_period(unsigned long nr_running) { u64 period = sysctl_sched_latency; |
b2be5e96d sched: reintroduc... |
451 |
unsigned long nr_latency = sched_nr_latency; |
4d78e7b65 sched: new task p... |
452 453 |
if (unlikely(nr_running > nr_latency)) { |
4bf0b7715 sched: remove do_... |
454 |
period = sysctl_sched_min_granularity; |
4d78e7b65 sched: new task p... |
455 |
period *= nr_running; |
4d78e7b65 sched: new task p... |
456 457 458 459 |
} return period; } |
647e7cac2 sched: vslice fix... |
460 461 462 463 |
/* * We calculate the wall-time slice from the period by taking a part * proportional to the weight. * |
f9c0b0950 sched: revert bac... |
464 |
* s = p*P[w/rw] |
647e7cac2 sched: vslice fix... |
465 |
*/ |
6d0f0ebd0 sched: simplify a... |
466 |
static u64 sched_slice(struct cfs_rq *cfs_rq, struct sched_entity *se) |
218050855 sched: adaptive s... |
467 |
{ |
0a582440f sched: fix sched_... |
468 |
u64 slice = __sched_period(cfs_rq->nr_running + !se->on_rq); |
f9c0b0950 sched: revert bac... |
469 |
|
0a582440f sched: fix sched_... |
470 |
for_each_sched_entity(se) { |
6272d68cc sched: sched_slic... |
471 |
struct load_weight *load; |
3104bf03a sched: Fix out of... |
472 |
struct load_weight lw; |
6272d68cc sched: sched_slic... |
473 474 475 |
cfs_rq = cfs_rq_of(se); load = &cfs_rq->load; |
f9c0b0950 sched: revert bac... |
476 |
|
0a582440f sched: fix sched_... |
477 |
if (unlikely(!se->on_rq)) { |
3104bf03a sched: Fix out of... |
478 |
lw = cfs_rq->load; |
0a582440f sched: fix sched_... |
479 480 481 482 483 484 485 |
update_load_add(&lw, se->load.weight); load = &lw; } slice = calc_delta_mine(slice, se->load.weight, load); } return slice; |
bf0f6f24a sched: cfs core, ... |
486 |
} |
647e7cac2 sched: vslice fix... |
487 |
/* |
ac884dec6 sched: fair-group... |
488 |
* We calculate the vruntime slice of a to be inserted task |
647e7cac2 sched: vslice fix... |
489 |
* |
f9c0b0950 sched: revert bac... |
490 |
* vs = s/w |
647e7cac2 sched: vslice fix... |
491 |
*/ |
f9c0b0950 sched: revert bac... |
492 |
static u64 sched_vslice(struct cfs_rq *cfs_rq, struct sched_entity *se) |
67e9fb2a3 sched: add vslice |
493 |
{ |
f9c0b0950 sched: revert bac... |
494 |
return calc_delta_fair(sched_slice(cfs_rq, se), se); |
a7be37ac8 sched: revert the... |
495 |
} |
d6b559182 sched: Allow upda... |
496 |
static void update_cfs_load(struct cfs_rq *cfs_rq, int global_update); |
6d5ab2932 sched: Simplify u... |
497 |
static void update_cfs_shares(struct cfs_rq *cfs_rq); |
3b3d190ec sched: Implement ... |
498 |
|
a7be37ac8 sched: revert the... |
499 |
/* |
bf0f6f24a sched: cfs core, ... |
500 501 502 503 |
* Update the current task's runtime statistics. Skip current tasks that * are not in our scheduling class. */ static inline void |
8ebc91d93 sched: remove sta... |
504 505 |
__update_curr(struct cfs_rq *cfs_rq, struct sched_entity *curr, unsigned long delta_exec) |
bf0f6f24a sched: cfs core, ... |
506 |
{ |
bbdba7c0e sched: remove wai... |
507 |
unsigned long delta_exec_weighted; |
bf0f6f24a sched: cfs core, ... |
508 |
|
41acab885 sched: Implement ... |
509 510 |
schedstat_set(curr->statistics.exec_max, max((u64)delta_exec, curr->statistics.exec_max)); |
bf0f6f24a sched: cfs core, ... |
511 512 |
curr->sum_exec_runtime += delta_exec; |
7a62eabc4 sched: debug: upd... |
513 |
schedstat_add(cfs_rq, exec_clock, delta_exec); |
a7be37ac8 sched: revert the... |
514 |
delta_exec_weighted = calc_delta_fair(delta_exec, curr); |
88ec22d3e sched: Remove the... |
515 |
|
e9acbff64 sched: introduce ... |
516 |
curr->vruntime += delta_exec_weighted; |
1af5f730f sched: more accur... |
517 |
update_min_vruntime(cfs_rq); |
3b3d190ec sched: Implement ... |
518 |
|
70caf8a6c sched: Fix UP bui... |
519 |
#if defined CONFIG_SMP && defined CONFIG_FAIR_GROUP_SCHED |
3b3d190ec sched: Implement ... |
520 |
cfs_rq->load_unacc_exec_time += delta_exec; |
3b3d190ec sched: Implement ... |
521 |
#endif |
bf0f6f24a sched: cfs core, ... |
522 |
} |
b7cc08965 sched: remove the... |
523 |
static void update_curr(struct cfs_rq *cfs_rq) |
bf0f6f24a sched: cfs core, ... |
524 |
{ |
429d43bcc sched: cleanup: s... |
525 |
struct sched_entity *curr = cfs_rq->curr; |
305e6835e sched: Do not acc... |
526 |
u64 now = rq_of(cfs_rq)->clock_task; |
bf0f6f24a sched: cfs core, ... |
527 528 529 530 531 532 533 534 535 536 |
unsigned long delta_exec; if (unlikely(!curr)) return; /* * Get the amount of time the current task was running * since the last time we changed load (this cannot * overflow on 32 bits): */ |
8ebc91d93 sched: remove sta... |
537 |
delta_exec = (unsigned long)(now - curr->exec_start); |
34f28ecd0 sched: optimize u... |
538 539 |
if (!delta_exec) return; |
bf0f6f24a sched: cfs core, ... |
540 |
|
8ebc91d93 sched: remove sta... |
541 542 |
__update_curr(cfs_rq, curr, delta_exec); curr->exec_start = now; |
d842de871 sched: cpu accoun... |
543 544 545 |
if (entity_is_task(curr)) { struct task_struct *curtask = task_of(curr); |
f977bb493 perf_counter, sch... |
546 |
trace_sched_stat_runtime(curtask, delta_exec, curr->vruntime); |
d842de871 sched: cpu accoun... |
547 |
cpuacct_charge(curtask, delta_exec); |
f06febc96 timers: fix itime... |
548 |
account_group_exec_runtime(curtask, delta_exec); |
d842de871 sched: cpu accoun... |
549 |
} |
bf0f6f24a sched: cfs core, ... |
550 551 552 |
} static inline void |
5870db5b8 sched: remove the... |
553 |
update_stats_wait_start(struct cfs_rq *cfs_rq, struct sched_entity *se) |
bf0f6f24a sched: cfs core, ... |
554 |
{ |
41acab885 sched: Implement ... |
555 |
schedstat_set(se->statistics.wait_start, rq_of(cfs_rq)->clock); |
bf0f6f24a sched: cfs core, ... |
556 |
} |
bf0f6f24a sched: cfs core, ... |
557 558 559 |
/* * Task is being enqueued - update stats: */ |
d2417e5a3 sched: remove the... |
560 |
static void update_stats_enqueue(struct cfs_rq *cfs_rq, struct sched_entity *se) |
bf0f6f24a sched: cfs core, ... |
561 |
{ |
bf0f6f24a sched: cfs core, ... |
562 563 564 565 |
/* * Are we enqueueing a waiting task? (for current tasks * a dequeue/enqueue event is a NOP) */ |
429d43bcc sched: cleanup: s... |
566 |
if (se != cfs_rq->curr) |
5870db5b8 sched: remove the... |
567 |
update_stats_wait_start(cfs_rq, se); |
bf0f6f24a sched: cfs core, ... |
568 |
} |
bf0f6f24a sched: cfs core, ... |
569 |
static void |
9ef0a9615 sched: remove the... |
570 |
update_stats_wait_end(struct cfs_rq *cfs_rq, struct sched_entity *se) |
bf0f6f24a sched: cfs core, ... |
571 |
{ |
41acab885 sched: Implement ... |
572 573 574 575 576 |
schedstat_set(se->statistics.wait_max, max(se->statistics.wait_max, rq_of(cfs_rq)->clock - se->statistics.wait_start)); schedstat_set(se->statistics.wait_count, se->statistics.wait_count + 1); schedstat_set(se->statistics.wait_sum, se->statistics.wait_sum + rq_of(cfs_rq)->clock - se->statistics.wait_start); |
768d0c272 sched: Add wait, ... |
577 578 579 |
#ifdef CONFIG_SCHEDSTATS if (entity_is_task(se)) { trace_sched_stat_wait(task_of(se), |
41acab885 sched: Implement ... |
580 |
rq_of(cfs_rq)->clock - se->statistics.wait_start); |
768d0c272 sched: Add wait, ... |
581 582 |
} #endif |
41acab885 sched: Implement ... |
583 |
schedstat_set(se->statistics.wait_start, 0); |
bf0f6f24a sched: cfs core, ... |
584 585 586 |
} static inline void |
19b6a2e37 sched: remove the... |
587 |
update_stats_dequeue(struct cfs_rq *cfs_rq, struct sched_entity *se) |
bf0f6f24a sched: cfs core, ... |
588 |
{ |
bf0f6f24a sched: cfs core, ... |
589 590 591 592 |
/* * Mark the end of the wait period if dequeueing a * waiting task: */ |
429d43bcc sched: cleanup: s... |
593 |
if (se != cfs_rq->curr) |
9ef0a9615 sched: remove the... |
594 |
update_stats_wait_end(cfs_rq, se); |
bf0f6f24a sched: cfs core, ... |
595 596 597 598 599 600 |
} /* * We are picking a new current task - update its stats: */ static inline void |
79303e9e0 sched: remove the... |
601 |
update_stats_curr_start(struct cfs_rq *cfs_rq, struct sched_entity *se) |
bf0f6f24a sched: cfs core, ... |
602 603 604 605 |
{ /* * We are starting a new run period: */ |
305e6835e sched: Do not acc... |
606 |
se->exec_start = rq_of(cfs_rq)->clock_task; |
bf0f6f24a sched: cfs core, ... |
607 |
} |
bf0f6f24a sched: cfs core, ... |
608 609 610 |
/************************************************** * Scheduling class queueing methods: */ |
c09595f63 sched: revert rev... |
611 612 613 614 615 616 617 618 619 620 621 622 |
#if defined CONFIG_SMP && defined CONFIG_FAIR_GROUP_SCHED static void add_cfs_task_weight(struct cfs_rq *cfs_rq, unsigned long weight) { cfs_rq->task_weight += weight; } #else static inline void add_cfs_task_weight(struct cfs_rq *cfs_rq, unsigned long weight) { } #endif |
30cfdcfc5 sched: do not kee... |
623 624 625 626 |
static void account_entity_enqueue(struct cfs_rq *cfs_rq, struct sched_entity *se) { update_load_add(&cfs_rq->load, se->load.weight); |
c09595f63 sched: revert rev... |
627 628 |
if (!parent_entity(se)) inc_cpu_load(rq_of(cfs_rq), se->load.weight); |
b87f17242 sched: maintain o... |
629 |
if (entity_is_task(se)) { |
c09595f63 sched: revert rev... |
630 |
add_cfs_task_weight(cfs_rq, se->load.weight); |
b87f17242 sched: maintain o... |
631 632 |
list_add(&se->group_node, &cfs_rq->tasks); } |
30cfdcfc5 sched: do not kee... |
633 |
cfs_rq->nr_running++; |
30cfdcfc5 sched: do not kee... |
634 635 636 637 638 639 |
} static void account_entity_dequeue(struct cfs_rq *cfs_rq, struct sched_entity *se) { update_load_sub(&cfs_rq->load, se->load.weight); |
c09595f63 sched: revert rev... |
640 641 |
if (!parent_entity(se)) dec_cpu_load(rq_of(cfs_rq), se->load.weight); |
b87f17242 sched: maintain o... |
642 |
if (entity_is_task(se)) { |
c09595f63 sched: revert rev... |
643 |
add_cfs_task_weight(cfs_rq, -se->load.weight); |
b87f17242 sched: maintain o... |
644 645 |
list_del_init(&se->group_node); } |
30cfdcfc5 sched: do not kee... |
646 |
cfs_rq->nr_running--; |
30cfdcfc5 sched: do not kee... |
647 |
} |
3ff6dcac7 sched: Fix poor i... |
648 649 |
#ifdef CONFIG_FAIR_GROUP_SCHED # ifdef CONFIG_SMP |
d6b559182 sched: Allow upda... |
650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 |
static void update_cfs_rq_load_contribution(struct cfs_rq *cfs_rq, int global_update) { struct task_group *tg = cfs_rq->tg; long load_avg; load_avg = div64_u64(cfs_rq->load_avg, cfs_rq->load_period+1); load_avg -= cfs_rq->load_contribution; if (global_update || abs(load_avg) > cfs_rq->load_contribution / 8) { atomic_add(load_avg, &tg->load_weight); cfs_rq->load_contribution += load_avg; } } static void update_cfs_load(struct cfs_rq *cfs_rq, int global_update) |
2069dd75c sched: Rewrite tg... |
666 |
{ |
a7a4f8a75 sched: Add sysctl... |
667 |
u64 period = sysctl_sched_shares_window; |
2069dd75c sched: Rewrite tg... |
668 |
u64 now, delta; |
e33078baa sched: Fix update... |
669 |
unsigned long load = cfs_rq->load.weight; |
2069dd75c sched: Rewrite tg... |
670 |
|
b815f1963 sched: Fix/remove... |
671 |
if (cfs_rq->tg == &root_task_group) |
2069dd75c sched: Rewrite tg... |
672 |
return; |
05ca62c6c sched: Use rq->cl... |
673 |
now = rq_of(cfs_rq)->clock_task; |
2069dd75c sched: Rewrite tg... |
674 |
delta = now - cfs_rq->load_stamp; |
e33078baa sched: Fix update... |
675 676 677 678 679 |
/* truncate load history at 4 idle periods */ if (cfs_rq->load_stamp > cfs_rq->load_last && now - cfs_rq->load_last > 4 * period) { cfs_rq->load_period = 0; cfs_rq->load_avg = 0; |
f07333bf6 sched: Avoid expe... |
680 |
delta = period - 1; |
e33078baa sched: Fix update... |
681 |
} |
2069dd75c sched: Rewrite tg... |
682 |
cfs_rq->load_stamp = now; |
3b3d190ec sched: Implement ... |
683 |
cfs_rq->load_unacc_exec_time = 0; |
2069dd75c sched: Rewrite tg... |
684 |
cfs_rq->load_period += delta; |
e33078baa sched: Fix update... |
685 686 687 688 |
if (load) { cfs_rq->load_last = now; cfs_rq->load_avg += delta * load; } |
2069dd75c sched: Rewrite tg... |
689 |
|
d6b559182 sched: Allow upda... |
690 691 692 693 |
/* consider updating load contribution on each fold or truncate */ if (global_update || cfs_rq->load_period > period || !cfs_rq->load_period) update_cfs_rq_load_contribution(cfs_rq, global_update); |
2069dd75c sched: Rewrite tg... |
694 695 696 697 698 699 700 701 702 703 |
while (cfs_rq->load_period > period) { /* * Inline assembly required to prevent the compiler * optimising this loop into a divmod call. * See __iter_div_u64_rem() for another example of this. */ asm("" : "+rm" (cfs_rq->load_period)); cfs_rq->load_period /= 2; cfs_rq->load_avg /= 2; } |
3d4b47b4b sched: Implement ... |
704 |
|
e33078baa sched: Fix update... |
705 706 |
if (!cfs_rq->curr && !cfs_rq->nr_running && !cfs_rq->load_avg) list_del_leaf_cfs_rq(cfs_rq); |
2069dd75c sched: Rewrite tg... |
707 |
} |
6d5ab2932 sched: Simplify u... |
708 |
static long calc_cfs_shares(struct cfs_rq *cfs_rq, struct task_group *tg) |
3ff6dcac7 sched: Fix poor i... |
709 710 |
{ long load_weight, load, shares; |
6d5ab2932 sched: Simplify u... |
711 |
load = cfs_rq->load.weight; |
3ff6dcac7 sched: Fix poor i... |
712 713 |
load_weight = atomic_read(&tg->load_weight); |
3ff6dcac7 sched: Fix poor i... |
714 |
load_weight += load; |
6d5ab2932 sched: Simplify u... |
715 |
load_weight -= cfs_rq->load_contribution; |
3ff6dcac7 sched: Fix poor i... |
716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 |
shares = (tg->shares * load); if (load_weight) shares /= load_weight; if (shares < MIN_SHARES) shares = MIN_SHARES; if (shares > tg->shares) shares = tg->shares; return shares; } static void update_entity_shares_tick(struct cfs_rq *cfs_rq) { if (cfs_rq->load_unacc_exec_time > sysctl_sched_shares_window) { update_cfs_load(cfs_rq, 0); |
6d5ab2932 sched: Simplify u... |
733 |
update_cfs_shares(cfs_rq); |
3ff6dcac7 sched: Fix poor i... |
734 735 736 737 738 739 |
} } # else /* CONFIG_SMP */ static void update_cfs_load(struct cfs_rq *cfs_rq, int global_update) { } |
6d5ab2932 sched: Simplify u... |
740 |
static inline long calc_cfs_shares(struct cfs_rq *cfs_rq, struct task_group *tg) |
3ff6dcac7 sched: Fix poor i... |
741 742 743 744 745 746 747 748 |
{ return tg->shares; } static inline void update_entity_shares_tick(struct cfs_rq *cfs_rq) { } # endif /* CONFIG_SMP */ |
2069dd75c sched: Rewrite tg... |
749 750 751 |
static void reweight_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, unsigned long weight) { |
19e5eebb8 sched: Fix intera... |
752 753 754 755 |
if (se->on_rq) { /* commit outstanding execution time */ if (cfs_rq->curr == se) update_curr(cfs_rq); |
2069dd75c sched: Rewrite tg... |
756 |
account_entity_dequeue(cfs_rq, se); |
19e5eebb8 sched: Fix intera... |
757 |
} |
2069dd75c sched: Rewrite tg... |
758 759 760 761 762 763 |
update_load_set(&se->load, weight); if (se->on_rq) account_entity_enqueue(cfs_rq, se); } |
6d5ab2932 sched: Simplify u... |
764 |
static void update_cfs_shares(struct cfs_rq *cfs_rq) |
2069dd75c sched: Rewrite tg... |
765 766 767 |
{ struct task_group *tg; struct sched_entity *se; |
3ff6dcac7 sched: Fix poor i... |
768 |
long shares; |
2069dd75c sched: Rewrite tg... |
769 |
|
2069dd75c sched: Rewrite tg... |
770 771 772 773 |
tg = cfs_rq->tg; se = tg->se[cpu_of(rq_of(cfs_rq))]; if (!se) return; |
3ff6dcac7 sched: Fix poor i... |
774 775 776 777 |
#ifndef CONFIG_SMP if (likely(se->load.weight == tg->shares)) return; #endif |
6d5ab2932 sched: Simplify u... |
778 |
shares = calc_cfs_shares(cfs_rq, tg); |
2069dd75c sched: Rewrite tg... |
779 780 781 782 |
reweight_entity(cfs_rq_of(se), se, shares); } #else /* CONFIG_FAIR_GROUP_SCHED */ |
d6b559182 sched: Allow upda... |
783 |
static void update_cfs_load(struct cfs_rq *cfs_rq, int global_update) |
2069dd75c sched: Rewrite tg... |
784 785 |
{ } |
6d5ab2932 sched: Simplify u... |
786 |
static inline void update_cfs_shares(struct cfs_rq *cfs_rq) |
2069dd75c sched: Rewrite tg... |
787 788 |
{ } |
43365bd7f sched: Move perio... |
789 790 791 792 |
static inline void update_entity_shares_tick(struct cfs_rq *cfs_rq) { } |
2069dd75c sched: Rewrite tg... |
793 |
#endif /* CONFIG_FAIR_GROUP_SCHED */ |
2396af69b sched: remove the... |
794 |
static void enqueue_sleeper(struct cfs_rq *cfs_rq, struct sched_entity *se) |
bf0f6f24a sched: cfs core, ... |
795 |
{ |
bf0f6f24a sched: cfs core, ... |
796 |
#ifdef CONFIG_SCHEDSTATS |
e414314cc sched: Fix latenc... |
797 798 799 800 |
struct task_struct *tsk = NULL; if (entity_is_task(se)) tsk = task_of(se); |
41acab885 sched: Implement ... |
801 802 |
if (se->statistics.sleep_start) { u64 delta = rq_of(cfs_rq)->clock - se->statistics.sleep_start; |
bf0f6f24a sched: cfs core, ... |
803 804 805 |
if ((s64)delta < 0) delta = 0; |
41acab885 sched: Implement ... |
806 807 |
if (unlikely(delta > se->statistics.sleep_max)) se->statistics.sleep_max = delta; |
bf0f6f24a sched: cfs core, ... |
808 |
|
41acab885 sched: Implement ... |
809 810 |
se->statistics.sleep_start = 0; se->statistics.sum_sleep_runtime += delta; |
9745512ce sched: latencytop... |
811 |
|
768d0c272 sched: Add wait, ... |
812 |
if (tsk) { |
e414314cc sched: Fix latenc... |
813 |
account_scheduler_latency(tsk, delta >> 10, 1); |
768d0c272 sched: Add wait, ... |
814 815 |
trace_sched_stat_sleep(tsk, delta); } |
bf0f6f24a sched: cfs core, ... |
816 |
} |
41acab885 sched: Implement ... |
817 818 |
if (se->statistics.block_start) { u64 delta = rq_of(cfs_rq)->clock - se->statistics.block_start; |
bf0f6f24a sched: cfs core, ... |
819 820 821 |
if ((s64)delta < 0) delta = 0; |
41acab885 sched: Implement ... |
822 823 |
if (unlikely(delta > se->statistics.block_max)) se->statistics.block_max = delta; |
bf0f6f24a sched: cfs core, ... |
824 |
|
41acab885 sched: Implement ... |
825 826 |
se->statistics.block_start = 0; se->statistics.sum_sleep_runtime += delta; |
30084fbd1 sched: fix profil... |
827 |
|
e414314cc sched: Fix latenc... |
828 |
if (tsk) { |
8f0dfc34e sched: Provide io... |
829 |
if (tsk->in_iowait) { |
41acab885 sched: Implement ... |
830 831 |
se->statistics.iowait_sum += delta; se->statistics.iowait_count++; |
768d0c272 sched: Add wait, ... |
832 |
trace_sched_stat_iowait(tsk, delta); |
8f0dfc34e sched: Provide io... |
833 |
} |
e414314cc sched: Fix latenc... |
834 835 836 837 838 839 840 841 842 843 844 |
/* * Blocking time is in units of nanosecs, so shift by * 20 to get a milliseconds-range estimation of the * amount of time that the task spent sleeping: */ if (unlikely(prof_on == SLEEP_PROFILING)) { profile_hits(SLEEP_PROFILING, (void *)get_wchan(tsk), delta >> 20); } account_scheduler_latency(tsk, delta >> 10, 0); |
30084fbd1 sched: fix profil... |
845 |
} |
bf0f6f24a sched: cfs core, ... |
846 847 848 |
} #endif } |
ddc972975 sched debug: chec... |
849 850 851 852 853 854 855 856 857 858 859 860 |
static void check_spread(struct cfs_rq *cfs_rq, struct sched_entity *se) { #ifdef CONFIG_SCHED_DEBUG s64 d = se->vruntime - cfs_rq->min_vruntime; if (d < 0) d = -d; if (d > 3*sysctl_sched_latency) schedstat_inc(cfs_rq, nr_spread_over); #endif } |
bf0f6f24a sched: cfs core, ... |
861 |
static void |
aeb73b040 sched: clean up n... |
862 863 |
place_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int initial) { |
1af5f730f sched: more accur... |
864 |
u64 vruntime = cfs_rq->min_vruntime; |
94dfb5e75 sched: add tree b... |
865 |
|
2cb8600e6 sched: documentat... |
866 867 868 869 870 871 |
/* * The 'current' period is already promised to the current tasks, * however the extra weight of the new task will slow them down a * little, place the new task so that it fits in the slot that * stays open at the end. */ |
