Blame view
block/blk-throttle.c
68.5 KB
b24413180
|
1 |
// SPDX-License-Identifier: GPL-2.0 |
e43473b7f
|
2 3 4 5 6 7 8 9 10 11 12 |
/* * Interface for controlling IO bandwidth on a request queue * * Copyright (C) 2010 Vivek Goyal <vgoyal@redhat.com> */ #include <linux/module.h> #include <linux/slab.h> #include <linux/blkdev.h> #include <linux/bio.h> #include <linux/blktrace_api.h> |
eea8f41cc
|
13 |
#include <linux/blk-cgroup.h> |
bc9fcbf9c
|
14 |
#include "blk.h" |
1d156646e
|
15 |
#include "blk-cgroup-rwstat.h" |
e43473b7f
|
16 17 |
/* Max dispatch from a group in 1 round */ |
e675df2ad
|
18 |
#define THROTL_GRP_QUANTUM 8 |
e43473b7f
|
19 20 |
/* Total max dispatch from all groups in one round */ |
e675df2ad
|
21 |
#define THROTL_QUANTUM 32 |
e43473b7f
|
22 |
|
d61fcfa4b
|
23 24 25 |
/* Throttling is performed over a slice and after that slice is renewed */ #define DFL_THROTL_SLICE_HD (HZ / 10) #define DFL_THROTL_SLICE_SSD (HZ / 50) |
297e3d854
|
26 |
#define MAX_THROTL_SLICE (HZ) |
9e234eeaf
|
27 |
#define MAX_IDLE_TIME (5L * 1000 * 1000) /* 5 s */ |
9bb67aeb9
|
28 29 |
#define MIN_THROTL_BPS (320 * 1024) #define MIN_THROTL_IOPS (10) |
b4f428ef2
|
30 31 |
#define DFL_LATENCY_TARGET (-1L) #define DFL_IDLE_THRESHOLD (0) |
6679a90c4
|
32 33 34 35 36 37 38 |
#define DFL_HD_BASELINE_LATENCY (4000L) /* 4ms */ #define LATENCY_FILTERED_SSD (0) /* * For HD, very small latency comes from sequential IO. Such IO is helpless to * help determine if its IO is impacted by others, hence we ignore the IO */ #define LATENCY_FILTERED_HD (1000L) /* 1ms */ |
e43473b7f
|
39 |
|
3c798398e
|
40 |
static struct blkcg_policy blkcg_policy_throtl; |
0381411e4
|
41 |
|
450adcbe5
|
42 43 |
/* A workqueue to queue throttle related work */ static struct workqueue_struct *kthrotld_workqueue; |
450adcbe5
|
44 |
|
c5cc2070b
|
45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 |
/* * To implement hierarchical throttling, throtl_grps form a tree and bios * are dispatched upwards level by level until they reach the top and get * issued. When dispatching bios from the children and local group at each * level, if the bios are dispatched into a single bio_list, there's a risk * of a local or child group which can queue many bios at once filling up * the list starving others. * * To avoid such starvation, dispatched bios are queued separately * according to where they came from. When they are again dispatched to * the parent, they're popped in round-robin order so that no single source * hogs the dispatch window. * * throtl_qnode is used to keep the queued bios separated by their sources. * Bios are queued to throtl_qnode which in turn is queued to * throtl_service_queue and then dispatched in round-robin order. * * It's also used to track the reference counts on blkg's. A qnode always * belongs to a throtl_grp and gets queued on itself or the parent, so * incrementing the reference of the associated throtl_grp when a qnode is * queued and decrementing when dequeued is enough to keep the whole blkg * tree pinned while bios are in flight. */ struct throtl_qnode { struct list_head node; /* service_queue->queued[] */ struct bio_list bios; /* queued bios */ struct throtl_grp *tg; /* tg this qnode belongs to */ }; |
c9e0332e8
|
73 |
struct throtl_service_queue { |
77216b048
|
74 |
struct throtl_service_queue *parent_sq; /* the parent service_queue */ |
73f0d49a9
|
75 76 77 78 |
/* * Bios queued directly to this service_queue or dispatched from * children throtl_grp's. */ |
c5cc2070b
|
79 |
struct list_head queued[2]; /* throtl_qnode [READ/WRITE] */ |
73f0d49a9
|
80 81 82 83 84 85 |
unsigned int nr_queued[2]; /* number of queued bios */ /* * RB tree of active children throtl_grp's, which are sorted by * their ->disptime. */ |
9ff01255a
|
86 |
struct rb_root_cached pending_tree; /* RB tree of active tgs */ |
c9e0332e8
|
87 88 |
unsigned int nr_pending; /* # queued in the tree */ unsigned long first_pending_disptime; /* disptime of the first tg */ |
69df0ab03
|
89 |
struct timer_list pending_timer; /* fires on first_pending_disptime */ |
e43473b7f
|
90 |
}; |
5b2c16aae
|
91 92 |
enum tg_state_flags { THROTL_TG_PENDING = 1 << 0, /* on parent's pending tree */ |
0e9f4164b
|
93 |
THROTL_TG_WAS_EMPTY = 1 << 1, /* bio_lists[] became non-empty */ |
5b2c16aae
|
94 |
}; |
e43473b7f
|
95 |
#define rb_entry_tg(node) rb_entry((node), struct throtl_grp, rb_node) |
9f626e372
|
96 |
enum { |
cd5ab1b0f
|
97 |
LIMIT_LOW, |
9f626e372
|
98 99 100 |
LIMIT_MAX, LIMIT_CNT, }; |
e43473b7f
|
101 |
struct throtl_grp { |
f95a04afa
|
102 103 |
/* must be the first member */ struct blkg_policy_data pd; |
c9e0332e8
|
104 |
/* active throtl group service_queue member */ |
e43473b7f
|
105 |
struct rb_node rb_node; |
0f3457f60
|
106 107 |
/* throtl_data this group belongs to */ struct throtl_data *td; |
49a2f1e3f
|
108 109 |
/* this group's service queue */ struct throtl_service_queue service_queue; |
e43473b7f
|
110 |
/* |
c5cc2070b
|
111 112 113 114 115 116 117 118 119 120 121 |
* qnode_on_