94dfb5e75 sched: add tree b... |
872 |
if (initial && sched_feat(START_DEBIT)) |
f9c0b0950 sched: revert bac... |
873 |
vruntime += sched_vslice(cfs_rq, se); |
aeb73b040 sched: clean up n... |
874 |
|
a2e7a7eb2 sched: Remove unn... |
875 |
/* sleeps up to a single latency don't count. */ |
5ca9880c6 sched: Remove FAI... |
876 |
if (!initial) { |
a2e7a7eb2 sched: Remove unn... |
877 |
unsigned long thresh = sysctl_sched_latency; |
a7be37ac8 sched: revert the... |
878 |
|
a2e7a7eb2 sched: Remove unn... |
879 |
/* |
a2e7a7eb2 sched: Remove unn... |
880 881 882 883 884 |
* Halve their sleep time's effect, to allow * for a gentler effect of sleepers: */ if (sched_feat(GENTLE_FAIR_SLEEPERS)) thresh >>= 1; |
51e0304ce sched: Implement ... |
885 |
|
a2e7a7eb2 sched: Remove unn... |
886 |
vruntime -= thresh; |
aeb73b040 sched: clean up n... |
887 |
} |
b5d9d734a sched: Ensure tha... |
888 889 |
/* ensure we never gain time by being placed backwards. */ vruntime = max_vruntime(se->vruntime, vruntime); |
67e9fb2a3 sched: add vslice |
890 |
se->vruntime = vruntime; |
aeb73b040 sched: clean up n... |
891 892 893 |
} static void |
88ec22d3e sched: Remove the... |
894 |
enqueue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int flags) |
bf0f6f24a sched: cfs core, ... |
895 896 |
{ /* |
88ec22d3e sched: Remove the... |
897 898 899 |
* Update the normalized vruntime before updating min_vruntime * through callig update_curr(). */ |
371fd7e7a sched: Add enqueu... |
900 |
if (!(flags & ENQUEUE_WAKEUP) || (flags & ENQUEUE_WAKING)) |
88ec22d3e sched: Remove the... |
901 902 903 |
se->vruntime += cfs_rq->min_vruntime; /* |
a2a2d6807 sched: cleanup, m... |
904 |
* Update run-time statistics of the 'current'. |
bf0f6f24a sched: cfs core, ... |
905 |
*/ |
b7cc08965 sched: remove the... |
906 |
update_curr(cfs_rq); |
d6b559182 sched: Allow upda... |
907 |
update_cfs_load(cfs_rq, 0); |
a992241de sched: fix normal... |
908 |
account_entity_enqueue(cfs_rq, se); |
6d5ab2932 sched: Simplify u... |
909 |
update_cfs_shares(cfs_rq); |
bf0f6f24a sched: cfs core, ... |
910 |
|
88ec22d3e sched: Remove the... |
911 |
if (flags & ENQUEUE_WAKEUP) { |
aeb73b040 sched: clean up n... |
912 |
place_entity(cfs_rq, se, 0); |
2396af69b sched: remove the... |
913 |
enqueue_sleeper(cfs_rq, se); |
e9acbff64 sched: introduce ... |
914 |
} |
bf0f6f24a sched: cfs core, ... |
915 |
|
d2417e5a3 sched: remove the... |
916 |
update_stats_enqueue(cfs_rq, se); |
ddc972975 sched debug: chec... |
917 |
check_spread(cfs_rq, se); |
83b699ed2 sched: revert rec... |
918 919 |
if (se != cfs_rq->curr) __enqueue_entity(cfs_rq, se); |
2069dd75c sched: Rewrite tg... |
920 |
se->on_rq = 1; |
3d4b47b4b sched: Implement ... |
921 922 923 |
if (cfs_rq->nr_running == 1) list_add_leaf_cfs_rq(cfs_rq); |
bf0f6f24a sched: cfs core, ... |
924 |
} |
2c13c919d sched: Limit the ... |
925 |
static void __clear_buddies_last(struct sched_entity *se) |
2002c6959 sched: release bu... |
926 |
{ |
2c13c919d sched: Limit the ... |
927 928 929 930 931 932 933 934 |
for_each_sched_entity(se) { struct cfs_rq *cfs_rq = cfs_rq_of(se); if (cfs_rq->last == se) cfs_rq->last = NULL; else break; } } |
2002c6959 sched: release bu... |
935 |
|
2c13c919d sched: Limit the ... |
936 937 938 939 940 941 942 943 944 |
static void __clear_buddies_next(struct sched_entity *se) { for_each_sched_entity(se) { struct cfs_rq *cfs_rq = cfs_rq_of(se); if (cfs_rq->next == se) cfs_rq->next = NULL; else break; } |
2002c6959 sched: release bu... |
945 |
} |
ac53db596 sched: Use a budd... |
946 947 948 949 950 951 952 953 954 955 |
static void __clear_buddies_skip(struct sched_entity *se) { for_each_sched_entity(se) { struct cfs_rq *cfs_rq = cfs_rq_of(se); if (cfs_rq->skip == se) cfs_rq->skip = NULL; else break; } } |
a571bbeaf sched: fix buddie... |
956 957 |
static void clear_buddies(struct cfs_rq *cfs_rq, struct sched_entity *se) { |
2c13c919d sched: Limit the ... |
958 959 960 961 962 |
if (cfs_rq->last == se) __clear_buddies_last(se); if (cfs_rq->next == se) __clear_buddies_next(se); |
ac53db596 sched: Use a budd... |
963 964 965 |
if (cfs_rq->skip == se) __clear_buddies_skip(se); |
a571bbeaf sched: fix buddie... |
966 |
} |
bf0f6f24a sched: cfs core, ... |
967 |
static void |
371fd7e7a sched: Add enqueu... |
968 |
dequeue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int flags) |
bf0f6f24a sched: cfs core, ... |
969 |
{ |
a2a2d6807 sched: cleanup, m... |
970 971 972 973 |
/* * Update run-time statistics of the 'current'. */ update_curr(cfs_rq); |
19b6a2e37 sched: remove the... |
974 |
update_stats_dequeue(cfs_rq, se); |
371fd7e7a sched: Add enqueu... |
975 |
if (flags & DEQUEUE_SLEEP) { |
67e9fb2a3 sched: add vslice |
976 |
#ifdef CONFIG_SCHEDSTATS |
bf0f6f24a sched: cfs core, ... |
977 978 979 980 |
if (entity_is_task(se)) { struct task_struct *tsk = task_of(se); if (tsk->state & TASK_INTERRUPTIBLE) |
41acab885 sched: Implement ... |
981 |
se->statistics.sleep_start = rq_of(cfs_rq)->clock; |
bf0f6f24a sched: cfs core, ... |
982 |
if (tsk->state & TASK_UNINTERRUPTIBLE) |
41acab885 sched: Implement ... |
983 |
se->statistics.block_start = rq_of(cfs_rq)->clock; |
bf0f6f24a sched: cfs core, ... |
984 |
} |
db36cc7d6 sched: clean up s... |
985 |
#endif |
67e9fb2a3 sched: add vslice |
986 |
} |
2002c6959 sched: release bu... |
987 |
clear_buddies(cfs_rq, se); |
4793241be sched: backward l... |
988 |
|
83b699ed2 sched: revert rec... |
989 |
if (se != cfs_rq->curr) |
30cfdcfc5 sched: do not kee... |
990 |
__dequeue_entity(cfs_rq, se); |
2069dd75c sched: Rewrite tg... |
991 |
se->on_rq = 0; |
d6b559182 sched: Allow upda... |
992 |
update_cfs_load(cfs_rq, 0); |
30cfdcfc5 sched: do not kee... |
993 |
account_entity_dequeue(cfs_rq, se); |
88ec22d3e sched: Remove the... |
994 995 996 997 998 999 |
/* * Normalize the entity after updating the min_vruntime because the * update can refer to the ->curr item and we need to reflect this * movement in our normalized position. */ |
371fd7e7a sched: Add enqueu... |
1000 |
if (!(flags & DEQUEUE_SLEEP)) |
88ec22d3e sched: Remove the... |
1001 |
se->vruntime -= cfs_rq->min_vruntime; |
1e8762317 sched: Fix ->min_... |
1002 1003 1004 |
update_min_vruntime(cfs_rq); update_cfs_shares(cfs_rq); |
bf0f6f24a sched: cfs core, ... |
1005 1006 1007 1008 1009 |
} /* * Preempt the current task with a newly woken task if needed: */ |
7c92e54f6 sched: simplify _... |
1010 |
static void |
2e09bf556 sched: wakeup gra... |
1011 |
check_preempt_tick(struct cfs_rq *cfs_rq, struct sched_entity *curr) |
bf0f6f24a sched: cfs core, ... |
1012 |
{ |
116978308 sched: fix ideal_... |
1013 |
unsigned long ideal_runtime, delta_exec; |
6d0f0ebd0 sched: simplify a... |
1014 |
ideal_runtime = sched_slice(cfs_rq, curr); |
116978308 sched: fix ideal_... |
1015 |
delta_exec = curr->sum_exec_runtime - curr->prev_sum_exec_runtime; |
a9f3e2b54 sched: clear budd... |
1016 |
if (delta_exec > ideal_runtime) { |
bf0f6f24a sched: cfs core, ... |
1017 |
resched_task(rq_of(cfs_rq)->curr); |
a9f3e2b54 sched: clear budd... |
1018 1019 1020 1021 1022 |
/* * The current task ran long enough, ensure it doesn't get * re-elected due to buddy favours. */ clear_buddies(cfs_rq, curr); |
f685ceaca sched: Strengthen... |
1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 |
return; } /* * Ensure that a task that missed wakeup preemption by a * narrow margin doesn't have to wait for a full slice. * This also mitigates buddy induced latencies under load. */ if (!sched_feat(WAKEUP_PREEMPT)) return; if (delta_exec < sysctl_sched_min_granularity) return; if (cfs_rq->nr_running > 1) { |
ac53db596 sched: Use a budd... |
1038 |
struct sched_entity *se = __pick_first_entity(cfs_rq); |
f685ceaca sched: Strengthen... |
1039 |
s64 delta = curr->vruntime - se->vruntime; |
d7d829441 sched: Fix signed... |
1040 1041 |
if (delta < 0) return; |
f685ceaca sched: Strengthen... |
1042 1043 |
if (delta > ideal_runtime) resched_task(rq_of(cfs_rq)->curr); |
a9f3e2b54 sched: clear budd... |
1044 |
} |
bf0f6f24a sched: cfs core, ... |
1045 |
} |
83b699ed2 sched: revert rec... |
1046 |
static void |
8494f412e sched: remove the... |
1047 |
set_next_entity(struct cfs_rq *cfs_rq, struct sched_entity *se) |
bf0f6f24a sched: cfs core, ... |
1048 |
{ |
83b699ed2 sched: revert rec... |
1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 |
/* 'current' is not kept within the tree. */ if (se->on_rq) { /* * Any task has to be enqueued before it get to execute on * a CPU. So account for the time it spent waiting on the * runqueue. */ update_stats_wait_end(cfs_rq, se); __dequeue_entity(cfs_rq, se); } |
79303e9e0 sched: remove the... |
1059 |
update_stats_curr_start(cfs_rq, se); |
429d43bcc sched: cleanup: s... |
1060 |
cfs_rq->curr = se; |
eba1ed4b7 sched: debug: tra... |
1061 1062 1063 1064 1065 1066 |
#ifdef CONFIG_SCHEDSTATS /* * Track our maximum slice length, if the CPU's load is at * least twice that of our own weight (i.e. dont track it * when there are only lesser-weight tasks around): */ |
495eca494 sched: clean up s... |
1067 |
if (rq_of(cfs_rq)->load.weight >= 2*se->load.weight) { |
41acab885 sched: Implement ... |
1068 |
se->statistics.slice_max = max(se->statistics.slice_max, |
eba1ed4b7 sched: debug: tra... |
1069 1070 1071 |
se->sum_exec_runtime - se->prev_sum_exec_runtime); } #endif |
4a55b4503 sched: improve pr... |
1072 |
se->prev_sum_exec_runtime = se->sum_exec_runtime; |
bf0f6f24a sched: cfs core, ... |
1073 |
} |
3f3a49048 sched: virtual ti... |
1074 1075 |
static int wakeup_preempt_entity(struct sched_entity *curr, struct sched_entity *se); |
ac53db596 sched: Use a budd... |
1076 1077 1078 1079 1080 1081 1082 |
/* * Pick the next process, keeping these things in mind, in this order: * 1) keep things fair between processes/task groups * 2) pick the "next" process, since someone really wants that to run * 3) pick the "last" process, for cache locality * 4) do not run the "skip" process, if something else is available */ |
f4b6755fb sched: cleanup fa... |
1083 |
static struct sched_entity *pick_next_entity(struct cfs_rq *cfs_rq) |
aa2ac2522 sched: fix overlo... |
1084 |
{ |
ac53db596 sched: Use a budd... |
1085 |
struct sched_entity *se = __pick_first_entity(cfs_rq); |
f685ceaca sched: Strengthen... |
1086 |
struct sched_entity *left = se; |
f4b6755fb sched: cleanup fa... |
1087 |
|
ac53db596 sched: Use a budd... |
1088 1089 1090 1091 1092 1093 1094 1095 1096 |
/* * Avoid running the skip buddy, if running something else can * be done without getting too unfair. */ if (cfs_rq->skip == se) { struct sched_entity *second = __pick_next_entity(se); if (second && wakeup_preempt_entity(second, left) < 1) se = second; } |
aa2ac2522 sched: fix overlo... |
1097 |
|
f685ceaca sched: Strengthen... |
1098 1099 1100 1101 1102 |
/* * Prefer last buddy, try to return the CPU to a preempted task. */ if (cfs_rq->last && wakeup_preempt_entity(cfs_rq->last, left) < 1) se = cfs_rq->last; |
ac53db596 sched: Use a budd... |
1103 1104 1105 1106 1107 |
/* * Someone really wants this to run. If it's not unfair, run it. */ if (cfs_rq->next && wakeup_preempt_entity(cfs_rq->next, left) < 1) se = cfs_rq->next; |
f685ceaca sched: Strengthen... |
1108 |
clear_buddies(cfs_rq, se); |
4793241be sched: backward l... |
1109 1110 |
return se; |
aa2ac2522 sched: fix overlo... |
1111 |
} |
ab6cde269 sched: remove the... |
1112 |
static void put_prev_entity(struct cfs_rq *cfs_rq, struct sched_entity *prev) |
bf0f6f24a sched: cfs core, ... |
1113 1114 1115 1116 1117 1118 |
{ /* * If still on the runqueue then deactivate_task() * was not called and update_curr() has to be done: */ if (prev->on_rq) |
b7cc08965 sched: remove the... |
1119 |
update_curr(cfs_rq); |
bf0f6f24a sched: cfs core, ... |
1120 |
|
ddc972975 sched debug: chec... |
1121 |
check_spread(cfs_rq, prev); |
30cfdcfc5 sched: do not kee... |
1122 |
if (prev->on_rq) { |
5870db5b8 sched: remove the... |
1123 |
update_stats_wait_start(cfs_rq, prev); |
30cfdcfc5 sched: do not kee... |
1124 1125 1126 |
/* Put 'current' back into the tree. */ __enqueue_entity(cfs_rq, prev); } |
429d43bcc sched: cleanup: s... |
1127 |
cfs_rq->curr = NULL; |
bf0f6f24a sched: cfs core, ... |
1128 |
} |
8f4d37ec0 sched: high-res p... |
1129 1130 |
static void entity_tick(struct cfs_rq *cfs_rq, struct sched_entity *curr, int queued) |
bf0f6f24a sched: cfs core, ... |
1131 |
{ |
bf0f6f24a sched: cfs core, ... |
1132 |
/* |
30cfdcfc5 sched: do not kee... |
1133 |
* Update run-time statistics of the 'current'. |
bf0f6f24a sched: cfs core, ... |
1134 |
*/ |
30cfdcfc5 sched: do not kee... |
1135 |
update_curr(cfs_rq); |
bf0f6f24a sched: cfs core, ... |
1136 |
|
43365bd7f sched: Move perio... |
1137 1138 1139 1140 |
/* * Update share accounting for long-running entities. */ update_entity_shares_tick(cfs_rq); |
8f4d37ec0 sched: high-res p... |
1141 1142 1143 1144 1145 |
#ifdef CONFIG_SCHED_HRTICK /* * queued ticks are scheduled to match the slice, so don't bother * validating it and just reschedule. */ |
983ed7a66 sched: add static... |
1146 1147 1148 1149 |
if (queued) { resched_task(rq_of(cfs_rq)->curr); return; } |
8f4d37ec0 sched: high-res p... |
1150 1151 1152 1153 1154 1155 1156 |
/* * don't let the period tick interfere with the hrtick preemption */ if (!sched_feat(DOUBLE_TICK) && hrtimer_active(&rq_of(cfs_rq)->hrtick_timer)) return; #endif |
ce6c13113 sched: disable fo... |
1157 |
if (cfs_rq->nr_running > 1 || !sched_feat(WAKEUP_PREEMPT)) |
2e09bf556 sched: wakeup gra... |
1158 |
check_preempt_tick(cfs_rq, curr); |
bf0f6f24a sched: cfs core, ... |
1159 1160 1161 1162 1163 |
} /************************************************** * CFS operations on tasks: */ |
8f4d37ec0 sched: high-res p... |
1164 1165 1166 |
#ifdef CONFIG_SCHED_HRTICK static void hrtick_start_fair(struct rq *rq, struct task_struct *p) { |
8f4d37ec0 sched: high-res p... |
1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 |
struct sched_entity *se = &p->se; struct cfs_rq *cfs_rq = cfs_rq_of(se); WARN_ON(task_rq(p) != rq); if (hrtick_enabled(rq) && cfs_rq->nr_running > 1) { u64 slice = sched_slice(cfs_rq, se); u64 ran = se->sum_exec_runtime - se->prev_sum_exec_runtime; s64 delta = slice - ran; if (delta < 0) { if (rq->curr == p) resched_task(p); return; } /* * Don't schedule slices shorter than 10000ns, that just * doesn't make sense. Rely on vruntime for fairness. */ |
31656519e sched, x86: clean... |
1187 |
if (rq->curr != p) |
157124c11 sched: fix warnin... |
1188 |
delta = max_t(s64, 10000LL, delta); |
8f4d37ec0 sched: high-res p... |
1189 |
|
31656519e sched, x86: clean... |
1190 |
hrtick_start(rq, delta); |
8f4d37ec0 sched: high-res p... |
1191 1192 |
} } |
a4c2f00f5 sched: fair sched... |
1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 |
/* * called from enqueue/dequeue and updates the hrtick when the * current task is from our class and nr_running is low enough * to matter. */ static void hrtick_update(struct rq *rq) { struct task_struct *curr = rq->curr; if (curr->sched_class != &fair_sched_class) return; if (cfs_rq_of(&curr->se)->nr_running < sched_nr_latency) hrtick_start_fair(rq, curr); } |
55e12e5e7 sched: make sched... |
1209 |
#else /* !CONFIG_SCHED_HRTICK */ |
8f4d37ec0 sched: high-res p... |
1210 1211 1212 1213 |
static inline void hrtick_start_fair(struct rq *rq, struct task_struct *p) { } |
a4c2f00f5 sched: fair sched... |
1214 1215 1216 1217 |
static inline void hrtick_update(struct rq *rq) { } |
8f4d37ec0 sched: high-res p... |
1218 |
#endif |
bf0f6f24a sched: cfs core, ... |
1219 1220 1221 1222 1223 |
/* * The enqueue_task method is called before nr_running is * increased. Here we update the fair scheduling stats and * then put the task into the rbtree: */ |
ea87bb785 sched: Extend enq... |
1224 |
static void |
371fd7e7a sched: Add enqueu... |
1225 |
enqueue_task_fair(struct rq *rq, struct task_struct *p, int flags) |
bf0f6f24a sched: cfs core, ... |
1226 1227 |
{ struct cfs_rq *cfs_rq; |
62fb18513 sched: revert loa... |
1228 |
struct sched_entity *se = &p->se; |
bf0f6f24a sched: cfs core, ... |
1229 1230 |
for_each_sched_entity(se) { |
62fb18513 sched: revert loa... |
1231 |
if (se->on_rq) |
bf0f6f24a sched: cfs core, ... |
1232 1233 |
break; cfs_rq = cfs_rq_of(se); |
88ec22d3e sched: Remove the... |
1234 1235 |
enqueue_entity(cfs_rq, se, flags); flags = ENQUEUE_WAKEUP; |
bf0f6f24a sched: cfs core, ... |
1236 |
} |
8f4d37ec0 sched: high-res p... |
1237 |
|
2069dd75c sched: Rewrite tg... |
1238 |
for_each_sched_entity(se) { |
0f3171438 sched: Cleanup du... |
1239 |
cfs_rq = cfs_rq_of(se); |
2069dd75c sched: Rewrite tg... |
1240 |
|
d6b559182 sched: Allow upda... |
1241 |
update_cfs_load(cfs_rq, 0); |
6d5ab2932 sched: Simplify u... |
1242 |
update_cfs_shares(cfs_rq); |
2069dd75c sched: Rewrite tg... |
1243 |
} |
a4c2f00f5 sched: fair sched... |
1244 |
hrtick_update(rq); |
bf0f6f24a sched: cfs core, ... |
1245 |
} |
2f36825b1 sched: Next buddy... |
1246 |
static void set_next_buddy(struct sched_entity *se); |
bf0f6f24a sched: cfs core, ... |
1247 1248 1249 1250 1251 |
/* * The dequeue_task method is called before nr_running is * decreased. We remove the task from the rbtree and * update the fair scheduling stats: */ |
371fd7e7a sched: Add enqueu... |
1252 |
static void dequeue_task_fair(struct rq *rq, struct task_struct *p, int flags) |
bf0f6f24a sched: cfs core, ... |
1253 1254 |
{ struct cfs_rq *cfs_rq; |
62fb18513 sched: revert loa... |
1255 |
struct sched_entity *se = &p->se; |
2f36825b1 sched: Next buddy... |
1256 |
int task_sleep = flags & DEQUEUE_SLEEP; |
bf0f6f24a sched: cfs core, ... |
1257 1258 1259 |
for_each_sched_entity(se) { cfs_rq = cfs_rq_of(se); |
371fd7e7a sched: Add enqueu... |
1260 |
dequeue_entity(cfs_rq, se, flags); |
2069dd75c sched: Rewrite tg... |
1261 |
|
bf0f6f24a sched: cfs core, ... |
1262 |
/* Don't dequeue parent if it has other entities besides us */ |
2f36825b1 sched: Next buddy... |
1263 1264 1265 1266 1267 1268 1269 |
if (cfs_rq->load.weight) { /* * Bias pick_next to pick a task from this cfs_rq, as * p is sleeping when it is within its sched_slice. */ if (task_sleep && parent_entity(se)) set_next_buddy(parent_entity(se)); |
9598c82dc sched: Don't upda... |
1270 1271 1272 |
/* avoid re-evaluating load for this entity */ se = parent_entity(se); |
bf0f6f24a sched: cfs core, ... |
1273 |
break; |
2f36825b1 sched: Next buddy... |
1274 |
} |
371fd7e7a sched: Add enqueu... |
1275 |
flags |= DEQUEUE_SLEEP; |
bf0f6f24a sched: cfs core, ... |
1276 |
} |
8f4d37ec0 sched: high-res p... |
1277 |
|
2069dd75c sched: Rewrite tg... |
1278 |
for_each_sched_entity(se) { |
0f3171438 sched: Cleanup du... |
1279 |
cfs_rq = cfs_rq_of(se); |
2069dd75c sched: Rewrite tg... |
1280 |
|
d6b559182 sched: Allow upda... |
1281 |
update_cfs_load(cfs_rq, 0); |
6d5ab2932 sched: Simplify u... |
1282 |
update_cfs_shares(cfs_rq); |
2069dd75c sched: Rewrite tg... |
1283 |
} |
a4c2f00f5 sched: fair sched... |
1284 |
hrtick_update(rq); |
bf0f6f24a sched: cfs core, ... |
1285 |
} |
e7693a362 sched: de-SCHED_O... |
1286 |
#ifdef CONFIG_SMP |
098fb9db2 sched: clean up w... |
1287 |
|
74f8e4b23 sched: Remove rq ... |
1288 |
static void task_waking_fair(struct task_struct *p) |
88ec22d3e sched: Remove the... |
1289 1290 1291 |
{ struct sched_entity *se = &p->se; struct cfs_rq *cfs_rq = cfs_rq_of(se); |
3fe1698b7 sched: Deal with ... |
1292 1293 1294 1295 |
u64 min_vruntime; #ifndef CONFIG_64BIT u64 min_vruntime_copy; |
88ec22d3e sched: Remove the... |
1296 |
|
3fe1698b7 sched: Deal with ... |
1297 1298 1299 1300 1301 1302 1303 1304 |
do { min_vruntime_copy = cfs_rq->min_vruntime_copy; smp_rmb(); min_vruntime = cfs_rq->min_vruntime; } while (min_vruntime != min_vruntime_copy); #else min_vruntime = cfs_rq->min_vruntime; #endif |
88ec22d3e sched: Remove the... |
1305 |
|
3fe1698b7 sched: Deal with ... |
1306 |
se->vruntime -= min_vruntime; |
88ec22d3e sched: Remove the... |
1307 |
} |
bb3469ac9 sched: hierarchic... |
1308 |
#ifdef CONFIG_FAIR_GROUP_SCHED |
f5bfb7d9f sched: bias effec... |
1309 1310 1311 1312 1313 1314 |
/* * effective_load() calculates the load change as seen from the root_task_group * * Adding load to a group doesn't make a group heavier, but can cause movement * of group shares between cpus. Assuming the shares were perfectly aligned one * can calculate the shift in shares. |
f5bfb7d9f sched: bias effec... |
1315 |
*/ |
2069dd75c sched: Rewrite tg... |
1316 |
static long effective_load(struct task_group *tg, int cpu, long wl, long wg) |
bb3469ac9 sched: hierarchic... |
1317 |
{ |
4be9daaa1 sched: fix task_h... |
1318 |
struct sched_entity *se = tg->se[cpu]; |
f1d239f73 sched: incrementa... |
1319 1320 1321 |
if (!tg->parent) return wl; |
4be9daaa1 sched: fix task_h... |
1322 |
for_each_sched_entity(se) { |
977dda7c9 sched: Update eff... |
1323 |
long lw, w; |
4be9daaa1 sched: fix task_h... |
1324 |
|
977dda7c9 sched: Update eff... |
1325 1326 |
tg = se->my_q->tg; w = se->my_q->load.weight; |
bb3469ac9 sched: hierarchic... |
1327 |
|
977dda7c9 sched: Update eff... |
1328 1329 1330 1331 |
/* use this cpu's instantaneous contribution */ lw = atomic_read(&tg->load_weight); lw -= se->my_q->load_contribution; lw += w + wg; |
4be9daaa1 sched: fix task_h... |
1332 |
|
977dda7c9 sched: Update eff... |
1333 |
wl += w; |
940959e93 sched: fixlet for... |
1334 |
|
977dda7c9 sched: Update eff... |
1335 1336 1337 1338 |
if (lw > 0 && wl < lw) wl = (wl * tg->shares) / lw; else wl = tg->shares; |
940959e93 sched: fixlet for... |
1339 |
|
977dda7c9 sched: Update eff... |
1340 1341 1342 1343 |
/* zero point is MIN_SHARES */ if (wl < MIN_SHARES) wl = MIN_SHARES; wl -= se->load.weight; |
4be9daaa1 sched: fix task_h... |
1344 |
wg = 0; |
4be9daaa1 sched: fix task_h... |
1345 |
} |
bb3469ac9 sched: hierarchic... |
1346 |
|
4be9daaa1 sched: fix task_h... |
1347 |
return wl; |
bb3469ac9 sched: hierarchic... |
1348 |
} |
4be9daaa1 sched: fix task_h... |
1349 |
|
bb3469ac9 sched: hierarchic... |
1350 |
#else |
4be9daaa1 sched: fix task_h... |
1351 |
|
83378269a sched: correct wa... |
1352 1353 |
static inline unsigned long effective_load(struct task_group *tg, int cpu, unsigned long wl, unsigned long wg) |
4be9daaa1 sched: fix task_h... |
1354 |
{ |
83378269a sched: correct wa... |
1355 |
return wl; |
bb3469ac9 sched: hierarchic... |
1356 |
} |
4be9daaa1 sched: fix task_h... |
1357 |
|
bb3469ac9 sched: hierarchic... |
1358 |
#endif |
c88d59108 sched: Merge sele... |
1359 |
static int wake_affine(struct sched_domain *sd, struct task_struct *p, int sync) |
098fb9db2 sched: clean up w... |
1360 |
{ |
e37b6a7b2 sched: Fix sign u... |
1361 |
s64 this_load, load; |
c88d59108 sched: Merge sele... |
1362 |
int idx, this_cpu, prev_cpu; |
098fb9db2 sched: clean up w... |
1363 |
unsigned long tl_per_task; |
c88d59108 sched: Merge sele... |
1364 |
struct task_group *tg; |
83378269a sched: correct wa... |
1365 |
unsigned long weight; |
b3137bc8e sched: stop wake_... |
1366 |
int balanced; |
098fb9db2 sched: clean up w... |
1367 |
|
c88d59108 sched: Merge sele... |
1368 1369 1370 1371 1372 |
idx = sd->wake_idx; this_cpu = smp_processor_id(); prev_cpu = task_cpu(p); load = source_load(prev_cpu, idx); this_load = target_load(this_cpu, idx); |
098fb9db2 sched: clean up w... |
1373 1374 |
/* |
b3137bc8e sched: stop wake_... |
1375 1376 1377 1378 |
* If sync wakeup then subtract the (maximum possible) * effect of the currently running task from the load * of the current CPU: */ |
83378269a sched: correct wa... |
1379 1380 1381 |
if (sync) { tg = task_group(current); weight = current->se.load.weight; |
c88d59108 sched: Merge sele... |
1382 |
this_load += effective_load(tg, this_cpu, -weight, -weight); |
83378269a sched: correct wa... |
1383 1384 |
load += effective_load(tg, prev_cpu, 0, -weight); } |
b3137bc8e sched: stop wake_... |
1385 |
|
83378269a sched: correct wa... |
1386 1387 |
tg = task_group(p); weight = p->se.load.weight; |
b3137bc8e sched: stop wake_... |
1388 |
|
71a29aa7b sched: Deal with ... |
1389 1390 |
/* * In low-load situations, where prev_cpu is idle and this_cpu is idle |
c88d59108 sched: Merge sele... |
1391 1392 1393 |
* due to the sync cause above having dropped this_load to 0, we'll * always have an imbalance, but there's really nothing you can do * about that, so that's good too. |
71a29aa7b sched: Deal with ... |
1394 1395 1396 1397 |
* * Otherwise check if either cpus are near enough in load to allow this * task to be woken on this_cpu. */ |
e37b6a7b2 sched: Fix sign u... |
1398 1399 |
if (this_load > 0) { s64 this_eff_load, prev_eff_load; |
e51fd5e22 sched: Fix wake_a... |
1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 |
this_eff_load = 100; this_eff_load *= power_of(prev_cpu); this_eff_load *= this_load + effective_load(tg, this_cpu, weight, weight); prev_eff_load = 100 + (sd->imbalance_pct - 100) / 2; prev_eff_load *= power_of(this_cpu); prev_eff_load *= load + effective_load(tg, prev_cpu, 0, weight); balanced = this_eff_load <= prev_eff_load; } else balanced = true; |
b3137bc8e sched: stop wake_... |
1413 1414 |
/* |
4ae7d5cef sched: improve af... |
1415 1416 1417 |
* If the currently running task will sleep within * a reasonable amount of time then attract this newly * woken task: |
098fb9db2 sched: clean up w... |
1418 |
*/ |
2fb7635c4 sched: sync wakeu... |
1419 1420 |
if (sync && balanced) return 1; |
098fb9db2 sched: clean up w... |
1421 |
|
41acab885 sched: Implement ... |
1422 |
schedstat_inc(p, se.statistics.nr_wakeups_affine_attempts); |
098fb9db2 sched: clean up w... |
1423 |
tl_per_task = cpu_avg_load_per_task(this_cpu); |
c88d59108 sched: Merge sele... |
1424 1425 1426 |
if (balanced || (this_load <= load && this_load + target_load(prev_cpu, idx) <= tl_per_task)) { |
098fb9db2 sched: clean up w... |
1427 1428 1429 1430 1431 |
/* * This domain has SD_WAKE_AFFINE and * p is cache cold in this domain, and * there is no bad imbalance. */ |
c88d59108 sched: Merge sele... |
1432 |
schedstat_inc(sd, ttwu_move_affine); |
41acab885 sched: Implement ... |
1433 |
schedstat_inc(p, se.statistics.nr_wakeups_affine); |
098fb9db2 sched: clean up w... |
1434 1435 1436 1437 1438 |
return 1; } return 0; } |
aaee1203c sched: Move sched... |
1439 1440 1441 1442 1443 |
/* * find_idlest_group finds and returns the least busy CPU group within the * domain. */ static struct sched_group * |
78e7ed53c sched: Tweak wake... |
1444 |
find_idlest_group(struct sched_domain *sd, struct task_struct *p, |
5158f4e44 sched: Clean up t... |
1445 |
int this_cpu, int load_idx) |
e7693a362 sched: de-SCHED_O... |
1446 |
{ |
b3bd3de66 gcc-4.6: kernel/*... |
1447 |
struct sched_group *idlest = NULL, *group = sd->groups; |
aaee1203c sched: Move sched... |
1448 |
unsigned long min_load = ULONG_MAX, this_load = 0; |
aaee1203c sched: Move sched... |
1449 |
int imbalance = 100 + (sd->imbalance_pct-100)/2; |
e7693a362 sched: de-SCHED_O... |
1450 |
|
aaee1203c sched: Move sched... |
1451 1452 1453 1454 |
do { unsigned long load, avg_load; int local_group; int i; |
e7693a362 sched: de-SCHED_O... |
1455 |
|
aaee1203c sched: Move sched... |
1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 |
/* Skip over this group if it has no CPUs allowed */ if (!cpumask_intersects(sched_group_cpus(group), &p->cpus_allowed)) continue; local_group = cpumask_test_cpu(this_cpu, sched_group_cpus(group)); /* Tally up the load of all CPUs in the group */ avg_load = 0; for_each_cpu(i, sched_group_cpus(group)) { /* Bias balancing toward cpus of our domain */ if (local_group) load = source_load(i, load_idx); else load = target_load(i, load_idx); avg_load += load; } /* Adjust by relative CPU power of the group */ |
9c3f75cbd sched: Break out ... |
1478 |
avg_load = (avg_load * SCHED_POWER_SCALE) / group->sgp->power; |
aaee1203c sched: Move sched... |
1479 1480 1481 |
if (local_group) { this_load = avg_load; |
aaee1203c sched: Move sched... |
1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 |
} else if (avg_load < min_load) { min_load = avg_load; idlest = group; } } while (group = group->next, group != sd->groups); if (!idlest || 100*this_load < imbalance*min_load) return NULL; return idlest; } /* * find_idlest_cpu - find the idlest cpu among the cpus in group. */ static int find_idlest_cpu(struct sched_group *group, struct task_struct *p, int this_cpu) { unsigned long load, min_load = ULONG_MAX; int idlest = -1; int i; /* Traverse only the allowed CPUs */ for_each_cpu_and(i, sched_group_cpus(group), &p->cpus_allowed) { load = weighted_cpuload(i); if (load < min_load || (load == min_load && i == this_cpu)) { min_load = load; idlest = i; |
e7693a362 sched: de-SCHED_O... |
1510 1511 |
} } |
aaee1203c sched: Move sched... |
1512 1513 |
return idlest; } |
e7693a362 sched: de-SCHED_O... |
1514 |
|
aaee1203c sched: Move sched... |
1515 |
/* |
a50bde513 sched: Cleanup se... |
1516 1517 |
* Try and locate an idle CPU in the sched_domain. */ |
99bd5e2f2 sched: Fix select... |
1518 |
static int select_idle_sibling(struct task_struct *p, int target) |
a50bde513 sched: Cleanup se... |
1519 1520 1521 |
{ int cpu = smp_processor_id(); int prev_cpu = task_cpu(p); |
99bd5e2f2 sched: Fix select... |
1522 |
struct sched_domain *sd; |
a50bde513 sched: Cleanup se... |
1523 1524 1525 |
int i; /* |
99bd5e2f2 sched: Fix select... |
1526 1527 |
* If the task is going to be woken-up on this cpu and if it is * already idle, then it is the right target. |
a50bde513 sched: Cleanup se... |
1528 |
*/ |
99bd5e2f2 sched: Fix select... |
1529 1530 1531 1532 1533 1534 1535 1536 |
if (target == cpu && idle_cpu(cpu)) return cpu; /* * If the task is going to be woken-up on the cpu where it previously * ran and if it is currently idle, then it the right target. */ if (target == prev_cpu && idle_cpu(prev_cpu)) |
fe3bcfe1f sched: More gener... |
1537 |
return prev_cpu; |
a50bde513 sched: Cleanup se... |
1538 1539 |
/* |
99bd5e2f2 sched: Fix select... |
1540 |
* Otherwise, iterate the domains and find an elegible idle cpu. |
a50bde513 sched: Cleanup se... |
1541 |
*/ |
dce840a08 sched: Dynamicall... |
1542 |
rcu_read_lock(); |
99bd5e2f2 sched: Fix select... |
1543 1544 |
for_each_domain(target, sd) { if (!(sd->flags & SD_SHARE_PKG_RESOURCES)) |
fe3bcfe1f sched: More gener... |
1545 |
break; |
99bd5e2f2 sched: Fix select... |
1546 1547 1548 1549 1550 1551 |
for_each_cpu_and(i, sched_domain_span(sd), &p->cpus_allowed) { if (idle_cpu(i)) { target = i; break; } |
a50bde513 sched: Cleanup se... |
1552 |
} |
99bd5e2f2 sched: Fix select... |
1553 1554 1555 1556 1557 1558 1559 1560 |
/* * Lets stop looking for an idle sibling when we reached * the domain that spans the current cpu and prev_cpu. */ if (cpumask_test_cpu(cpu, sched_domain_span(sd)) && cpumask_test_cpu(prev_cpu, sched_domain_span(sd))) break; |
a50bde513 sched: Cleanup se... |
1561 |
} |
dce840a08 sched: Dynamicall... |
1562 |
rcu_read_unlock(); |
a50bde513 sched: Cleanup se... |
1563 1564 1565 1566 1567 |
return target; } /* |
aaee1203c sched: Move sched... |
1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 |
* sched_balance_self: balance the current task (running on cpu) in domains * that have the 'flag' flag set. In practice, this is SD_BALANCE_FORK and * SD_BALANCE_EXEC. * * Balance, ie. select the least loaded group. * * Returns the target CPU number, or the same CPU if no balancing is needed. * * preempt must be disabled. */ |
0017d7350 sched: Fix TASK_W... |
1578 |
static int |
7608dec2c sched: Drop the r... |
1579 |
select_task_rq_fair(struct task_struct *p, int sd_flag, int wake_flags) |
aaee1203c sched: Move sched... |
1580 |
{ |
29cd8bae3 sched: Fix SD_POW... |
1581 |
struct sched_domain *tmp, *affine_sd = NULL, *sd = NULL; |
c88d59108 sched: Merge sele... |
1582 1583 1584 |
int cpu = smp_processor_id(); int prev_cpu = task_cpu(p); int new_cpu = cpu; |
99bd5e2f2 sched: Fix select... |
1585 |
int want_affine = 0; |
29cd8bae3 sched: Fix SD_POW... |
1586 |
int want_sd = 1; |
5158f4e44 sched: Clean up t... |
1587 |
int sync = wake_flags & WF_SYNC; |
c88d59108 sched: Merge sele... |
1588 |
|
0763a660a sched: Rename sel... |
1589 |
if (sd_flag & SD_BALANCE_WAKE) { |
beac4c7e4 sched: Remove AFF... |
1590 |
if (cpumask_test_cpu(cpu, &p->cpus_allowed)) |
c88d59108 sched: Merge sele... |
1591 1592 1593 |
want_affine = 1; new_cpu = prev_cpu; } |
aaee1203c sched: Move sched... |
1594 |
|
dce840a08 sched: Dynamicall... |
1595 |
rcu_read_lock(); |
aaee1203c sched: Move sched... |
1596 |
for_each_domain(cpu, tmp) { |
e4f428884 sched: Select_tas... |
1597 1598 |
if (!(tmp->flags & SD_LOAD_BALANCE)) continue; |
aaee1203c sched: Move sched... |
1599 |
/* |
ae154be1f sched: Weaken SD_... |
1600 1601 |
* If power savings logic is enabled for a domain, see if we * are not overloaded, if so, don't balance wider. |
aaee1203c sched: Move sched... |
1602 |
*/ |
59abf0264 sched: Add SD_PRE... |
1603 |
if (tmp->flags & (SD_POWERSAVINGS_BALANCE|SD_PREFER_LOCAL)) { |
ae154be1f sched: Weaken SD_... |
1604 1605 1606 1607 1608 1609 1610 1611 1612 |
unsigned long power = 0; unsigned long nr_running = 0; unsigned long capacity; int i; for_each_cpu(i, sched_domain_span(tmp)) { power += power_of(i); nr_running += cpu_rq(i)->cfs.nr_running; } |
1399fa780 sched: Introduce ... |
1613 |
capacity = DIV_ROUND_CLOSEST(power, SCHED_POWER_SCALE); |
ae154be1f sched: Weaken SD_... |
1614 |
|
59abf0264 sched: Add SD_PRE... |
1615 1616 1617 1618 |
if (tmp->flags & SD_POWERSAVINGS_BALANCE) nr_running /= 2; if (nr_running < capacity) |
29cd8bae3 sched: Fix SD_POW... |
1619 |
want_sd = 0; |
ae154be1f sched: Weaken SD_... |
1620 |
} |
aaee1203c sched: Move sched... |
1621 |
|
fe3bcfe1f sched: More gener... |
1622 |
/* |
99bd5e2f2 sched: Fix select... |
1623 1624 |
* If both cpu and prev_cpu are part of this domain, * cpu is a valid SD_WAKE_AFFINE target. |
fe3bcfe1f sched: More gener... |
1625 |
*/ |
99bd5e2f2 sched: Fix select... |
1626 1627 1628 1629 |
if (want_affine && (tmp->flags & SD_WAKE_AFFINE) && cpumask_test_cpu(prev_cpu, sched_domain_span(tmp))) { affine_sd = tmp; want_affine = 0; |
c88d59108 sched: Merge sele... |
1630 |
} |
29cd8bae3 sched: Fix SD_POW... |
1631 1632 |
if (!want_sd && !want_affine) break; |
0763a660a sched: Rename sel... |
1633 |
if (!(tmp->flags & sd_flag)) |
c88d59108 sched: Merge sele... |
1634 |
continue; |
29cd8bae3 sched: Fix SD_POW... |
1635 1636 1637 |
if (want_sd) sd = tmp; } |
8b911acdf sched: Fix select... |
1638 |
if (affine_sd) { |
99bd5e2f2 sched: Fix select... |
1639 |
if (cpu == prev_cpu || wake_affine(affine_sd, p, sync)) |
dce840a08 sched: Dynamicall... |
1640 1641 1642 1643 |
prev_cpu = cpu; new_cpu = select_idle_sibling(p, prev_cpu); goto unlock; |
8b911acdf sched: Fix select... |
1644 |
} |
e7693a362 sched: de-SCHED_O... |
1645 |
|
aaee1203c sched: Move sched... |
1646 |
while (sd) { |
5158f4e44 sched: Clean up t... |
1647 |
int load_idx = sd->forkexec_idx; |
aaee1203c sched: Move sched... |
1648 |
struct sched_group *group; |
c88d59108 sched: Merge sele... |
1649 |
int weight; |
098fb9db2 sched: clean up w... |
1650 |
|
0763a660a sched: Rename sel... |
1651 |
if (!(sd->flags & sd_flag)) { |
aaee1203c sched: Move sched... |
1652 1653 1654 |
sd = sd->child; continue; } |
098fb9db2 sched: clean up w... |
1655 |
|
5158f4e44 sched: Clean up t... |
1656 1657 |
if (sd_flag & SD_BALANCE_WAKE) load_idx = sd->wake_idx; |
098fb9db2 sched: clean up w... |
1658 |
|
5158f4e44 sched: Clean up t... |
1659 |
group = find_idlest_group(sd, p, cpu, load_idx); |
aaee1203c sched: Move sched... |
1660 1661 1662 1663 |
if (!group) { sd = sd->child; continue; } |
4ae7d5cef sched: improve af... |
1664 |
|
d7c33c493 sched: Fix task a... |
1665 |
new_cpu = find_idlest_cpu(group, p, cpu); |
aaee1203c sched: Move sched... |
1666 1667 1668 1669 |
if (new_cpu == -1 || new_cpu == cpu) { /* Now try balancing at a lower domain level of cpu */ sd = sd->child; continue; |
e7693a362 sched: de-SCHED_O... |
1670 |
} |
aaee1203c sched: Move sched... |
1671 1672 1673 |
/* Now try balancing at a lower domain level of new_cpu */ cpu = new_cpu; |
669c55e9f sched: Pre-comput... |
1674 |
weight = sd->span_weight; |
aaee1203c sched: Move sched... |
1675 1676 |
sd = NULL; for_each_domain(cpu, tmp) { |
669c55e9f sched: Pre-comput... |
1677 |
if (weight <= tmp->span_weight) |
aaee1203c sched: Move sched... |
1678 |
break; |
0763a660a sched: Rename sel... |
1679 |
if (tmp->flags & sd_flag) |
aaee1203c sched: Move sched... |
1680 1681 1682 |
sd = tmp; } /* while loop will break here if sd == NULL */ |
e7693a362 sched: de-SCHED_O... |
1683 |
} |
dce840a08 sched: Dynamicall... |
1684 1685 |
unlock: rcu_read_unlock(); |
e7693a362 sched: de-SCHED_O... |
1686 |
|
c88d59108 sched: Merge sele... |
1687 |
return new_cpu; |
e7693a362 sched: de-SCHED_O... |
1688 1689 |
} #endif /* CONFIG_SMP */ |
e52fb7c09 sched: prefer wakers |
1690 1691 |
static unsigned long wakeup_gran(struct sched_entity *curr, struct sched_entity *se) |
0bbd3336e sched: fix wakeup... |
1692 1693 1694 1695 |
{ unsigned long gran = sysctl_sched_wakeup_granularity; /* |
e52fb7c09 sched: prefer wakers |
1696 1697 |
* Since its curr running now, convert the gran from real-time * to virtual-time in his units. |
13814d42e sched: Remove ASY... |
1698 1699 1700 1701 1702 1703 1704 1705 1706 |
* * By using 'se' instead of 'curr' we penalize light tasks, so * they get preempted easier. That is, if 'se' < 'curr' then * the resulting gran will be larger, therefore penalizing the * lighter, if otoh 'se' > 'curr' then the resulting gran will * be smaller, again penalizing the lighter task. * * This is especially important for buddies when the leftmost * task is higher priority than the buddy. |
0bbd3336e sched: fix wakeup... |
1707 |
*/ |
f4ad9bd20 sched: Eliminate ... |
1708 |
return calc_delta_fair(gran, se); |
0bbd3336e sched: fix wakeup... |
1709 1710 1711 |
} /* |
464b75273 sched: re-instate... |
1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 |
* Should 'se' preempt 'curr'. * * |s1 * |s2 * |s3 * g * |<--->|c * * w(c, s1) = -1 * w(c, s2) = 0 * w(c, s3) = 1 * */ static int wakeup_preempt_entity(struct sched_entity *curr, struct sched_entity *se) { s64 gran, vdiff = curr->vruntime - se->vruntime; if (vdiff <= 0) return -1; |
e52fb7c09 sched: prefer wakers |
1732 |
gran = wakeup_gran(curr, se); |
464b75273 sched: re-instate... |
1733 1734 1735 1736 1737 |
if (vdiff > gran) return 1; return 0; } |
02479099c sched: fix buddie... |
1738 1739 |
static void set_last_buddy(struct sched_entity *se) { |
69c80f3e9 sched: Make set_*... |
1740 1741 1742 1743 1744 |
if (entity_is_task(se) && unlikely(task_of(se)->policy == SCHED_IDLE)) return; for_each_sched_entity(se) cfs_rq_of(se)->last = se; |
02479099c sched: fix buddie... |
1745 1746 1747 1748 |
} static void set_next_buddy(struct sched_entity *se) { |
69c80f3e9 sched: Make set_*... |
1749 1750 1751 1752 1753 |
if (entity_is_task(se) && unlikely(task_of(se)->policy == SCHED_IDLE)) return; for_each_sched_entity(se) cfs_rq_of(se)->next = se; |
02479099c sched: fix buddie... |
1754 |
} |
ac53db596 sched: Use a budd... |
1755 1756 |
static void set_skip_buddy(struct sched_entity *se) { |
69c80f3e9 sched: Make set_*... |
1757 1758 |
for_each_sched_entity(se) cfs_rq_of(se)->skip = se; |
ac53db596 sched: Use a budd... |
1759 |
} |
464b75273 sched: re-instate... |
1760 |
/* |
bf0f6f24a sched: cfs core, ... |
1761 1762 |
* Preempt the current task with a newly woken task if needed: */ |
5a9b86f64 sched: Rename fla... |
1763 |
static void check_preempt_wakeup(struct rq *rq, struct task_struct *p, int wake_flags) |
bf0f6f24a sched: cfs core, ... |
1764 1765 |
{ struct task_struct *curr = rq->curr; |
8651a86c3 sched: group sche... |
1766 |
struct sched_entity *se = &curr->se, *pse = &p->se; |
03e89e457 sched: fix wakeup... |
1767 |
struct cfs_rq *cfs_rq = task_cfs_rq(curr); |
f685ceaca sched: Strengthen... |
1768 |
int scale = cfs_rq->nr_running >= sched_nr_latency; |
2f36825b1 sched: Next buddy... |
1769 |
int next_buddy_marked = 0; |
bf0f6f24a sched: cfs core, ... |
1770 |
|
4ae7d5cef sched: improve af... |
1771 1772 |
if (unlikely(se == pse)) return; |
2f36825b1 sched: Next buddy... |
1773 |
if (sched_feat(NEXT_BUDDY) && scale && !(wake_flags & WF_FORK)) { |
3cb63d527 sched: Complete b... |
1774 |
set_next_buddy(pse); |
2f36825b1 sched: Next buddy... |
1775 1776 |
next_buddy_marked = 1; } |
57fdc26d4 sched: fixup budd... |
1777 |
|
aec0a5142 sched: call resch... |
1778 1779 1780 1781 1782 1783 |
/* * We can come here with TIF_NEED_RESCHED already set from new task * wake up path. */ if (test_tsk_need_resched(curr)) return; |
a2f5c9ab7 sched: Allow SCHE... |
1784 1785 1786 1787 |
/* Idle tasks are by definition preempted by non-idle tasks. */ if (unlikely(curr->policy == SCHED_IDLE) && likely(p->policy != SCHED_IDLE)) goto preempt; |
91c234b4e sched: do not wak... |
1788 |
/* |
a2f5c9ab7 sched: Allow SCHE... |
1789 1790 |
* Batch and idle tasks do not preempt non-idle tasks (their preemption * is driven by the tick): |
91c234b4e sched: do not wak... |
1791 |
*/ |
6bc912b71 sched: SCHED_OTHE... |
1792 |
if (unlikely(p->policy != SCHED_NORMAL)) |
91c234b4e sched: do not wak... |
1793 |
return; |
bf0f6f24a sched: cfs core, ... |
1794 |
|
bf0f6f24a sched: cfs core, ... |
1795 |
|
ad4b78bbc sched: Add new wa... |
1796 1797 |
if (!sched_feat(WAKEUP_PREEMPT)) return; |
464b75273 sched: re-instate... |
1798 |
find_matching_se(&se, &pse); |
9bbd73743 sched: update cor... |
1799 |
update_curr(cfs_rq_of(se)); |
002f128b4 sched: remove red... |
1800 |
BUG_ON(!pse); |
2f36825b1 sched: Next buddy... |
1801 1802 1803 1804 1805 1806 1807 |
if (wakeup_preempt_entity(se, pse) == 1) { /* * Bias pick_next to pick the sched entity that is * triggering this preemption. */ if (!next_buddy_marked) set_next_buddy(pse); |
3a7e73a2e sched: Clean up c... |
1808 |
goto preempt; |
2f36825b1 sched: Next buddy... |
1809 |
} |
464b75273 sched: re-instate... |
1810 |
|
3a7e73a2e sched: Clean up c... |
1811 |
return; |
a65ac745e sched: Move updat... |
1812 |
|
3a7e73a2e sched: Clean up c... |
1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 |
preempt: resched_task(curr); /* * Only set the backward buddy when the current task is still * on the rq. This can happen when a wakeup gets interleaved * with schedule on the ->pre_schedule() or idle_balance() * point, either of which can * drop the rq lock. * * Also, during early boot the idle thread is in the fair class, * for obvious reasons its a bad idea to schedule back to it. */ if (unlikely(!se->on_rq || curr == rq->idle)) return; if (sched_feat(LAST_BUDDY) && scale && entity_is_task(se)) set_last_buddy(se); |
bf0f6f24a sched: cfs core, ... |
1829 |
} |
fb8d47240 sched: remove the... |
1830 |
static struct task_struct *pick_next_task_fair(struct rq *rq) |
bf0f6f24a sched: cfs core, ... |
1831 |
{ |
8f4d37ec0 sched: high-res p... |
1832 |
struct task_struct *p; |
bf0f6f24a sched: cfs core, ... |
1833 1834 |
struct cfs_rq *cfs_rq = &rq->cfs; struct sched_entity *se; |
36ace27e3 sched: Optimize b... |
1835 |
if (!cfs_rq->nr_running) |
bf0f6f24a sched: cfs core, ... |
1836 1837 1838 |
return NULL; do { |
9948f4b2a sched: remove the... |
1839 |
se = pick_next_entity(cfs_rq); |
f4b6755fb sched: cleanup fa... |
1840 |
set_next_entity(cfs_rq, se); |
bf0f6f24a sched: cfs core, ... |
1841 1842 |
cfs_rq = group_cfs_rq(se); } while (cfs_rq); |
8f4d37ec0 sched: high-res p... |
1843 1844 1845 1846 |
p = task_of(se); hrtick_start_fair(rq, p); return p; |
bf0f6f24a sched: cfs core, ... |
1847 1848 1849 1850 1851 |
} /* * Account for a descheduled task: */ |
31ee529cc sched: remove the... |
1852 |
static void put_prev_task_fair(struct rq *rq, struct task_struct *prev) |
bf0f6f24a sched: cfs core, ... |
1853 1854 1855 1856 1857 1858 |
{ struct sched_entity *se = &prev->se; struct cfs_rq *cfs_rq; for_each_sched_entity(se) { cfs_rq = cfs_rq_of(se); |
ab6cde269 sched: remove the... |
1859 |
put_prev_entity(cfs_rq, se); |
bf0f6f24a sched: cfs core, ... |
1860 1861 |
} } |
ac53db596 sched: Use a budd... |
1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 |
/* * sched_yield() is very simple * * The magic of dealing with the ->skip buddy is in pick_next_entity. */ static void yield_task_fair(struct rq *rq) { struct task_struct *curr = rq->curr; struct cfs_rq *cfs_rq = task_cfs_rq(curr); struct sched_entity *se = &curr->se; /* * Are we the only task in the tree? */ if (unlikely(rq->nr_running == 1)) return; clear_buddies(cfs_rq, se); if (curr->policy != SCHED_BATCH) { update_rq_clock(rq); /* * Update run-time statistics of the 'current'. */ update_curr(cfs_rq); } set_skip_buddy(se); } |
d95f41220 sched: Add yield_... |
1891 1892 1893 1894 1895 1896 1897 1898 1899 |
static bool yield_to_task_fair(struct rq *rq, struct task_struct *p, bool preempt) { struct sched_entity *se = &p->se; if (!se->on_rq) return false; /* Tell the scheduler that we'd really like pse to run next. */ set_next_buddy(se); |
d95f41220 sched: Add yield_... |
1900 1901 1902 1903 |
yield_task_fair(rq); return true; } |
681f3e685 sched: isolate SM... |
1904 |
#ifdef CONFIG_SMP |
bf0f6f24a sched: cfs core, ... |
1905 1906 1907 |
/************************************************** * Fair scheduling class load-balancing methods: */ |
1e3c88bde sched: Move load ... |
1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 |
/* * pull_task - move a task from a remote runqueue to the local runqueue. * Both runqueues must be locked. */ static void pull_task(struct rq *src_rq, struct task_struct *p, struct rq *this_rq, int this_cpu) { deactivate_task(src_rq, p, 0); set_task_cpu(p, this_cpu); activate_task(this_rq, p, 0); check_preempt_curr(this_rq, p, 0); } /* * can_migrate_task - may task p from runqueue rq be migrated to this_cpu? */ static int can_migrate_task(struct task_struct *p, struct rq *rq, int this_cpu, struct sched_domain *sd, enum cpu_idle_type idle, int *all_pinned) { int tsk_cache_hot = 0; /* * We do not migrate tasks that are: * 1) running (obviously), or * 2) cannot be migrated to this CPU due to cpus_allowed, or * 3) are cache-hot on their current CPU. */ if (!cpumask_test_cpu(this_cpu, &p->cpus_allowed)) { |
41acab885 sched: Implement ... |
1937 |
schedstat_inc(p, se.statistics.nr_failed_migrations_affine); |
1e3c88bde sched: Move load ... |
1938 1939 1940 1941 1942 |
return 0; } *all_pinned = 0; if (task_running(rq, p)) { |
41acab885 sched: Implement ... |
1943 |
schedstat_inc(p, se.statistics.nr_failed_migrations_running); |
1e3c88bde sched: Move load ... |
1944 1945 1946 1947 1948 1949 1950 1951 |
return 0; } /* * Aggressive migration if: * 1) task is cache cold, or * 2) too many balance attempts have failed. */ |
305e6835e sched: Do not acc... |
1952 |
tsk_cache_hot = task_hot(p, rq->clock_task, sd); |
1e3c88bde sched: Move load ... |
1953 1954 1955 1956 1957 |
if (!tsk_cache_hot || sd->nr_balance_failed > sd->cache_nice_tries) { #ifdef CONFIG_SCHEDSTATS if (tsk_cache_hot) { schedstat_inc(sd, lb_hot_gained[idle]); |
41acab885 sched: Implement ... |
1958 |
schedstat_inc(p, se.statistics.nr_forced_migrations); |
1e3c88bde sched: Move load ... |
1959 1960 1961 1962 1963 1964 |
} #endif return 1; } if (tsk_cache_hot) { |
41acab885 sched: Implement ... |
1965 |
schedstat_inc(p, se.statistics.nr_failed_migrations_hot); |
1e3c88bde sched: Move load ... |
1966 1967 1968 1969 |
return 0; } return 1; } |
897c395f4 sched: Remove rq_... |
1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 |
/* * move_one_task tries to move exactly one task from busiest to this_rq, as * part of active balancing operations within "domain". * Returns 1 if successful and 0 otherwise. * * Called with both runqueues locked. */ static int move_one_task(struct rq *this_rq, int this_cpu, struct rq *busiest, struct sched_domain *sd, enum cpu_idle_type idle) { struct task_struct *p, *n; struct cfs_rq *cfs_rq; int pinned = 0; for_each_leaf_cfs_rq(busiest, cfs_rq) { list_for_each_entry_safe(p, n, &cfs_rq->tasks, se.group_node) { if (!can_migrate_task(p, busiest, this_cpu, sd, idle, &pinned)) continue; pull_task(busiest, p, this_rq, this_cpu); /* * Right now, this is only the second place pull_task() * is called, so we can safely collect pull_task() * stats here rather than inside pull_task(). */ schedstat_inc(sd, lb_gained[idle]); return 1; } } return 0; } |
1e3c88bde sched: Move load ... |
2005 2006 2007 2008 |
static unsigned long balance_tasks(struct rq *this_rq, int this_cpu, struct rq *busiest, unsigned long max_load_move, struct sched_domain *sd, enum cpu_idle_type idle, int *all_pinned, |
931aeeda0 sched: Remove unu... |
2009 |
struct cfs_rq *busiest_cfs_rq) |
1e3c88bde sched: Move load ... |
2010 |
{ |
b30aef17f sched: Fix errone... |
2011 |
int loops = 0, pulled = 0; |
1e3c88bde sched: Move load ... |
2012 |
long rem_load_move = max_load_move; |
ee00e66ff sched: Remove rq_... |
2013 |
struct task_struct *p, *n; |
1e3c88bde sched: Move load ... |
2014 2015 2016 |
if (max_load_move == 0) goto out; |
ee00e66ff sched: Remove rq_... |
2017 2018 2019 |
list_for_each_entry_safe(p, n, &busiest_cfs_rq->tasks, se.group_node) { if (loops++ > sysctl_sched_nr_migrate) break; |
1e3c88bde sched: Move load ... |
2020 |
|
ee00e66ff sched: Remove rq_... |
2021 |
if ((p->se.load.weight >> 1) > rem_load_move || |
b30aef17f sched: Fix errone... |
2022 2023 |
!can_migrate_task(p, busiest, this_cpu, sd, idle, all_pinned)) |
ee00e66ff sched: Remove rq_... |
2024 |
continue; |
1e3c88bde sched: Move load ... |
2025 |
|
ee00e66ff sched: Remove rq_... |
2026 2027 2028 |
pull_task(busiest, p, this_rq, this_cpu); pulled++; rem_load_move -= p->se.load.weight; |
1e3c88bde sched: Move load ... |
2029 2030 |
#ifdef CONFIG_PREEMPT |
ee00e66ff sched: Remove rq_... |
2031 2032 2033 2034 2035 2036 2037 |
/* * NEWIDLE balancing is a source of latency, so preemptible * kernels will stop after the first task is pulled to minimize * the critical section. */ if (idle == CPU_NEWLY_IDLE) break; |
1e3c88bde sched: Move load ... |
2038 |
#endif |
ee00e66ff sched: Remove rq_... |
2039 2040 2041 2042 2043 2044 |
/* * We only want to steal up to the prescribed amount of * weighted load. */ if (rem_load_move <= 0) break; |
1e3c88bde sched: Move load ... |
2045 2046 2047 2048 2049 2050 2051 2052 |
} out: /* * Right now, this is one of only two places pull_task() is called, * so we can safely collect pull_task() stats here rather than * inside pull_task(). */ schedstat_add(sd, lb_gained[idle], pulled); |
1e3c88bde sched: Move load ... |
2053 2054 |
return max_load_move - rem_load_move; } |
230059de7 sched: Remove fro... |
2055 |
#ifdef CONFIG_FAIR_GROUP_SCHED |
9e3081ca6 sched: Make tg_sh... |
2056 2057 2058 |
/* * update tg->load_weight by folding this cpu's load_avg */ |
67e86250f sched: Introduce ... |
2059 |
static int update_shares_cpu(struct task_group *tg, int cpu) |
9e3081ca6 sched: Make tg_sh... |
2060 2061 2062 2063 |
{ struct cfs_rq *cfs_rq; unsigned long flags; struct rq *rq; |
9e3081ca6 sched: Make tg_sh... |
2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 |
if (!tg->se[cpu]) return 0; rq = cpu_rq(cpu); cfs_rq = tg->cfs_rq[cpu]; raw_spin_lock_irqsave(&rq->lock, flags); update_rq_clock(rq); |
d6b559182 sched: Allow upda... |
2074 |
update_cfs_load(cfs_rq, 1); |
9e3081ca6 sched: Make tg_sh... |
2075 2076 2077 2078 2079 |
/* * We need to update shares after updating tg->load_weight in * order to adjust the weight of groups with long running tasks. */ |
6d5ab2932 sched: Simplify u... |
2080 |
update_cfs_shares(cfs_rq); |
9e3081ca6 sched: Make tg_sh... |
2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 |
raw_spin_unlock_irqrestore(&rq->lock, flags); return 0; } static void update_shares(int cpu) { struct cfs_rq *cfs_rq; struct rq *rq = cpu_rq(cpu); rcu_read_lock(); |
9763b67fb sched, cgroup: Op... |
2093 2094 2095 2096 |
/* * Iterates the task_group tree in a bottom up fashion, see * list_add_leaf_cfs_rq() for details. */ |
67e86250f sched: Introduce ... |
2097 2098 |
for_each_leaf_cfs_rq(rq, cfs_rq) update_shares_cpu(cfs_rq->tg, cpu); |
9e3081ca6 sched: Make tg_sh... |
2099 2100 |
rcu_read_unlock(); } |
9763b67fb sched, cgroup: Op... |
2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 |
/* * Compute the cpu's hierarchical load factor for each task group. * This needs to be done in a top-down fashion because the load of a child * group is a fraction of its parents load. */ static int tg_load_down(struct task_group *tg, void *data) { unsigned long load; long cpu = (long)data; if (!tg->parent) { load = cpu_rq(cpu)->load.weight; } else { load = tg->parent->cfs_rq[cpu]->h_load; load *= tg->se[cpu]->load.weight; load /= tg->parent->cfs_rq[cpu]->load.weight + 1; } tg->cfs_rq[cpu]->h_load = load; return 0; } static void update_h_load(long cpu) { walk_tg_tree(tg_load_down, tg_nop, (void *)cpu); } |
230059de7 sched: Remove fro... |
2128 2129 2130 2131 |
static unsigned long load_balance_fair(struct rq *this_rq, int this_cpu, struct rq *busiest, unsigned long max_load_move, struct sched_domain *sd, enum cpu_idle_type idle, |
931aeeda0 sched: Remove unu... |
2132 |
int *all_pinned) |
230059de7 sched: Remove fro... |
2133 2134 |
{ long rem_load_move = max_load_move; |
9763b67fb sched, cgroup: Op... |
2135 |
struct cfs_rq *busiest_cfs_rq; |
230059de7 sched: Remove fro... |
2136 2137 |
rcu_read_lock(); |
9763b67fb sched, cgroup: Op... |
2138 |
update_h_load(cpu_of(busiest)); |
230059de7 sched: Remove fro... |
2139 |
|
9763b67fb sched, cgroup: Op... |
2140 |
for_each_leaf_cfs_rq(busiest, busiest_cfs_rq) { |
230059de7 sched: Remove fro... |
2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 |
unsigned long busiest_h_load = busiest_cfs_rq->h_load; unsigned long busiest_weight = busiest_cfs_rq->load.weight; u64 rem_load, moved_load; /* * empty group */ if (!busiest_cfs_rq->task_weight) continue; rem_load = (u64)rem_load_move * busiest_weight; rem_load = div_u64(rem_load, busiest_h_load + 1); moved_load = balance_tasks(this_rq, this_cpu, busiest, |
931aeeda0 sched: Remove unu... |
2155 |
rem_load, sd, idle, all_pinned, |
230059de7 sched: Remove fro... |
2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 |
busiest_cfs_rq); if (!moved_load) continue; moved_load *= busiest_h_load; moved_load = div_u64(moved_load, busiest_weight + 1); rem_load_move -= moved_load; if (rem_load_move < 0) break; } rcu_read_unlock(); return max_load_move - rem_load_move; } #else |
9e3081ca6 sched: Make tg_sh... |
2173 2174 2175 |
static inline void update_shares(int cpu) { } |
230059de7 sched: Remove fro... |
2176 2177 2178 2179 |
static unsigned long load_balance_fair(struct rq *this_rq, int this_cpu, struct rq *busiest, unsigned long max_load_move, struct sched_domain *sd, enum cpu_idle_type idle, |
931aeeda0 sched: Remove unu... |
2180 |
int *all_pinned) |
230059de7 sched: Remove fro... |
2181 2182 2183 |
{ return balance_tasks(this_rq, this_cpu, busiest, max_load_move, sd, idle, all_pinned, |
931aeeda0 sched: Remove unu... |
2184 |
&busiest->cfs); |
230059de7 sched: Remove fro... |
2185 2186 |
} #endif |
1e3c88bde sched: Move load ... |
2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 |
/* * move_tasks tries to move up to max_load_move weighted load from busiest to * this_rq, as part of a balancing operation within domain "sd". * Returns 1 if successful and 0 otherwise. * * Called with both runqueues locked. */ static int move_tasks(struct rq *this_rq, int this_cpu, struct rq *busiest, unsigned long max_load_move, struct sched_domain *sd, enum cpu_idle_type idle, int *all_pinned) { |
3d45fd804 sched: Remove the... |
2199 |
unsigned long total_load_moved = 0, load_moved; |
1e3c88bde sched: Move load ... |
2200 2201 |
do { |
3d45fd804 sched: Remove the... |
2202 |
load_moved = load_balance_fair(this_rq, this_cpu, busiest, |
1e3c88bde sched: Move load ... |
2203 |
max_load_move - total_load_moved, |
931aeeda0 sched: Remove unu... |
2204 |
sd, idle, all_pinned); |
3d45fd804 sched: Remove the... |
2205 2206 |
total_load_moved += load_moved; |
1e3c88bde sched: Move load ... |
2207 2208 2209 2210 2211 2212 2213 2214 2215 |
#ifdef CONFIG_PREEMPT /* * NEWIDLE balancing is a source of latency, so preemptible * kernels will stop after the first task is pulled to minimize * the critical section. */ if (idle == CPU_NEWLY_IDLE && this_rq->nr_running) break; |
baa8c1102 sched: Add a lock... |
2216 2217 2218 2219 |
if (raw_spin_is_contended(&this_rq->lock) || raw_spin_is_contended(&busiest->lock)) break; |
1e3c88bde sched: Move load ... |
2220 |
#endif |
3d45fd804 sched: Remove the... |
2221 |
} while (load_moved && max_load_move > total_load_moved); |
1e3c88bde sched: Move load ... |
2222 2223 2224 |
return total_load_moved > 0; } |
1e3c88bde sched: Move load ... |
2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 |
/********** Helpers for find_busiest_group ************************/ /* * sd_lb_stats - Structure to store the statistics of a sched_domain * during load balancing. */ struct sd_lb_stats { struct sched_group *busiest; /* Busiest group in this sd */ struct sched_group *this; /* Local group in this sd */ unsigned long total_load; /* Total load of all groups in sd */ unsigned long total_pwr; /* Total power of all groups in sd */ unsigned long avg_load; /* Average load across all groups in sd */ /** Statistics of this group */ unsigned long this_load; unsigned long this_load_per_task; unsigned long this_nr_running; |
fab476228 sched: Force bala... |
2241 |
unsigned long this_has_capacity; |
aae6d3ddd sched: Use group ... |
2242 |
unsigned int this_idle_cpus; |
1e3c88bde sched: Move load ... |
2243 2244 |
/* Statistics of the busiest group */ |
aae6d3ddd sched: Use group ... |
2245 |
unsigned int busiest_idle_cpus; |
1e3c88bde sched: Move load ... |
2246 2247 2248 |
unsigned long max_load; unsigned long busiest_load_per_task; unsigned long busiest_nr_running; |
dd5feea14 sched: Fix SCHED_... |
2249 |
unsigned long busiest_group_capacity; |
fab476228 sched: Force bala... |
2250 |
unsigned long busiest_has_capacity; |
aae6d3ddd sched: Use group ... |
2251 |
unsigned int busiest_group_weight; |
1e3c88bde sched: Move load ... |
2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 |
int group_imb; /* Is there imbalance in this sd */ #if defined(CONFIG_SCHED_MC) || defined(CONFIG_SCHED_SMT) int power_savings_balance; /* Is powersave balance needed for this sd */ struct sched_group *group_min; /* Least loaded group in sd */ struct sched_group *group_leader; /* Group which relieves group_min */ unsigned long min_load_per_task; /* load_per_task in group_min */ unsigned long leader_nr_running; /* Nr running of group_leader */ unsigned long min_nr_running; /* Nr running of group_min */ #endif }; /* * sg_lb_stats - stats of a sched_group required for load_balancing */ struct sg_lb_stats { unsigned long avg_load; /*Avg load across the CPUs of the group */ unsigned long group_load; /* Total load over the CPUs of the group */ unsigned long sum_nr_running; /* Nr tasks running in the group */ unsigned long sum_weighted_load; /* Weighted load of group's tasks */ unsigned long group_capacity; |
aae6d3ddd sched: Use group ... |
2273 2274 |
unsigned long idle_cpus; unsigned long group_weight; |
1e3c88bde sched: Move load ... |
2275 |
int group_imb; /* Is there an imbalance in the group ? */ |
fab476228 sched: Force bala... |
2276 |
int group_has_capacity; /* Is there extra capacity in the group? */ |
1e3c88bde sched: Move load ... |
2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 |
}; /** * group_first_cpu - Returns the first cpu in the cpumask of a sched_group. * @group: The group whose first cpu is to be returned. */ static inline unsigned int group_first_cpu(struct sched_group *group) { return cpumask_first(sched_group_cpus(group)); } /** * get_sd_load_idx - Obtain the load index for a given sched domain. * @sd: The sched_domain whose load_idx is to be obtained. * @idle: The Idle status of the CPU for whose sd load_icx is obtained. */ static inline int get_sd_load_idx(struct sched_domain *sd, enum cpu_idle_type idle) { int load_idx; switch (idle) { case CPU_NOT_IDLE: load_idx = sd->busy_idx; break; case CPU_NEWLY_IDLE: load_idx = sd->newidle_idx; break; default: load_idx = sd->idle_idx; break; } return load_idx; } #if defined(CONFIG_SCHED_MC) || defined(CONFIG_SCHED_SMT) /** * init_sd_power_savings_stats - Initialize power savings statistics for * the given sched_domain, during load balancing. * * @sd: Sched domain whose power-savings statistics are to be initialized. * @sds: Variable containing the statistics for sd. * @idle: Idle status of the CPU at which we're performing load-balancing. */ static inline void init_sd_power_savings_stats(struct sched_domain *sd, struct sd_lb_stats *sds, enum cpu_idle_type idle) { /* * Busy processors will not participate in power savings * balance. */ if (idle == CPU_NOT_IDLE || !(sd->flags & SD_POWERSAVINGS_BALANCE)) sds->power_savings_balance = 0; else { sds->power_savings_balance = 1; sds->min_nr_running = ULONG_MAX; sds->leader_nr_running = 0; } } /** * update_sd_power_savings_stats - Update the power saving stats for a * sched_domain while performing load balancing. * * @group: sched_group belonging to the sched_domain under consideration. * @sds: Variable containing the statistics of the sched_domain * @local_group: Does group contain the CPU for which we're performing * load balancing ? * @sgs: Variable containing the statistics of the group. */ static inline void update_sd_power_savings_stats(struct sched_group *group, struct sd_lb_stats *sds, int local_group, struct sg_lb_stats *sgs) { if (!sds->power_savings_balance) return; /* * If the local group is idle or completely loaded * no need to do power savings balance at this domain */ if (local_group && (sds->this_nr_running >= sgs->group_capacity || !sds->this_nr_running)) sds->power_savings_balance = 0; /* * If a group is already running at full capacity or idle, * don't include that group in power savings calculations */ if (!sds->power_savings_balance || sgs->sum_nr_running >= sgs->group_capacity || !sgs->sum_nr_running) return; /* * Calculate the group which has the least non-idle load. * This is the group from where we need to pick up the load * for saving power */ if ((sgs->sum_nr_running < sds->min_nr_running) || (sgs->sum_nr_running == sds->min_nr_running && group_first_cpu(group) > group_first_cpu(sds->group_min))) { sds->group_min = group; sds->min_nr_running = sgs->sum_nr_running; sds->min_load_per_task = sgs->sum_weighted_load / sgs->sum_nr_running; } /* * Calculate the group which is almost near its * capacity but still has some space to pick up some load * from other group and save more power */ if (sgs->sum_nr_running + 1 > sgs->group_capacity) return; if (sgs->sum_nr_running > sds->leader_nr_running || (sgs->sum_nr_running == sds->leader_nr_running && group_first_cpu(group) < group_first_cpu(sds->group_leader))) { sds->group_leader = group; sds->leader_nr_running = sgs->sum_nr_running; } } /** * check_power_save_busiest_group - see if there is potential for some power-savings balance * @sds: Variable containing the statistics of the sched_domain * under consideration. * @this_cpu: Cpu at which we're currently performing load-balancing. * @imbalance: Variable to store the imbalance. * * Description: * Check if we have potential to perform some power-savings balance. * If yes, set the busiest group to be the least loaded group in the * sched_domain, so that it's CPUs can be put to idle. * * Returns 1 if there is potential to perform power-savings balance. * Else returns 0. */ static inline int check_power_save_busiest_group(struct sd_lb_stats *sds, int this_cpu, unsigned long *imbalance) { if (!sds->power_savings_balance) return 0; if (sds->this != sds->group_leader || sds->group_leader == sds->group_min) return 0; *imbalance = sds->min_load_per_task; sds->busiest = sds->group_min; return 1; } #else /* CONFIG_SCHED_MC || CONFIG_SCHED_SMT */ static inline void init_sd_power_savings_stats(struct sched_domain *sd, struct sd_lb_stats *sds, enum cpu_idle_type idle) { return; } static inline void update_sd_power_savings_stats(struct sched_group *group, struct sd_lb_stats *sds, int local_group, struct sg_lb_stats *sgs) { return; } static inline int check_power_save_busiest_group(struct sd_lb_stats *sds, int this_cpu, unsigned long *imbalance) { return 0; } #endif /* CONFIG_SCHED_MC || CONFIG_SCHED_SMT */ unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu) { |
1399fa780 sched: Introduce ... |
2458 |
return SCHED_POWER_SCALE; |
1e3c88bde sched: Move load ... |
2459 2460 2461 2462 2463 2464 2465 2466 2467 |
} unsigned long __weak arch_scale_freq_power(struct sched_domain *sd, int cpu) { return default_scale_freq_power(sd, cpu); } unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu) { |
669c55e9f sched: Pre-comput... |
2468 |
unsigned long weight = sd->span_weight; |
1e3c88bde sched: Move load ... |
2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 |
unsigned long smt_gain = sd->smt_gain; smt_gain /= weight; return smt_gain; } unsigned long __weak arch_scale_smt_power(struct sched_domain *sd, int cpu) { return default_scale_smt_power(sd, cpu); } unsigned long scale_rt_power(int cpu) { struct rq *rq = cpu_rq(cpu); u64 total, available; |
1e3c88bde sched: Move load ... |
2485 |
total = sched_avg_period() + (rq->clock - rq->age_stamp); |
aa4838085 sched: Remove irq... |
2486 2487 2488 2489 2490 2491 2492 |
if (unlikely(total < rq->rt_avg)) { /* Ensures that power won't end up being negative */ available = 0; } else { available = total - rq->rt_avg; } |
1e3c88bde sched: Move load ... |
2493 |
|
1399fa780 sched: Introduce ... |
2494 2495 |
if (unlikely((s64)total < SCHED_POWER_SCALE)) total = SCHED_POWER_SCALE; |
1e3c88bde sched: Move load ... |
2496 |
|
1399fa780 sched: Introduce ... |
2497 |
total >>= SCHED_POWER_SHIFT; |
1e3c88bde sched: Move load ... |
2498 2499 2500 2501 2502 2503 |
return div_u64(available, total); } static void update_cpu_power(struct sched_domain *sd, int cpu) { |
669c55e9f sched: Pre-comput... |
2504 |
unsigned long weight = sd->span_weight; |
1399fa780 sched: Introduce ... |
2505 |
unsigned long power = SCHED_POWER_SCALE; |
1e3c88bde sched: Move load ... |
2506 |
struct sched_group *sdg = sd->groups; |
1e3c88bde sched: Move load ... |
2507 2508 2509 2510 2511 |
if ((sd->flags & SD_SHARE_CPUPOWER) && weight > 1) { if (sched_feat(ARCH_POWER)) power *= arch_scale_smt_power(sd, cpu); else power *= default_scale_smt_power(sd, cpu); |
1399fa780 sched: Introduce ... |
2512 |
power >>= SCHED_POWER_SHIFT; |
1e3c88bde sched: Move load ... |
2513 |
} |
9c3f75cbd sched: Break out ... |
2514 |
sdg->sgp->power_orig = power; |
9d5efe05e sched: Fix capaci... |
2515 2516 2517 2518 2519 |
if (sched_feat(ARCH_POWER)) power *= arch_scale_freq_power(sd, cpu); else power *= default_scale_freq_power(sd, cpu); |
1399fa780 sched: Introduce ... |
2520 |
power >>= SCHED_POWER_SHIFT; |
9d5efe05e sched: Fix capaci... |
2521 |
|
1e3c88bde sched: Move load ... |
2522 |
power *= scale_rt_power(cpu); |
1399fa780 sched: Introduce ... |
2523 |
power >>= SCHED_POWER_SHIFT; |
1e3c88bde sched: Move load ... |
2524 2525 2526 |
if (!power) power = 1; |
e51fd5e22 sched: Fix wake_a... |
2527 |
cpu_rq(cpu)->cpu_power = power; |
9c3f75cbd sched: Break out ... |
2528 |
sdg->sgp->power = power; |
1e3c88bde sched: Move load ... |
2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 |
} static void update_group_power(struct sched_domain *sd, int cpu) { struct sched_domain *child = sd->child; struct sched_group *group, *sdg = sd->groups; unsigned long power; if (!child) { update_cpu_power(sd, cpu); return; } power = 0; group = child->groups; do { |
9c3f75cbd sched: Break out ... |
2546 |
power += group->sgp->power; |
1e3c88bde sched: Move load ... |
2547 2548 |
group = group->next; } while (group != child->groups); |
9c3f75cbd sched: Break out ... |
2549 |
sdg->sgp->power = power; |
1e3c88bde sched: Move load ... |
2550 |
} |
9d5efe05e sched: Fix capaci... |
2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 |
/* * Try and fix up capacity for tiny siblings, this is needed when * things like SD_ASYM_PACKING need f_b_g to select another sibling * which on its own isn't powerful enough. * * See update_sd_pick_busiest() and check_asym_packing(). */ static inline int fix_small_capacity(struct sched_domain *sd, struct sched_group *group) { /* |
1399fa780 sched: Introduce ... |
2562 |
* Only siblings can have significantly less than SCHED_POWER_SCALE |
9d5efe05e sched: Fix capaci... |
2563 |
*/ |
a6c75f2f8 sched: Avoid usin... |
2564 |
if (!(sd->flags & SD_SHARE_CPUPOWER)) |
9d5efe05e sched: Fix capaci... |
2565 2566 2567 2568 2569 |
return 0; /* * If ~90% of the cpu_power is still there, we're good. */ |
9c3f75cbd sched: Break out ... |
2570 |
if (group->sgp->power * 32 > group->sgp->power_orig * 29) |
9d5efe05e sched: Fix capaci... |
2571 2572 2573 2574 |
return 1; return 0; } |
1e3c88bde sched: Move load ... |
2575 2576 2577 2578 2579 2580 2581 |
/** * update_sg_lb_stats - Update sched_group's statistics for load balancing. * @sd: The sched_domain whose statistics are to be updated. * @group: sched_group whose statistics are to be updated. * @this_cpu: Cpu for which load balance is currently performed. * @idle: Idle status of this_cpu * @load_idx: Load index of sched_domain of this_cpu for load calc. |
1e3c88bde sched: Move load ... |
2582 2583 2584 2585 2586 2587 2588 |
* @local_group: Does group contain this_cpu. * @cpus: Set of cpus considered for load balancing. * @balance: Should we balance. * @sgs: variable to hold the statistics for this group. */ static inline void update_sg_lb_stats(struct sched_domain *sd, struct sched_group *group, int this_cpu, |
46e49b383 sched: Wholesale ... |
2589 |
enum cpu_idle_type idle, int load_idx, |
1e3c88bde sched: Move load ... |
2590 2591 2592 |
int local_group, const struct cpumask *cpus, int *balance, struct sg_lb_stats *sgs) { |
2582f0eba sched: Set group_... |
2593 |
unsigned long load, max_cpu_load, min_cpu_load, max_nr_running; |
1e3c88bde sched: Move load ... |
2594 2595 |
int i; unsigned int balance_cpu = -1, first_idle_cpu = 0; |
dd5feea14 sched: Fix SCHED_... |
2596 |
unsigned long avg_load_per_task = 0; |
1e3c88bde sched: Move load ... |
2597 |
|
871e35bc9 sched: Fix the pl... |
2598 |
if (local_group) |
1e3c88bde sched: Move load ... |
2599 |
balance_cpu = group_first_cpu(group); |
1e3c88bde sched: Move load ... |
2600 2601 |
/* Tally up the load of all CPUs in the group */ |
1e3c88bde sched: Move load ... |
2602 2603 |
max_cpu_load = 0; min_cpu_load = ~0UL; |
2582f0eba sched: Set group_... |
2604 |
max_nr_running = 0; |
1e3c88bde sched: Move load ... |
2605 2606 2607 |
for_each_cpu_and(i, sched_group_cpus(group), cpus) { struct rq *rq = cpu_rq(i); |
1e3c88bde sched: Move load ... |
2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 |
/* Bias balancing toward cpus of our domain */ if (local_group) { if (idle_cpu(i) && !first_idle_cpu) { first_idle_cpu = 1; balance_cpu = i; } load = target_load(i, load_idx); } else { load = source_load(i, load_idx); |
2582f0eba sched: Set group_... |
2618 |
if (load > max_cpu_load) { |
1e3c88bde sched: Move load ... |
2619 |
max_cpu_load = load; |
2582f0eba sched: Set group_... |
2620 2621 |
max_nr_running = rq->nr_running; } |
1e3c88bde sched: Move load ... |
2622 2623 2624 2625 2626 2627 2628 |
if (min_cpu_load > load) min_cpu_load = load; } sgs->group_load += load; sgs->sum_nr_running += rq->nr_running; sgs->sum_weighted_load += weighted_cpuload(i); |
aae6d3ddd sched: Use group ... |
2629 2630 |
if (idle_cpu(i)) sgs->idle_cpus++; |
1e3c88bde sched: Move load ... |
2631 2632 2633 2634 2635 2636 2637 2638 |
} /* * First idle cpu or the first cpu(busiest) in this sched group * is eligible for doing load balancing at this and above * domains. In the newly idle case, we will allow all the cpu's * to do the newly idle load balance. */ |
bbc8cb5ba sched: Reduce upd... |
2639 2640 2641 2642 2643 2644 |
if (idle != CPU_NEWLY_IDLE && local_group) { if (balance_cpu != this_cpu) { *balance = 0; return; } update_group_power(sd, this_cpu); |
1e3c88bde sched: Move load ... |
2645 2646 2647 |
} /* Adjust by relative CPU power of the group */ |
9c3f75cbd sched: Break out ... |
2648 |
sgs->avg_load = (sgs->group_load*SCHED_POWER_SCALE) / group->sgp->power; |
1e3c88bde sched: Move load ... |
2649 |
|
1e3c88bde sched: Move load ... |
2650 2651 |
/* * Consider the group unbalanced when the imbalance is larger |
866ab43ef sched: Fix the gr... |
2652 |
* than the average weight of a task. |
1e3c88bde sched: Move load ... |
2653 2654 2655 2656 2657 2658 |
* * APZ: with cgroup the avg task weight can vary wildly and * might not be a suitable number - should we keep a * normalized nr_running number somewhere that negates * the hierarchy? */ |
dd5feea14 sched: Fix SCHED_... |
2659 2660 |
if (sgs->sum_nr_running) avg_load_per_task = sgs->sum_weighted_load / sgs->sum_nr_running; |
1e3c88bde sched: Move load ... |
2661 |
|
866ab43ef sched: Fix the gr... |
2662 |
if ((max_cpu_load - min_cpu_load) >= avg_load_per_task && max_nr_running > 1) |
1e3c88bde sched: Move load ... |
2663 |
sgs->group_imb = 1; |
9c3f75cbd sched: Break out ... |
2664 |
sgs->group_capacity = DIV_ROUND_CLOSEST(group->sgp->power, |
1399fa780 sched: Introduce ... |
2665 |
SCHED_POWER_SCALE); |
9d5efe05e sched: Fix capaci... |
2666 2667 |
if (!sgs->group_capacity) sgs->group_capacity = fix_small_capacity(sd, group); |
aae6d3ddd sched: Use group ... |
2668 |
sgs->group_weight = group->group_weight; |
fab476228 sched: Force bala... |
2669 2670 2671 |
if (sgs->group_capacity > sgs->sum_nr_running) sgs->group_has_capacity = 1; |
1e3c88bde sched: Move load ... |
2672 2673 2674 |
} /** |
532cb4c40 sched: Add asymme... |
2675 2676 2677 2678 |
* update_sd_pick_busiest - return 1 on busiest group * @sd: sched_domain whose statistics are to be checked * @sds: sched_domain statistics * @sg: sched_group candidate to be checked for being the busiest |
b6b122944 sched: Fix commen... |
2679 2680 |
* @sgs: sched_group statistics * @this_cpu: the current cpu |
532cb4c40 sched: Add asymme... |
2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 |
* * Determine if @sg is a busier group than the previously selected * busiest group. */ static bool update_sd_pick_busiest(struct sched_domain *sd, struct sd_lb_stats *sds, struct sched_group *sg, struct sg_lb_stats *sgs, int this_cpu) { if (sgs->avg_load <= sds->max_load) return false; if (sgs->sum_nr_running > sgs->group_capacity) return true; if (sgs->group_imb) return true; /* * ASYM_PACKING needs to move all the work to the lowest * numbered CPUs in the group, therefore mark all groups * higher than ourself as busy. */ if ((sd->flags & SD_ASYM_PACKING) && sgs->sum_nr_running && this_cpu < group_first_cpu(sg)) { if (!sds->busiest) return true; if (group_first_cpu(sds->busiest) > group_first_cpu(sg)) return true; } return false; } /** |
1e3c88bde sched: Move load ... |
2718 2719 2720 2721 |
* update_sd_lb_stats - Update sched_group's statistics for load balancing. * @sd: sched_domain whose statistics are to be updated. * @this_cpu: Cpu for which load balance is currently performed. * @idle: Idle status of this_cpu |
1e3c88bde sched: Move load ... |
2722 2723 2724 2725 2726 |
* @cpus: Set of cpus considered for load balancing. * @balance: Should we balance. * @sds: variable to hold the statistics for this sched_domain. */ static inline void update_sd_lb_stats(struct sched_domain *sd, int this_cpu, |
46e49b383 sched: Wholesale ... |
2727 2728 |
enum cpu_idle_type idle, const struct cpumask *cpus, int *balance, struct sd_lb_stats *sds) |
1e3c88bde sched: Move load ... |
2729 2730 |
{ struct sched_domain *child = sd->child; |
532cb4c40 sched: Add asymme... |
2731 |
struct sched_group *sg = sd->groups; |
1e3c88bde sched: Move load ... |
2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 |
struct sg_lb_stats sgs; int load_idx, prefer_sibling = 0; if (child && child->flags & SD_PREFER_SIBLING) prefer_sibling = 1; init_sd_power_savings_stats(sd, sds, idle); load_idx = get_sd_load_idx(sd, idle); do { int local_group; |
532cb4c40 sched: Add asymme... |
2743 |
local_group = cpumask_test_cpu(this_cpu, sched_group_cpus(sg)); |
1e3c88bde sched: Move load ... |
2744 |
memset(&sgs, 0, sizeof(sgs)); |
46e49b383 sched: Wholesale ... |
2745 |
update_sg_lb_stats(sd, sg, this_cpu, idle, load_idx, |
1e3c88bde sched: Move load ... |
2746 |
local_group, cpus, balance, &sgs); |
8f190fb3f sched: Assume *ba... |
2747 |
if (local_group && !(*balance)) |
1e3c88bde sched: Move load ... |
2748 2749 2750 |
return; sds->total_load += sgs.group_load; |
9c3f75cbd sched: Break out ... |
2751 |
sds->total_pwr += sg->sgp->power; |
1e3c88bde sched: Move load ... |
2752 2753 2754 |
/* * In case the child domain prefers tasks go to siblings |
532cb4c40 sched: Add asymme... |
2755 |
* first, lower the sg capacity to one so that we'll try |
75dd321d7 sched: Drop group... |
2756 2757 2758 2759 2760 2761 |
* and move all the excess tasks away. We lower the capacity * of a group only if the local group has the capacity to fit * these excess tasks, i.e. nr_running < group_capacity. The * extra check prevents the case where you always pull from the * heaviest group when it is already under-utilized (possible * with a large weight task outweighs the tasks on the system). |
1e3c88bde sched: Move load ... |
2762 |
*/ |
75dd321d7 sched: Drop group... |
2763 |
if (prefer_sibling && !local_group && sds->this_has_capacity) |
1e3c88bde sched: Move load ... |
2764 2765 2766 2767 |
sgs.group_capacity = min(sgs.group_capacity, 1UL); if (local_group) { sds->this_load = sgs.avg_load; |
532cb4c40 sched: Add asymme... |
2768 |
sds->this = sg; |
1e3c88bde sched: Move load ... |
2769 2770 |
sds->this_nr_running = sgs.sum_nr_running; sds->this_load_per_task = sgs.sum_weighted_load; |
fab476228 sched: Force bala... |
2771 |
sds->this_has_capacity = sgs.group_has_capacity; |
aae6d3ddd sched: Use group ... |
2772 |
sds->this_idle_cpus = sgs.idle_cpus; |
532cb4c40 sched: Add asymme... |
2773 |
} else if (update_sd_pick_busiest(sd, sds, sg, &sgs, this_cpu)) { |
1e3c88bde sched: Move load ... |
2774 |
sds->max_load = sgs.avg_load; |
532cb4c40 sched: Add asymme... |
2775 |
sds->busiest = sg; |
1e3c88bde sched: Move load ... |
2776 |
sds->busiest_nr_running = sgs.sum_nr_running; |
aae6d3ddd sched: Use group ... |
2777 |
sds->busiest_idle_cpus = sgs.idle_cpus; |
dd5feea14 sched: Fix SCHED_... |
2778 |
sds->busiest_group_capacity = sgs.group_capacity; |
1e3c88bde sched: Move load ... |
2779 |
sds->busiest_load_per_task = sgs.sum_weighted_load; |
fab476228 sched: Force bala... |
2780 |
sds->busiest_has_capacity = sgs.group_has_capacity; |
aae6d3ddd sched: Use group ... |
2781 |
sds->busiest_group_weight = sgs.group_weight; |
1e3c88bde sched: Move load ... |
2782 2783 |
sds->group_imb = sgs.group_imb; } |
532cb4c40 sched: Add asymme... |
2784 2785 2786 2787 |
update_sd_power_savings_stats(sg, sds, local_group, &sgs); sg = sg->next; } while (sg != sd->groups); } |
2ec57d448 sched: Fix spelli... |
2788 |
int __weak arch_sd_sibling_asym_packing(void) |
532cb4c40 sched: Add asymme... |
2789 2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 |
{ return 0*SD_ASYM_PACKING; } /** * check_asym_packing - Check to see if the group is packed into the * sched doman. * * This is primarily intended to used at the sibling level. Some * cores like POWER7 prefer to use lower numbered SMT threads. In the * case of POWER7, it can move to lower SMT modes only when higher * threads are idle. When in lower SMT modes, the threads will * perform better since they share less core resources. Hence when we * have idle threads, we want them to be the higher ones. * * This packing function is run on idle threads. It checks to see if * the busiest CPU in this domain (core in the P7 case) has a higher * CPU number than the packing function is being run on. Here we are * assuming lower CPU number will be equivalent to lower a SMT thread * number. * |
b6b122944 sched: Fix commen... |
2810 2811 2812 |
* Returns 1 when packing is required and a task should be moved to * this CPU. The amount of the imbalance is returned in *imbalance. * |
532cb4c40 sched: Add asymme... |
2813 2814 2815 2816 |
* @sd: The sched_domain whose packing is to be checked. * @sds: Statistics of the sched_domain which is to be packed * @this_cpu: The cpu at whose sched_domain we're performing load-balance. * @imbalance: returns amount of imbalanced due to packing. |
532cb4c40 sched: Add asymme... |
2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 |
*/ static int check_asym_packing(struct sched_domain *sd, struct sd_lb_stats *sds, int this_cpu, unsigned long *imbalance) { int busiest_cpu; if (!(sd->flags & SD_ASYM_PACKING)) return 0; if (!sds->busiest) return 0; busiest_cpu = group_first_cpu(sds->busiest); if (this_cpu > busiest_cpu) return 0; |
9c3f75cbd sched: Break out ... |
2833 |
*imbalance = DIV_ROUND_CLOSEST(sds->max_load * sds->busiest->sgp->power, |
1399fa780 sched: Introduce ... |
2834 |
SCHED_POWER_SCALE); |
532cb4c40 sched: Add asymme... |
2835 |
return 1; |
1e3c88bde sched: Move load ... |
2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 |
} /** * fix_small_imbalance - Calculate the minor imbalance that exists * amongst the groups of a sched_domain, during * load balancing. * @sds: Statistics of the sched_domain whose imbalance is to be calculated. * @this_cpu: The cpu at whose sched_domain we're performing load-balance. * @imbalance: Variable to store the imbalance. */ static inline void fix_small_imbalance(struct sd_lb_stats *sds, int this_cpu, unsigned long *imbalance) { unsigned long tmp, pwr_now = 0, pwr_move = 0; unsigned int imbn = 2; |
dd5feea14 sched: Fix SCHED_... |
2851 |
unsigned long scaled_busy_load_per_task; |
1e3c88bde sched: Move load ... |
2852 2853 2854 2855 2856 2857 2858 2859 2860 |
if (sds->this_nr_running) { sds->this_load_per_task /= sds->this_nr_running; if (sds->busiest_load_per_task > sds->this_load_per_task) imbn = 1; } else sds->this_load_per_task = cpu_avg_load_per_task(this_cpu); |
dd5feea14 sched: Fix SCHED_... |
2861 |
scaled_busy_load_per_task = sds->busiest_load_per_task |
1399fa780 sched: Introduce ... |
2862 |
* SCHED_POWER_SCALE; |
9c3f75cbd sched: Break out ... |
2863 |
scaled_busy_load_per_task /= sds->busiest->sgp->power; |
dd5feea14 sched: Fix SCHED_... |
2864 2865 2866 |
if (sds->max_load - sds->this_load + scaled_busy_load_per_task >= (scaled_busy_load_per_task * imbn)) { |
1e3c88bde sched: Move load ... |
2867 2868 2869 2870 2871 2872 2873 2874 2875 |
*imbalance = sds->busiest_load_per_task; return; } /* * OK, we don't have enough imbalance to justify moving tasks, * however we may be able to increase total CPU power used by * moving them. */ |
9c3f75cbd sched: Break out ... |
2876 |
pwr_now += sds->busiest->sgp->power * |
1e3c88bde sched: Move load ... |
2877 |
min(sds->busiest_load_per_task, sds->max_load); |
9c3f75cbd sched: Break out ... |
2878 |
pwr_now += sds->this->sgp->power * |
1e3c88bde sched: Move load ... |
2879 |
min(sds->this_load_per_task, sds->this_load); |
1399fa780 sched: Introduce ... |
2880 |
pwr_now /= SCHED_POWER_SCALE; |
1e3c88bde sched: Move load ... |
2881 2882 |
/* Amount of load we'd subtract */ |
1399fa780 sched: Introduce ... |
2883 |
tmp = (sds->busiest_load_per_task * SCHED_POWER_SCALE) / |
9c3f75cbd sched: Break out ... |
2884 |
sds->busiest->sgp->power; |
1e3c88bde sched: Move load ... |
2885 |
if (sds->max_load > tmp) |
9c3f75cbd sched: Break out ... |
2886 |
pwr_move += sds->busiest->sgp->power * |
1e3c88bde sched: Move load ... |
2887 2888 2889 |
min(sds->busiest_load_per_task, sds->max_load - tmp); /* Amount of load we'd add */ |
9c3f75cbd sched: Break out ... |
2890 |
if (sds->max_load * sds->busiest->sgp->power < |
1399fa780 sched: Introduce ... |
2891 |
sds->busiest_load_per_task * SCHED_POWER_SCALE) |
9c3f75cbd sched: Break out ... |
2892 2893 |
tmp = (sds->max_load * sds->busiest->sgp->power) / sds->this->sgp->power; |
1e3c88bde sched: Move load ... |
2894 |
else |
1399fa780 sched: Introduce ... |
2895 |
tmp = (sds->busiest_load_per_task * SCHED_POWER_SCALE) / |
9c3f75cbd sched: Break out ... |
2896 2897 |
sds->this->sgp->power; pwr_move += sds->this->sgp->power * |
1e3c88bde sched: Move load ... |
2898 |
min(sds->this_load_per_task, sds->this_load + tmp); |
1399fa780 sched: Introduce ... |
2899 |
pwr_move /= SCHED_POWER_SCALE; |
1e3c88bde sched: Move load ... |
2900 2901 2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 |
/* Move if we gain throughput */ if (pwr_move > pwr_now) *imbalance = sds->busiest_load_per_task; } /** * calculate_imbalance - Calculate the amount of imbalance present within the * groups of a given sched_domain during load balance. * @sds: statistics of the sched_domain whose imbalance is to be calculated. * @this_cpu: Cpu for which currently load balance is being performed. * @imbalance: The variable to store the imbalance. */ static inline void calculate_imbalance(struct sd_lb_stats *sds, int this_cpu, unsigned long *imbalance) { |
dd5feea14 sched: Fix SCHED_... |
2916 2917 2918 2919 2920 2921 2922 |
unsigned long max_pull, load_above_capacity = ~0UL; sds->busiest_load_per_task /= sds->busiest_nr_running; if (sds->group_imb) { sds->busiest_load_per_task = min(sds->busiest_load_per_task, sds->avg_load); } |
1e3c88bde sched: Move load ... |
2923 2924 2925 2926 2927 2928 2929 2930 2931 |
/* * In the presence of smp nice balancing, certain scenarios can have * max load less than avg load(as we skip the groups at or below * its cpu_power, while calculating max_load..) */ if (sds->max_load < sds->avg_load) { *imbalance = 0; return fix_small_imbalance(sds, this_cpu, imbalance); } |
dd5feea14 sched: Fix SCHED_... |
2932 2933 2934 2935 2936 2937 |
if (!sds->group_imb) { /* * Don't want to pull so many tasks that a group would go idle. */ load_above_capacity = (sds->busiest_nr_running - sds->busiest_group_capacity); |
1399fa780 sched: Introduce ... |
2938 |
load_above_capacity *= (SCHED_LOAD_SCALE * SCHED_POWER_SCALE); |
dd5feea14 sched: Fix SCHED_... |
2939 |
|
9c3f75cbd sched: Break out ... |
2940 |
load_above_capacity /= sds->busiest->sgp->power; |
dd5feea14 sched: Fix SCHED_... |
2941 2942 2943 2944 2945 2946 2947 2948 2949 2950 2951 2952 2953 |
} /* * We're trying to get all the cpus to the average_load, so we don't * want to push ourselves above the average load, nor do we wish to * reduce the max loaded cpu below the average load. At the same time, * we also don't want to reduce the group load below the group capacity * (so that we can implement power-savings policies etc). Thus we look * for the minimum possible imbalance. * Be careful of negative numbers as they'll appear as very large values * with unsigned longs. */ max_pull = min(sds->max_load - sds->avg_load, load_above_capacity); |
1e3c88bde sched: Move load ... |
2954 2955 |
/* How much load to actually move to equalise the imbalance */ |
9c3f75cbd sched: Break out ... |
2956 2957 |
*imbalance = min(max_pull * sds->busiest->sgp->power, (sds->avg_load - sds->this_load) * sds->this->sgp->power) |
1399fa780 sched: Introduce ... |
2958 |
/ SCHED_POWER_SCALE; |
1e3c88bde sched: Move load ... |
2959 2960 2961 |
/* * if *imbalance is less than the average load per runnable task |
25985edce Fix common misspe... |
2962 |
* there is no guarantee that any tasks will be moved so we'll have |
1e3c88bde sched: Move load ... |
2963 2964 2965 2966 2967 2968 2969 |
* a think about bumping its value to force at least one task to be * moved */ if (*imbalance < sds->busiest_load_per_task) return fix_small_imbalance(sds, this_cpu, imbalance); } |
fab476228 sched: Force bala... |
2970 |
|
1e3c88bde sched: Move load ... |
2971 2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 |
/******* find_busiest_group() helpers end here *********************/ /** * find_busiest_group - Returns the busiest group within the sched_domain * if there is an imbalance. If there isn't an imbalance, and * the user has opted for power-savings, it returns a group whose * CPUs can be put to idle by rebalancing those tasks elsewhere, if * such a group exists. * * Also calculates the amount of weighted load which should be moved * to restore balance. * * @sd: The sched_domain whose busiest group is to be returned. * @this_cpu: The cpu for which load balancing is currently being performed. * @imbalance: Variable which stores amount of weighted load which should * be moved to restore balance/put a group to idle. * @idle: The idle status of this_cpu. |
1e3c88bde sched: Move load ... |
2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 |
* @cpus: The set of CPUs under consideration for load-balancing. * @balance: Pointer to a variable indicating if this_cpu * is the appropriate cpu to perform load balancing at this_level. * * Returns: - the busiest group if imbalance exists. * - If no imbalance and user has opted for power-savings balance, * return the least loaded group whose CPUs can be * put to idle by rebalancing its tasks onto our group. */ static struct sched_group * find_busiest_group(struct sched_domain *sd, int this_cpu, unsigned long *imbalance, enum cpu_idle_type idle, |
46e49b383 sched: Wholesale ... |
3000 |
const struct cpumask *cpus, int *balance) |
1e3c88bde sched: Move load ... |
3001 3002 3003 3004 3005 3006 3007 3008 3009 |
{ struct sd_lb_stats sds; memset(&sds, 0, sizeof(sds)); /* * Compute the various statistics relavent for load balancing at * this level. */ |
46e49b383 sched: Wholesale ... |
3010 |
update_sd_lb_stats(sd, this_cpu, idle, cpus, balance, &sds); |
1e3c88bde sched: Move load ... |
3011 |
|
cc57aa8f4 sched: Clean up s... |
3012 3013 3014 |
/* * this_cpu is not the appropriate cpu to perform load balancing at * this level. |
1e3c88bde sched: Move load ... |
3015 |
*/ |
8f190fb3f sched: Assume *ba... |
3016 |
if (!(*balance)) |
1e3c88bde sched: Move load ... |
3017 |
goto ret; |
532cb4c40 sched: Add asymme... |
3018 3019 3020 |
if ((idle == CPU_IDLE || idle == CPU_NEWLY_IDLE) && check_asym_packing(sd, &sds, this_cpu, imbalance)) return sds.busiest; |
cc57aa8f4 sched: Clean up s... |
3021 |
/* There is no busy sibling group to pull tasks from */ |
1e3c88bde sched: Move load ... |
3022 3023 |
if (!sds.busiest || sds.busiest_nr_running == 0) goto out_balanced; |
1399fa780 sched: Introduce ... |
3024 |
sds.avg_load = (SCHED_POWER_SCALE * sds.total_load) / sds.total_pwr; |
b0432d8f1 sched: Fix sched-... |
3025 |
|
866ab43ef sched: Fix the gr... |
3026 3027 3028 3029 3030 3031 3032 |
/* * If the busiest group is imbalanced the below checks don't * work because they assumes all things are equal, which typically * isn't true due to cpus_allowed constraints and the like. */ if (sds.group_imb) goto force_balance; |
cc57aa8f4 sched: Clean up s... |
3033 |
/* SD_BALANCE_NEWIDLE trumps SMP nice when underutilized */ |
fab476228 sched: Force bala... |
3034 3035 3036 |
if (idle == CPU_NEWLY_IDLE && sds.this_has_capacity && !sds.busiest_has_capacity) goto force_balance; |
cc57aa8f4 sched: Clean up s... |
3037 3038 3039 3040 |
/* * If the local group is more busy than the selected busiest group * don't try and pull any tasks. */ |
1e3c88bde sched: Move load ... |
3041 3042 |
if (sds.this_load >= sds.max_load) goto out_balanced; |
cc57aa8f4 sched: Clean up s... |
3043 3044 3045 3046 |
/* * Don't pull any tasks if this group is already above the domain * average load. */ |
1e3c88bde sched: Move load ... |
3047 3048 |
if (sds.this_load >= sds.avg_load) goto out_balanced; |
c186fafe9 sched: Clean up r... |
3049 |
if (idle == CPU_IDLE) { |
aae6d3ddd sched: Use group ... |
3050 3051 3052 3053 3054 3055 |
/* * This cpu is idle. If the busiest group load doesn't * have more tasks than the number of available cpu's and * there is no imbalance between this and busiest group * wrt to idle cpu's, it is balanced. */ |
c186fafe9 sched: Clean up r... |
3056 |
if ((sds.this_idle_cpus <= sds.busiest_idle_cpus + 1) && |
aae6d3ddd sched: Use group ... |
3057 3058 |
sds.busiest_nr_running <= sds.busiest_group_weight) goto out_balanced; |
c186fafe9 sched: Clean up r... |
3059 3060 3061 3062 3063 3064 3065 |
} else { /* * In the CPU_NEWLY_IDLE, CPU_NOT_IDLE cases, use * imbalance_pct to be conservative. */ if (100 * sds.max_load <= sd->imbalance_pct * sds.this_load) goto out_balanced; |
aae6d3ddd sched: Use group ... |
3066 |
} |
1e3c88bde sched: Move load ... |
3067 |
|
fab476228 sched: Force bala... |
3068 |
force_balance: |
1e3c88bde sched: Move load ... |
3069 3070 3071 3072 3073 3074 3075 3076 3077 3078 3079 3080 3081 3082 3083 3084 3085 3086 3087 3088 |
/* Looks like there is an imbalance. Compute it */ calculate_imbalance(&sds, this_cpu, imbalance); return sds.busiest; out_balanced: /* * There is no obvious imbalance. But check if we can do some balancing * to save power. */ if (check_power_save_busiest_group(&sds, this_cpu, imbalance)) return sds.busiest; ret: *imbalance = 0; return NULL; } /* * find_busiest_queue - find the busiest runqueue among the cpus in group. */ static struct rq * |
9d5efe05e sched: Fix capaci... |
3089 3090 3091 |
find_busiest_queue(struct sched_domain *sd, struct sched_group *group, enum cpu_idle_type idle, unsigned long imbalance, const struct cpumask *cpus) |
1e3c88bde sched: Move load ... |
3092 3093 3094 3095 3096 3097 3098 |
{ struct rq *busiest = NULL, *rq; unsigned long max_load = 0; int i; for_each_cpu(i, sched_group_cpus(group)) { unsigned long power = power_of(i); |
1399fa780 sched: Introduce ... |
3099 3100 |
unsigned long capacity = DIV_ROUND_CLOSEST(power, SCHED_POWER_SCALE); |
1e3c88bde sched: Move load ... |
3101 |
unsigned long wl; |
9d5efe05e sched: Fix capaci... |
3102 3103 |
if (!capacity) capacity = fix_small_capacity(sd, group); |
1e3c88bde sched: Move load ... |
3104 3105 3106 3107 |
if (!cpumask_test_cpu(i, cpus)) continue; rq = cpu_rq(i); |
6e40f5bbb Merge branch 'sch... |
3108 |
wl = weighted_cpuload(i); |
1e3c88bde sched: Move load ... |
3109 |
|
6e40f5bbb Merge branch 'sch... |
3110 3111 3112 3113 |
/* * When comparing with imbalance, use weighted_cpuload() * which is not scaled with the cpu power. */ |
1e3c88bde sched: Move load ... |
3114 3115 |
if (capacity && rq->nr_running == 1 && wl > imbalance) continue; |
6e40f5bbb Merge branch 'sch... |
3116 3117 3118 3119 3120 3121 |
/* * For the load comparisons with the other cpu's, consider * the weighted_cpuload() scaled with the cpu power, so that * the load can be moved away from the cpu that is potentially * running at a lower capacity. */ |
1399fa780 sched: Introduce ... |
3122 |
wl = (wl * SCHED_POWER_SCALE) / power; |
6e40f5bbb Merge branch 'sch... |
3123 |
|
1e3c88bde sched: Move load ... |
3124 3125 3126 3127 3128 3129 3130 3131 3132 3133 3134 3135 3136 3137 3138 3139 3140 |
if (wl > max_load) { max_load = wl; busiest = rq; } } return busiest; } /* * Max backoff if we encounter pinned tasks. Pretty arbitrary value, but * so long as it is large enough. */ #define MAX_PINNED_INTERVAL 512 /* Working cpumask for load_balance and load_balance_newidle. */ static DEFINE_PER_CPU(cpumask_var_t, load_balance_tmpmask); |
46e49b383 sched: Wholesale ... |
3141 |
static int need_active_balance(struct sched_domain *sd, int idle, |
532cb4c40 sched: Add asymme... |
3142 |
int busiest_cpu, int this_cpu) |
1af3ed3dd sched: Unify load... |
3143 3144 |
{ if (idle == CPU_NEWLY_IDLE) { |
532cb4c40 sched: Add asymme... |
3145 3146 3147 3148 3149 3150 3151 3152 |
/* * ASYM_PACKING needs to force migrate tasks from busy but * higher numbered CPUs in order to pack all tasks in the * lowest numbered CPUs. */ if ((sd->flags & SD_ASYM_PACKING) && busiest_cpu > this_cpu) return 1; |
1af3ed3dd sched: Unify load... |
3153 3154 3155 3156 3157 3158 3159 3160 3161 3162 3163 3164 3165 3166 3167 3168 3169 3170 3171 |
/* * The only task running in a non-idle cpu can be moved to this * cpu in an attempt to completely freeup the other CPU * package. * * The package power saving logic comes from * find_busiest_group(). If there are no imbalance, then * f_b_g() will return NULL. However when sched_mc={1,2} then * f_b_g() will select a group from which a running task may be * pulled to this cpu in order to make the other package idle. * If there is no opportunity to make a package idle and if * there are no imbalance, then f_b_g() will return NULL and no * action will be taken in load_balance_newidle(). * * Under normal task pull operation due to imbalance, there * will be more than one task in the source run queue and * move_tasks() will succeed. ld_moved will be true and this * active balance code will not be triggered. */ |
1af3ed3dd sched: Unify load... |
3172 3173 3174 3175 3176 3177 |
if (sched_mc_power_savings < POWERSAVINGS_BALANCE_WAKEUP) return 0; } return unlikely(sd->nr_balance_failed > sd->cache_nice_tries+2); } |
969c79215 sched: replace mi... |
3178 |
static int active_load_balance_cpu_stop(void *data); |
1e3c88bde sched: Move load ... |
3179 3180 3181 3182 3183 3184 3185 3186 |
/* * Check this_cpu to ensure it is balanced within domain. Attempt to move * tasks if there is an imbalance. */ static int load_balance(int this_cpu, struct rq *this_rq, struct sched_domain *sd, enum cpu_idle_type idle, int *balance) { |
46e49b383 sched: Wholesale ... |
3187 |
int ld_moved, all_pinned = 0, active_balance = 0; |
1e3c88bde sched: Move load ... |
3188 3189 3190 3191 3192 3193 3194 |
struct sched_group *group; unsigned long imbalance; struct rq *busiest; unsigned long flags; struct cpumask *cpus = __get_cpu_var(load_balance_tmpmask); cpumask_copy(cpus, cpu_active_mask); |
1e3c88bde sched: Move load ... |
3195 3196 3197 |
schedstat_inc(sd, lb_count[idle]); redo: |
46e49b383 sched: Wholesale ... |
3198 |
group = find_busiest_group(sd, this_cpu, &imbalance, idle, |
1e3c88bde sched: Move load ... |
3199 3200 3201 3202 3203 3204 3205 3206 3207 |
cpus, balance); if (*balance == 0) goto out_balanced; if (!group) { schedstat_inc(sd, lb_nobusyg[idle]); goto out_balanced; } |
9d5efe05e sched: Fix capaci... |
3208 |
busiest = find_busiest_queue(sd, group, idle, imbalance, cpus); |
1e3c88bde sched: Move load ... |
3209 3210 3211 3212 3213 3214 3215 3216 3217 3218 3219 3220 3221 3222 3223 3224 3225 |
if (!busiest) { schedstat_inc(sd, lb_nobusyq[idle]); goto out_balanced; } BUG_ON(busiest == this_rq); schedstat_add(sd, lb_imbalance[idle], imbalance); ld_moved = 0; if (busiest->nr_running > 1) { /* * Attempt to move tasks. If find_busiest_group has found * an imbalance but busiest->nr_running <= 1, the group is * still unbalanced. ld_moved simply stays zero, so it is * correctly treated as an imbalance. */ |
b30aef17f sched: Fix errone... |
3226 |
all_pinned = 1; |
1e3c88bde sched: Move load ... |
3227 3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 |
local_irq_save(flags); double_rq_lock(this_rq, busiest); ld_moved = move_tasks(this_rq, this_cpu, busiest, imbalance, sd, idle, &all_pinned); double_rq_unlock(this_rq, busiest); local_irq_restore(flags); /* * some other cpu did the load balance for us. */ if (ld_moved && this_cpu != smp_processor_id()) resched_cpu(this_cpu); /* All tasks on this runqueue were pinned by CPU affinity */ if (unlikely(all_pinned)) { cpumask_clear_cpu(cpu_of(busiest), cpus); if (!cpumask_empty(cpus)) goto redo; goto out_balanced; } } if (!ld_moved) { schedstat_inc(sd, lb_failed[idle]); |
58b26c4c0 sched: Increment ... |
3251 3252 3253 3254 3255 3256 3257 3258 |
/* * Increment the failure counter only on periodic balance. * We do not want newidle balance, which can be very * frequent, pollute the failure counter causing * excessive cache_hot migrations and active balances. */ if (idle != CPU_NEWLY_IDLE) sd->nr_balance_failed++; |
1e3c88bde sched: Move load ... |
3259 |
|
46e49b383 sched: Wholesale ... |
3260 |
if (need_active_balance(sd, idle, cpu_of(busiest), this_cpu)) { |
1e3c88bde sched: Move load ... |
3261 |
raw_spin_lock_irqsave(&busiest->lock, flags); |
969c79215 sched: replace mi... |
3262 3263 3264 |
/* don't kick the active_load_balance_cpu_stop, * if the curr task on busiest cpu can't be * moved to this_cpu |
1e3c88bde sched: Move load ... |
3265 3266 3267 3268 3269 3270 3271 3272 |
*/ if (!cpumask_test_cpu(this_cpu, &busiest->curr->cpus_allowed)) { raw_spin_unlock_irqrestore(&busiest->lock, flags); all_pinned = 1; goto out_one_pinned; } |
969c79215 sched: replace mi... |
3273 3274 3275 3276 3277 |
/* * ->active_balance synchronizes accesses to * ->active_balance_work. Once set, it's cleared * only after active load balance is finished. */ |
1e3c88bde sched: Move load ... |
3278 3279 3280 3281 3282 3283 |
if (!busiest->active_balance) { busiest->active_balance = 1; busiest->push_cpu = this_cpu; active_balance = 1; } raw_spin_unlock_irqrestore(&busiest->lock, flags); |
969c79215 sched: replace mi... |
3284 |
|
1e3c88bde sched: Move load ... |
3285 |
if (active_balance) |
969c79215 sched: replace mi... |
3286 3287 3288 |
stop_one_cpu_nowait(cpu_of(busiest), active_load_balance_cpu_stop, busiest, &busiest->active_balance_work); |
1e3c88bde sched: Move load ... |
3289 3290 3291 3292 3293 3294 3295 3296 3297 3298 3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 |
/* * We've kicked active balancing, reset the failure * counter. */ sd->nr_balance_failed = sd->cache_nice_tries+1; } } else sd->nr_balance_failed = 0; if (likely(!active_balance)) { /* We were unbalanced, so reset the balancing interval */ sd->balance_interval = sd->min_interval; } else { /* * If we've begun active balancing, start to back off. This * case may not be covered by the all_pinned logic if there * is only 1 task on the busy runqueue (because we don't call * move_tasks). */ if (sd->balance_interval < sd->max_interval) sd->balance_interval *= 2; } |
1e3c88bde sched: Move load ... |
3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 |
goto out; out_balanced: schedstat_inc(sd, lb_balanced[idle]); sd->nr_balance_failed = 0; out_one_pinned: /* tune up the balancing interval */ if ((all_pinned && sd->balance_interval < MAX_PINNED_INTERVAL) || (sd->balance_interval < sd->max_interval)) sd->balance_interval *= 2; |
46e49b383 sched: Wholesale ... |
3324 |
ld_moved = 0; |
1e3c88bde sched: Move load ... |
3325 |
out: |
1e3c88bde sched: Move load ... |
3326 3327 3328 3329 |
return ld_moved; } /* |
1e3c88bde sched: Move load ... |
3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 |
* idle_balance is called by schedule() if this_cpu is about to become * idle. Attempts to pull tasks from other CPUs. */ static void idle_balance(int this_cpu, struct rq *this_rq) { struct sched_domain *sd; int pulled_task = 0; unsigned long next_balance = jiffies + HZ; this_rq->idle_stamp = this_rq->clock; if (this_rq->avg_idle < sysctl_sched_migration_cost) return; |
f492e12ef sched: Remove loa... |
3343 3344 3345 3346 |
/* * Drop the rq->lock, but keep IRQ/preempt disabled. */ raw_spin_unlock(&this_rq->lock); |
c66eaf619 sched: Update sha... |
3347 |
update_shares(this_cpu); |
dce840a08 sched: Dynamicall... |
3348 |
rcu_read_lock(); |
1e3c88bde sched: Move load ... |
3349 3350 |
for_each_domain(this_cpu, sd) { unsigned long interval; |
f492e12ef sched: Remove loa... |
3351 |
int balance = 1; |
1e3c88bde sched: Move load ... |
3352 3353 3354 |
if (!(sd->flags & SD_LOAD_BALANCE)) continue; |
f492e12ef sched: Remove loa... |
3355 |
if (sd->flags & SD_BALANCE_NEWIDLE) { |
1e3c88bde sched: Move load ... |
3356 |
/* If we've pulled tasks over stop searching: */ |
f492e12ef sched: Remove loa... |
3357 3358 3359 |
pulled_task = load_balance(this_cpu, this_rq, sd, CPU_NEWLY_IDLE, &balance); } |
1e3c88bde sched: Move load ... |
3360 3361 3362 3363 |
interval = msecs_to_jiffies(sd->balance_interval); if (time_after(next_balance, sd->last_balance + interval)) next_balance = sd->last_balance + interval; |
d5ad140bc sched: Fix idle b... |
3364 3365 |
if (pulled_task) { this_rq->idle_stamp = 0; |
1e3c88bde sched: Move load ... |
3366 |
break; |
d5ad140bc sched: Fix idle b... |
3367 |
} |
1e3c88bde sched: Move load ... |
3368 |
} |
dce840a08 sched: Dynamicall... |
3369 |
rcu_read_unlock(); |
f492e12ef sched: Remove loa... |
3370 3371 |
raw_spin_lock(&this_rq->lock); |
1e3c88bde sched: Move load ... |
3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 |
if (pulled_task || time_after(jiffies, this_rq->next_balance)) { /* * We are going idle. next_balance may be set based on * a busy processor. So reset next_balance. */ this_rq->next_balance = next_balance; } } /* |
969c79215 sched: replace mi... |
3382 3383 3384 3385 |
* active_load_balance_cpu_stop is run by cpu stopper. It pushes * running tasks off the busiest CPU onto idle CPUs. It requires at * least 1 task to be running on each physical CPU where possible, and * avoids physical / logical imbalances. |
1e3c88bde sched: Move load ... |
3386 |
*/ |
969c79215 sched: replace mi... |
3387 |
static int active_load_balance_cpu_stop(void *data) |
1e3c88bde sched: Move load ... |
3388 |
{ |
969c79215 sched: replace mi... |
3389 3390 |
struct rq *busiest_rq = data; int busiest_cpu = cpu_of(busiest_rq); |
1e3c88bde sched: Move load ... |
3391 |
int target_cpu = busiest_rq->push_cpu; |
969c79215 sched: replace mi... |
3392 |
struct rq *target_rq = cpu_rq(target_cpu); |
1e3c88bde sched: Move load ... |
3393 |
struct sched_domain *sd; |
969c79215 sched: replace mi... |
3394 3395 3396 3397 3398 3399 3400 |
raw_spin_lock_irq(&busiest_rq->lock); /* make sure the requested cpu hasn't gone down in the meantime */ if (unlikely(busiest_cpu != smp_processor_id() || !busiest_rq->active_balance)) goto out_unlock; |
1e3c88bde sched: Move load ... |
3401 3402 3403 |
/* Is there any task to move? */ if (busiest_rq->nr_running <= 1) |
969c79215 sched: replace mi... |
3404 |
goto out_unlock; |
1e3c88bde sched: Move load ... |
3405 3406 3407 3408 3409 3410 3411 3412 3413 3414 |
/* * This condition is "impossible", if it occurs * we need to fix it. Originally reported by * Bjorn Helgaas on a 128-cpu setup. */ BUG_ON(busiest_rq == target_rq); /* move a task from busiest_rq to target_rq */ double_lock_balance(busiest_rq, target_rq); |
1e3c88bde sched: Move load ... |
3415 3416 |
/* Search for an sd spanning us and the target CPU. */ |
dce840a08 sched: Dynamicall... |
3417 |
rcu_read_lock(); |
1e3c88bde sched: Move load ... |
3418 3419 3420 3421 3422 3423 3424 3425 3426 3427 3428 3429 3430 3431 3432 |
for_each_domain(target_cpu, sd) { if ((sd->flags & SD_LOAD_BALANCE) && cpumask_test_cpu(busiest_cpu, sched_domain_span(sd))) break; } if (likely(sd)) { schedstat_inc(sd, alb_count); if (move_one_task(target_rq, target_cpu, busiest_rq, sd, CPU_IDLE)) schedstat_inc(sd, alb_pushed); else schedstat_inc(sd, alb_failed); } |
dce840a08 sched: Dynamicall... |
3433 |
rcu_read_unlock(); |
1e3c88bde sched: Move load ... |
3434 |
double_unlock_balance(busiest_rq, target_rq); |
969c79215 sched: replace mi... |
3435 3436 3437 3438 |
out_unlock: busiest_rq->active_balance = 0; raw_spin_unlock_irq(&busiest_rq->lock); return 0; |
1e3c88bde sched: Move load ... |
3439 3440 3441 |
} #ifdef CONFIG_NO_HZ |
83cd4fe27 sched: Change noh... |
3442 3443 3444 3445 3446 3447 3448 3449 3450 3451 3452 3453 3454 3455 3456 3457 3458 3459 3460 3461 3462 3463 3464 3465 3466 3467 |
static DEFINE_PER_CPU(struct call_single_data, remote_sched_softirq_cb); static void trigger_sched_softirq(void *data) { raise_softirq_irqoff(SCHED_SOFTIRQ); } static inline void init_sched_softirq_csd(struct call_single_data *csd) { csd->func = trigger_sched_softirq; csd->info = NULL; csd->flags = 0; csd->priv = 0; } /* * idle load balancing details * - One of the idle CPUs nominates itself as idle load_balancer, while * entering idle. * - This idle load balancer CPU will also go into tickless mode when * it is idle, just like all other idle CPUs * - When one of the busy CPUs notice that there may be an idle rebalancing * needed, they will kick the idle load balancer, which then does idle * load balancing for all the idle CPUs. */ |
1e3c88bde sched: Move load ... |
3468 3469 |
static struct { atomic_t load_balancer; |
83cd4fe27 sched: Change noh... |
3470 3471 3472 3473 3474 3475 |
atomic_t first_pick_cpu; atomic_t second_pick_cpu; cpumask_var_t idle_cpus_mask; cpumask_var_t grp_idle_mask; unsigned long next_balance; /* in jiffy units */ } nohz ____cacheline_aligned; |
1e3c88bde sched: Move load ... |
3476 3477 3478 3479 3480 3481 3482 3483 3484 3485 3486 3487 3488 3489 3490 3491 3492 3493 3494 3495 3496 3497 3498 3499 3500 3501 3502 3503 3504 3505 3506 3507 3508 3509 3510 3511 3512 3513 3514 3515 3516 3517 3518 3519 3520 3521 3522 3523 3524 3525 3526 3527 3528 |
int get_nohz_load_balancer(void) { return atomic_read(&nohz.load_balancer); } #if defined(CONFIG_SCHED_MC) || defined(CONFIG_SCHED_SMT) /** * lowest_flag_domain - Return lowest sched_domain containing flag. * @cpu: The cpu whose lowest level of sched domain is to * be returned. * @flag: The flag to check for the lowest sched_domain * for the given cpu. * * Returns the lowest sched_domain of a cpu which contains the given flag. */ static inline struct sched_domain *lowest_flag_domain(int cpu, int flag) { struct sched_domain *sd; for_each_domain(cpu, sd) if (sd && (sd->flags & flag)) break; return sd; } /** * for_each_flag_domain - Iterates over sched_domains containing the flag. * @cpu: The cpu whose domains we're iterating over. * @sd: variable holding the value of the power_savings_sd * for cpu. * @flag: The flag to filter the sched_domains to be iterated. * * Iterates over all the scheduler domains for a given cpu that has the 'flag' * set, starting from the lowest sched_domain to the highest. */ #define for_each_flag_domain(cpu, sd, flag) \ for (sd = lowest_flag_domain(cpu, flag); \ (sd && (sd->flags & flag)); sd = sd->parent) /** * is_semi_idle_group - Checks if the given sched_group is semi-idle. * @ilb_group: group to be checked for semi-idleness * * Returns: 1 if the group is semi-idle. 0 otherwise. * * We define a sched_group to be semi idle if it has atleast one idle-CPU * and atleast one non-idle CPU. This helper function checks if the given * sched_group is semi-idle or not. */ static inline int is_semi_idle_group(struct sched_group *ilb_group) { |
83cd4fe27 sched: Change noh... |
3529 |
cpumask_and(nohz.grp_idle_mask, nohz.idle_cpus_mask, |
1e3c88bde sched: Move load ... |
3530 3531 3532 3533 3534 3535 |
sched_group_cpus(ilb_group)); /* * A sched_group is semi-idle when it has atleast one busy cpu * and atleast one idle cpu. */ |
83cd4fe27 sched: Change noh... |
3536 |
if (cpumask_empty(nohz.grp_idle_mask)) |
1e3c88bde sched: Move load ... |
3537 |
return 0; |
83cd4fe27 sched: Change noh... |
3538 |
if (cpumask_equal(nohz.grp_idle_mask, sched_group_cpus(ilb_group))) |
1e3c88bde sched: Move load ... |
3539 3540 3541 3542 3543 3544 3545 3546 3547 3548 3549 3550 3551 3552 3553 3554 3555 3556 3557 3558 |
return 0; return 1; } /** * find_new_ilb - Finds the optimum idle load balancer for nomination. * @cpu: The cpu which is nominating a new idle_load_balancer. * * Returns: Returns the id of the idle load balancer if it exists, * Else, returns >= nr_cpu_ids. * * This algorithm picks the idle load balancer such that it belongs to a * semi-idle powersavings sched_domain. The idea is to try and avoid * completely idle packages/cores just for the purpose of idle load balancing * when there are other idle cpu's which are better suited for that job. */ static int find_new_ilb(int cpu) { struct sched_domain *sd; struct sched_group *ilb_group; |
dce840a08 sched: Dynamicall... |
3559 |
int ilb = nr_cpu_ids; |
1e3c88bde sched: Move load ... |
3560 3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 3571 |
/* * Have idle load balancer selection from semi-idle packages only * when power-aware load balancing is enabled */ if (!(sched_smt_power_savings || sched_mc_power_savings)) goto out_done; /* * Optimize for the case when we have no idle CPUs or only one * idle CPU. Don't walk the sched_domain hierarchy in such cases */ |
83cd4fe27 sched: Change noh... |
3572 |
if (cpumask_weight(nohz.idle_cpus_mask) < 2) |
1e3c88bde sched: Move load ... |
3573 |
goto out_done; |
dce840a08 sched: Dynamicall... |
3574 |
rcu_read_lock(); |
1e3c88bde sched: Move load ... |
3575 3576 3577 3578 |
for_each_flag_domain(cpu, sd, SD_POWERSAVINGS_BALANCE) { ilb_group = sd->groups; do { |
dce840a08 sched: Dynamicall... |
3579 3580 3581 3582 |
if (is_semi_idle_group(ilb_group)) { ilb = cpumask_first(nohz.grp_idle_mask); goto unlock; } |
1e3c88bde sched: Move load ... |
3583 3584 3585 3586 3587 |
ilb_group = ilb_group->next; } while (ilb_group != sd->groups); } |
dce840a08 sched: Dynamicall... |
3588 3589 |
unlock: rcu_read_unlock(); |
1e3c88bde sched: Move load ... |
3590 3591 |
out_done: |
dce840a08 sched: Dynamicall... |
3592 |
return ilb; |
1e3c88bde sched: Move load ... |
3593 3594 3595 3596 |
} #else /* (CONFIG_SCHED_MC || CONFIG_SCHED_SMT) */ static inline int find_new_ilb(int call_cpu) { |
83cd4fe27 sched: Change noh... |
3597 |
return nr_cpu_ids; |
1e3c88bde sched: Move load ... |
3598 3599 3600 3601 |
} #endif /* |
83cd4fe27 sched: Change noh... |
3602 3603 3604 3605 3606 3607 3608 3609 3610 3611 3612 3613 3614 3615 3616 3617 3618 3619 3620 3621 3622 3623 3624 3625 3626 3627 3628 3629 3630 |
* Kick a CPU to do the nohz balancing, if it is time for it. We pick the * nohz_load_balancer CPU (if there is one) otherwise fallback to any idle * CPU (if there is one). */ static void nohz_balancer_kick(int cpu) { int ilb_cpu; nohz.next_balance++; ilb_cpu = get_nohz_load_balancer(); if (ilb_cpu >= nr_cpu_ids) { ilb_cpu = cpumask_first(nohz.idle_cpus_mask); if (ilb_cpu >= nr_cpu_ids) return; } if (!cpu_rq(ilb_cpu)->nohz_balance_kick) { struct call_single_data *cp; cpu_rq(ilb_cpu)->nohz_balance_kick = 1; cp = &per_cpu(remote_sched_softirq_cb, cpu); __smp_call_function_single(ilb_cpu, cp, 0); } return; } /* |
1e3c88bde sched: Move load ... |
3631 3632 |
* This routine will try to nominate the ilb (idle load balancing) * owner among the cpus whose ticks are stopped. ilb owner will do the idle |
83cd4fe27 sched: Change noh... |
3633 |
* load balancing on behalf of all those cpus. |
1e3c88bde sched: Move load ... |
3634 |
* |
83cd4fe27 sched: Change noh... |
3635 3636 3637 |
* When the ilb owner becomes busy, we will not have new ilb owner until some * idle CPU wakes up and goes back to idle or some busy CPU tries to kick * idle load balancing by kicking one of the idle CPUs. |
1e3c88bde sched: Move load ... |
3638 |
* |
83cd4fe27 sched: Change noh... |
3639 3640 3641 |
* Ticks are stopped for the ilb owner as well, with busy CPU kicking this * ilb owner CPU in future (when there is a need for idle load balancing on * behalf of all idle CPUs). |
1e3c88bde sched: Move load ... |
3642 |
*/ |
83cd4fe27 sched: Change noh... |
3643 |
void select_nohz_load_balancer(int stop_tick) |
1e3c88bde sched: Move load ... |
3644 3645 3646 3647 |
{ int cpu = smp_processor_id(); if (stop_tick) { |
1e3c88bde sched: Move load ... |
3648 3649 |
if (!cpu_active(cpu)) { if (atomic_read(&nohz.load_balancer) != cpu) |
83cd4fe27 sched: Change noh... |
3650 |
return; |
1e3c88bde sched: Move load ... |
3651 3652 3653 3654 3655 |
/* * If we are going offline and still the leader, * give up! */ |
83cd4fe27 sched: Change noh... |
3656 3657 |
if (atomic_cmpxchg(&nohz.load_balancer, cpu, nr_cpu_ids) != cpu) |
1e3c88bde sched: Move load ... |
3658 |
BUG(); |
83cd4fe27 sched: Change noh... |
3659 |
return; |
1e3c88bde sched: Move load ... |
3660 |
} |
83cd4fe27 sched: Change noh... |
3661 |
cpumask_set_cpu(cpu, nohz.idle_cpus_mask); |
1e3c88bde sched: Move load ... |
3662 |
|
83cd4fe27 sched: Change noh... |
3663 3664 3665 3666 |
if (atomic_read(&nohz.first_pick_cpu) == cpu) atomic_cmpxchg(&nohz.first_pick_cpu, cpu, nr_cpu_ids); if (atomic_read(&nohz.second_pick_cpu) == cpu) atomic_cmpxchg(&nohz.second_pick_cpu, cpu, nr_cpu_ids); |
1e3c88bde sched: Move load ... |
3667 |
|
83cd4fe27 sched: Change noh... |
3668 |
if (atomic_read(&nohz.load_balancer) >= nr_cpu_ids) { |
1e3c88bde sched: Move load ... |
3669 |
int new_ilb; |
83cd4fe27 sched: Change noh... |
3670 3671 3672 3673 |
/* make me the ilb owner */ if (atomic_cmpxchg(&nohz.load_balancer, nr_cpu_ids, cpu) != nr_cpu_ids) return; |
1e3c88bde sched: Move load ... |
3674 3675 3676 3677 3678 3679 |
/* * Check to see if there is a more power-efficient * ilb. */ new_ilb = find_new_ilb(cpu); if (new_ilb < nr_cpu_ids && new_ilb != cpu) { |
83cd4fe27 sched: Change noh... |
3680 |
atomic_set(&nohz.load_balancer, nr_cpu_ids); |
1e3c88bde sched: Move load ... |
3681 |
resched_cpu(new_ilb); |
83cd4fe27 sched: Change noh... |
3682 |
return; |
1e3c88bde sched: Move load ... |
3683 |
} |
83cd4fe27 sched: Change noh... |
3684 |
return; |
1e3c88bde sched: Move load ... |
3685 3686 |
} } else { |
83cd4fe27 sched: Change noh... |
3687 3688 |
if (!cpumask_test_cpu(cpu, nohz.idle_cpus_mask)) return; |
1e3c88bde sched: Move load ... |
3689 |
|
83cd4fe27 sched: Change noh... |
3690 |
cpumask_clear_cpu(cpu, nohz.idle_cpus_mask); |
1e3c88bde sched: Move load ... |
3691 3692 |
if (atomic_read(&nohz.load_balancer) == cpu) |
83cd4fe27 sched: Change noh... |
3693 3694 |
if (atomic_cmpxchg(&nohz.load_balancer, cpu, nr_cpu_ids) != cpu) |
1e3c88bde sched: Move load ... |
3695 3696 |
BUG(); } |
83cd4fe27 sched: Change noh... |
3697 |
return; |
1e3c88bde sched: Move load ... |
3698 3699 3700 3701 |
} #endif static DEFINE_SPINLOCK(balancing); |
49c022e65 sched: Clean up r... |
3702 3703 3704 3705 3706 3707 3708 3709 3710 3711 |
static unsigned long __read_mostly max_load_balance_interval = HZ/10; /* * Scale the max load_balance interval with the number of CPUs in the system. * This trades load-balance latency on larger machines for less cross talk. */ static void update_max_interval(void) { max_load_balance_interval = HZ*num_online_cpus()/10; } |
1e3c88bde sched: Move load ... |
3712 3713 3714 3715 3716 3717 3718 3719 3720 3721 3722 3723 3724 3725 3726 3727 |
/* * It checks each scheduling domain to see if it is due to be balanced, * and initiates a balancing operation if so. * * Balancing parameters are set up in arch_init_sched_domains. */ static void rebalance_domains(int cpu, enum cpu_idle_type idle) { int balance = 1; struct rq *rq = cpu_rq(cpu); unsigned long interval; struct sched_domain *sd; /* Earliest time when we have to do rebalance again */ unsigned long next_balance = jiffies + 60*HZ; int update_next_balance = 0; int need_serialize; |
2069dd75c sched: Rewrite tg... |
3728 |
update_shares(cpu); |
dce840a08 sched: Dynamicall... |
3729 |
rcu_read_lock(); |
1e3c88bde sched: Move load ... |
3730 3731 3732 3733 3734 3735 3736 3737 3738 3739 |
for_each_domain(cpu, sd) { if (!(sd->flags & SD_LOAD_BALANCE)) continue; interval = sd->balance_interval; if (idle != CPU_IDLE) interval *= sd->busy_factor; /* scale ms to jiffies */ interval = msecs_to_jiffies(interval); |
49c022e65 sched: Clean up r... |
3740 |
interval = clamp(interval, 1UL, max_load_balance_interval); |
1e3c88bde sched: Move load ... |
3741 3742 3743 3744 3745 3746 3747 3748 3749 3750 3751 3752 |
need_serialize = sd->flags & SD_SERIALIZE; if (need_serialize) { if (!spin_trylock(&balancing)) goto out; } if (time_after_eq(jiffies, sd->last_balance + interval)) { if (load_balance(cpu, rq, sd, idle, &balance)) { /* * We've pulled tasks over so either we're no |
c186fafe9 sched: Clean up r... |
3753 |
* longer idle. |
1e3c88bde sched: Move load ... |
3754 3755 3756 3757 3758 3759 3760 3761 3762 3763 3764 3765 3766 3767 3768 3769 3770 3771 3772 3773 3774 |
*/ idle = CPU_NOT_IDLE; } sd->last_balance = jiffies; } if (need_serialize) spin_unlock(&balancing); out: if (time_after(next_balance, sd->last_balance + interval)) { next_balance = sd->last_balance + interval; update_next_balance = 1; } /* * Stop the load balance at this level. There is another * CPU in our sched group which is doing load balancing more * actively. */ if (!balance) break; } |
dce840a08 sched: Dynamicall... |
3775 |
rcu_read_unlock(); |
1e3c88bde sched: Move load ... |
3776 3777 3778 3779 3780 3781 3782 3783 3784 |
/* * next_balance will be updated only when there is a need. * When the cpu is attached to null domain for ex, it will not be * updated. */ if (likely(update_next_balance)) rq->next_balance = next_balance; } |
83cd4fe27 sched: Change noh... |
3785 |
#ifdef CONFIG_NO_HZ |
1e3c88bde sched: Move load ... |
3786 |
/* |
83cd4fe27 sched: Change noh... |
3787 |
* In CONFIG_NO_HZ case, the idle balance kickee will do the |
1e3c88bde sched: Move load ... |
3788 3789 |
* rebalancing for all the cpus for whom scheduler ticks are stopped. */ |
83cd4fe27 sched: Change noh... |
3790 3791 3792 3793 3794 3795 3796 3797 3798 3799 3800 3801 3802 3803 3804 3805 3806 3807 3808 3809 3810 3811 3812 3813 |
static void nohz_idle_balance(int this_cpu, enum cpu_idle_type idle) { struct rq *this_rq = cpu_rq(this_cpu); struct rq *rq; int balance_cpu; if (idle != CPU_IDLE || !this_rq->nohz_balance_kick) return; for_each_cpu(balance_cpu, nohz.idle_cpus_mask) { if (balance_cpu == this_cpu) continue; /* * If this cpu gets work to do, stop the load balancing * work being done for other cpus. Next load * balancing owner will pick it up. */ if (need_resched()) { this_rq->nohz_balance_kick = 0; break; } raw_spin_lock_irq(&this_rq->lock); |
5343bdb8f sched: Update rq-... |
3814 |
update_rq_clock(this_rq); |
83cd4fe27 sched: Change noh... |
3815 3816 3817 3818 3819 3820 3821 3822 3823 3824 3825 3826 3827 3828 3829 3830 3831 3832 3833 3834 3835 3836 3837 3838 3839 3840 3841 3842 3843 3844 3845 3846 3847 |
update_cpu_load(this_rq); raw_spin_unlock_irq(&this_rq->lock); rebalance_domains(balance_cpu, CPU_IDLE); rq = cpu_rq(balance_cpu); if (time_after(this_rq->next_balance, rq->next_balance)) this_rq->next_balance = rq->next_balance; } nohz.next_balance = this_rq->next_balance; this_rq->nohz_balance_kick = 0; } /* * Current heuristic for kicking the idle load balancer * - first_pick_cpu is the one of the busy CPUs. It will kick * idle load balancer when it has more than one process active. This * eliminates the need for idle load balancing altogether when we have * only one running process in the system (common case). * - If there are more than one busy CPU, idle load balancer may have * to run for active_load_balance to happen (i.e., two busy CPUs are * SMT or core siblings and can run better if they move to different * physical CPUs). So, second_pick_cpu is the second of the busy CPUs * which will kick idle load balancer as soon as it has any load. */ static inline int nohz_kick_needed(struct rq *rq, int cpu) { unsigned long now = jiffies; int ret; int first_pick_cpu, second_pick_cpu; if (time_before(now, nohz.next_balance)) return 0; |
f6c3f1686 sched: Fix nohz b... |
3848 |
if (rq->idle_at_tick) |
83cd4fe27 sched: Change noh... |
3849 3850 3851 3852 3853 3854 3855 3856 3857 3858 3859 3860 3861 3862 3863 3864 3865 3866 3867 3868 3869 3870 3871 3872 3873 3874 3875 3876 3877 3878 3879 |
return 0; first_pick_cpu = atomic_read(&nohz.first_pick_cpu); second_pick_cpu = atomic_read(&nohz.second_pick_cpu); if (first_pick_cpu < nr_cpu_ids && first_pick_cpu != cpu && second_pick_cpu < nr_cpu_ids && second_pick_cpu != cpu) return 0; ret = atomic_cmpxchg(&nohz.first_pick_cpu, nr_cpu_ids, cpu); if (ret == nr_cpu_ids || ret == cpu) { atomic_cmpxchg(&nohz.second_pick_cpu, cpu, nr_cpu_ids); if (rq->nr_running > 1) return 1; } else { ret = atomic_cmpxchg(&nohz.second_pick_cpu, nr_cpu_ids, cpu); if (ret == nr_cpu_ids || ret == cpu) { if (rq->nr_running) return 1; } } return 0; } #else static void nohz_idle_balance(int this_cpu, enum cpu_idle_type idle) { } #endif /* * run_rebalance_domains is triggered when needed from the scheduler tick. * Also triggered for nohz idle balancing (with nohz_balancing_kick set). */ |
1e3c88bde sched: Move load ... |
3880 3881 3882 3883 3884 3885 3886 3887 |
static void run_rebalance_domains(struct softirq_action *h) { int this_cpu = smp_processor_id(); struct rq *this_rq = cpu_rq(this_cpu); enum cpu_idle_type idle = this_rq->idle_at_tick ? CPU_IDLE : CPU_NOT_IDLE; rebalance_domains(this_cpu, idle); |
1e3c88bde sched: Move load ... |
3888 |
/* |
83cd4fe27 sched: Change noh... |
3889 |
* If this cpu has a pending nohz_balance_kick, then do the |
1e3c88bde sched: Move load ... |
3890 3891 3892 |
* balancing on behalf of the other idle cpus whose ticks are * stopped. */ |
83cd4fe27 sched: Change noh... |
3893 |
nohz_idle_balance(this_cpu, idle); |
1e3c88bde sched: Move load ... |
3894 3895 3896 3897 |
} static inline int on_null_domain(int cpu) { |
90a6501f9 sched, rcu: Fix r... |
3898 |
return !rcu_dereference_sched(cpu_rq(cpu)->sd); |
1e3c88bde sched: Move load ... |
3899 3900 3901 3902 |
} /* * Trigger the SCHED_SOFTIRQ if it is time to do periodic load balancing. |
1e3c88bde sched: Move load ... |
3903 3904 3905 |
*/ static inline void trigger_load_balance(struct rq *rq, int cpu) { |
1e3c88bde sched: Move load ... |
3906 3907 3908 3909 |
/* Don't need to rebalance while attached to NULL domain */ if (time_after_eq(jiffies, rq->next_balance) && likely(!on_null_domain(cpu))) raise_softirq(SCHED_SOFTIRQ); |
83cd4fe27 sched: Change noh... |
3910 3911 3912 3913 |
#ifdef CONFIG_NO_HZ else if (nohz_kick_needed(rq, cpu) && likely(!on_null_domain(cpu))) nohz_balancer_kick(cpu); #endif |
1e3c88bde sched: Move load ... |
3914 |
} |
0bcdcf28c sched: Fix missin... |
3915 3916 3917 3918 3919 3920 3921 3922 3923 |
static void rq_online_fair(struct rq *rq) { update_sysctl(); } static void rq_offline_fair(struct rq *rq) { update_sysctl(); } |
1e3c88bde sched: Move load ... |
3924 3925 3926 3927 3928 3929 3930 3931 |
#else /* CONFIG_SMP */ /* * on UP we do not need to balance between CPUs: */ static inline void idle_balance(int cpu, struct rq *rq) { } |
55e12e5e7 sched: make sched... |
3932 |
#endif /* CONFIG_SMP */ |
e1d1484f7 sched: reduce bal... |
3933 |
|
bf0f6f24a sched: cfs core, ... |
3934 3935 3936 |
/* * scheduler tick hitting a task of our scheduling class: */ |
8f4d37ec0 sched: high-res p... |
3937 |
static void task_tick_fair(struct rq *rq, struct task_struct *curr, int queued) |
bf0f6f24a sched: cfs core, ... |
3938 3939 3940 3941 3942 3943 |
{ struct cfs_rq *cfs_rq; struct sched_entity *se = &curr->se; for_each_sched_entity(se) { cfs_rq = cfs_rq_of(se); |
8f4d37ec0 sched: high-res p... |
3944 |
entity_tick(cfs_rq, se, queued); |
bf0f6f24a sched: cfs core, ... |
3945 3946 3947 3948 |
} } /* |
cd29fe6f2 sched: Sanitize f... |
3949 3950 3951 |
* called on fork with the child task as argument from the parent's context * - child not yet on the tasklist * - preemption disabled |
bf0f6f24a sched: cfs core, ... |
3952 |
*/ |
cd29fe6f2 sched: Sanitize f... |
3953 |
static void task_fork_fair(struct task_struct *p) |
bf0f6f24a sched: cfs core, ... |
3954 |
{ |
cd29fe6f2 sched: Sanitize f... |
3955 |
struct cfs_rq *cfs_rq = task_cfs_rq(current); |
429d43bcc sched: cleanup: s... |
3956 |
struct sched_entity *se = &p->se, *curr = cfs_rq->curr; |
00bf7bfc2 sched: fix: move ... |
3957 |
int this_cpu = smp_processor_id(); |
cd29fe6f2 sched: Sanitize f... |
3958 3959 |
struct rq *rq = this_rq(); unsigned long flags; |
05fa785cf sched: Convert rq... |
3960 |
raw_spin_lock_irqsave(&rq->lock, flags); |
bf0f6f24a sched: cfs core, ... |
3961 |
|
861d034ee sched: Fix rq->cl... |
3962 |
update_rq_clock(rq); |
b0a0f667a sched: suppress R... |
3963 3964 |
if (unlikely(task_cpu(p) != this_cpu)) { rcu_read_lock(); |
cd29fe6f2 sched: Sanitize f... |
3965 |
__set_task_cpu(p, this_cpu); |
b0a0f667a sched: suppress R... |
3966 3967 |
rcu_read_unlock(); } |
bf0f6f24a sched: cfs core, ... |
3968 |
|
7109c4429 sched: call updat... |
3969 |
update_curr(cfs_rq); |
cd29fe6f2 sched: Sanitize f... |
3970 |
|
b5d9d734a sched: Ensure tha... |
3971 3972 |
if (curr) se->vruntime = curr->vruntime; |
aeb73b040 sched: clean up n... |
3973 |
place_entity(cfs_rq, se, 1); |
4d78e7b65 sched: new task p... |
3974 |
|
cd29fe6f2 sched: Sanitize f... |
3975 |
if (sysctl_sched_child_runs_first && curr && entity_before(curr, se)) { |
87fefa381 sched: optimize t... |
3976 |
/* |
edcb60a30 sched: kernel/sch... |
3977 3978 3979 |
* Upon rescheduling, sched_class::put_prev_task() will place * 'current' within the tree based on its new key value. */ |
4d78e7b65 sched: new task p... |
3980 |
swap(curr->vruntime, se->vruntime); |
aec0a5142 sched: call resch... |
3981 |
resched_task(rq->curr); |
4d78e7b65 sched: new task p... |
3982 |
} |
bf0f6f24a sched: cfs core, ... |
3983 |
|
88ec22d3e sched: Remove the... |
3984 |
se->vruntime -= cfs_rq->min_vruntime; |
05fa785cf sched: Convert rq... |
3985 |
raw_spin_unlock_irqrestore(&rq->lock, flags); |
bf0f6f24a sched: cfs core, ... |
3986 |
} |
cb4698450 sched: RT-balance... |
3987 3988 3989 3990 |
/* * Priority of the task has changed. Check to see if we preempt * the current task. */ |
da7a735e5 sched: Fix switch... |
3991 3992 |
static void prio_changed_fair(struct rq *rq, struct task_struct *p, int oldprio) |
cb4698450 sched: RT-balance... |
3993 |
{ |
da7a735e5 sched: Fix switch... |
3994 3995 |
if (!p->se.on_rq) return; |
cb4698450 sched: RT-balance... |
3996 3997 3998 3999 4000 |
/* * Reschedule if we are currently running on this runqueue and * our priority decreased, or if we are not currently running on * this runqueue and our priority is higher than the current's */ |
da7a735e5 sched: Fix switch... |
4001 |
if (rq->curr == p) { |
cb4698450 sched: RT-balance... |
4002 4003 4004 |
if (p->prio > oldprio) resched_task(rq->curr); } else |
15afe09bf sched: wakeup pre... |
4005 |
check_preempt_curr(rq, p, 0); |
cb4698450 sched: RT-balance... |
4006 |
} |
da7a735e5 sched: Fix switch... |
4007 4008 4009 4010 4011 4012 4013 4014 4015 4016 4017 4018 4019 4020 4021 4022 4023 4024 4025 4026 4027 4028 4029 |
static void switched_from_fair(struct rq *rq, struct task_struct *p) { struct sched_entity *se = &p->se; struct cfs_rq *cfs_rq = cfs_rq_of(se); /* * Ensure the task's vruntime is normalized, so that when its * switched back to the fair class the enqueue_entity(.flags=0) will * do the right thing. * * If it was on_rq, then the dequeue_entity(.flags=0) will already * have normalized the vruntime, if it was !on_rq, then only when * the task is sleeping will it still have non-normalized vruntime. */ if (!se->on_rq && p->state != TASK_RUNNING) { /* * Fix up our vruntime so that the current sleep doesn't * cause 'unlimited' sleep bonus. */ place_entity(cfs_rq, se, 0); se->vruntime -= cfs_rq->min_vruntime; } } |
cb4698450 sched: RT-balance... |
4030 4031 4032 |
/* * We switched to the sched_fair class. */ |
da7a735e5 sched: Fix switch... |
4033 |
static void switched_to_fair(struct rq *rq, struct task_struct *p) |
cb4698450 sched: RT-balance... |
4034 |
{ |
da7a735e5 sched: Fix switch... |
4035 4036 |
if (!p->se.on_rq) return; |
cb4698450 sched: RT-balance... |
4037 4038 4039 4040 4041 |
/* * We were most likely switched from sched_rt, so * kick off the schedule if running, otherwise just see * if we can still preempt the current task. */ |
da7a735e5 sched: Fix switch... |
4042 |
if (rq->curr == p) |
cb4698450 sched: RT-balance... |
4043 4044 |
resched_task(rq->curr); else |
15afe09bf sched: wakeup pre... |
4045 |
check_preempt_curr(rq, p, 0); |
cb4698450 sched: RT-balance... |
4046 |
} |
83b699ed2 sched: revert rec... |
4047 4048 4049 4050 4051 4052 4053 4054 4055 4056 4057 4058 |
/* Account for a task changing its policy or group. * * This routine is mostly called to set cfs_rq->curr field when a task * migrates between groups/classes. */ static void set_curr_task_fair(struct rq *rq) { struct sched_entity *se = &rq->curr->se; for_each_sched_entity(se) set_next_entity(cfs_rq_of(se), se); } |
810b38179 sched: retain vru... |
4059 |
#ifdef CONFIG_FAIR_GROUP_SCHED |
b2b5ce022 sched, cgroup: Fi... |
4060 |
static void task_move_group_fair(struct task_struct *p, int on_rq) |
810b38179 sched: retain vru... |
4061 |
{ |
b2b5ce022 sched, cgroup: Fi... |
4062 4063 4064 4065 4066 4067 4068 4069 4070 4071 4072 4073 4074 4075 4076 4077 |
/* * If the task was not on the rq at the time of this cgroup movement * it must have been asleep, sleeping tasks keep their ->vruntime * absolute on their old rq until wakeup (needed for the fair sleeper * bonus in place_entity()). * * If it was on the rq, we've just 'preempted' it, which does convert * ->vruntime to a relative base. * * Make sure both cases convert their relative position when migrating * to another cgroup's rq. This does somewhat interfere with the * fair sleeper stuff for the first placement, but who cares. */ if (!on_rq) p->se.vruntime -= cfs_rq_of(&p->se)->min_vruntime; set_task_rq(p, task_cpu(p)); |
88ec22d3e sched: Remove the... |
4078 |
if (!on_rq) |
b2b5ce022 sched, cgroup: Fi... |
4079 |
p->se.vruntime += cfs_rq_of(&p->se)->min_vruntime; |
810b38179 sched: retain vru... |
4080 4081 |
} #endif |
6d686f456 sched: Don't expo... |
4082 |
static unsigned int get_rr_interval_fair(struct rq *rq, struct task_struct *task) |
0d721cead sched: Simplify s... |
4083 4084 |
{ struct sched_entity *se = &task->se; |
0d721cead sched: Simplify s... |
4085 4086 4087 4088 4089 4090 |
unsigned int rr_interval = 0; /* * Time slice is 0 for SCHED_OTHER tasks that are on an otherwise * idle runqueue: */ |
0d721cead sched: Simplify s... |
4091 4092 |
if (rq->cfs.load.weight) rr_interval = NS_TO_JIFFIES(sched_slice(&rq->cfs, se)); |
0d721cead sched: Simplify s... |
4093 4094 4095 |
return rr_interval; } |
bf0f6f24a sched: cfs core, ... |
4096 4097 4098 |
/* * All the scheduling class methods: */ |
5522d5d5f sched: mark sched... |
4099 4100 |
static const struct sched_class fair_sched_class = { .next = &idle_sched_class, |
bf0f6f24a sched: cfs core, ... |
4101 4102 4103 |
.enqueue_task = enqueue_task_fair, .dequeue_task = dequeue_task_fair, .yield_task = yield_task_fair, |
d95f41220 sched: Add yield_... |
4104 |
.yield_to_task = yield_to_task_fair, |
bf0f6f24a sched: cfs core, ... |
4105 |
|
2e09bf556 sched: wakeup gra... |
4106 |
.check_preempt_curr = check_preempt_wakeup, |
bf0f6f24a sched: cfs core, ... |
4107 4108 4109 |
.pick_next_task = pick_next_task_fair, .put_prev_task = put_prev_task_fair, |
681f3e685 sched: isolate SM... |
4110 |
#ifdef CONFIG_SMP |
4ce72a2c0 sched: add CONFIG... |
4111 |
.select_task_rq = select_task_rq_fair, |
0bcdcf28c sched: Fix missin... |
4112 4113 |
.rq_online = rq_online_fair, .rq_offline = rq_offline_fair, |
88ec22d3e sched: Remove the... |
4114 4115 |
.task_waking = task_waking_fair, |
681f3e685 sched: isolate SM... |
4116 |
#endif |
bf0f6f24a sched: cfs core, ... |
4117 |
|
83b699ed2 sched: revert rec... |
4118 |
.set_curr_task = set_curr_task_fair, |
bf0f6f24a sched: cfs core, ... |
4119 |
.task_tick = task_tick_fair, |
cd29fe6f2 sched: Sanitize f... |
4120 |
.task_fork = task_fork_fair, |
cb4698450 sched: RT-balance... |
4121 4122 |
.prio_changed = prio_changed_fair, |
da7a735e5 sched: Fix switch... |
4123 |
.switched_from = switched_from_fair, |
cb4698450 sched: RT-balance... |
4124 |
.switched_to = switched_to_fair, |
810b38179 sched: retain vru... |
4125 |
|
0d721cead sched: Simplify s... |
4126 |
.get_rr_interval = get_rr_interval_fair, |
810b38179 sched: retain vru... |
4127 |
#ifdef CONFIG_FAIR_GROUP_SCHED |
b2b5ce022 sched, cgroup: Fi... |
4128 |
.task_move_group = task_move_group_fair, |
810b38179 sched: retain vru... |
4129 |
#endif |
bf0f6f24a sched: cfs core, ... |
4130 4131 4132 |
}; #ifdef CONFIG_SCHED_DEBUG |
5cef9eca3 sched: remove the... |
4133 |
static void print_cfs_stats(struct seq_file *m, int cpu) |
bf0f6f24a sched: cfs core, ... |
4134 |
{ |
bf0f6f24a sched: cfs core, ... |
4135 |
struct cfs_rq *cfs_rq; |
5973e5b95 sched: fix: don't... |
4136 |
rcu_read_lock(); |
c3b64f1e4 sched: clean up s... |
4137 |
for_each_leaf_cfs_rq(cpu_rq(cpu), cfs_rq) |
5cef9eca3 sched: remove the... |
4138 |
print_cfs_rq(m, cpu, cfs_rq); |
5973e5b95 sched: fix: don't... |
4139 |
rcu_read_unlock(); |
bf0f6f24a sched: cfs core, ... |
4140 4141 |
} #endif |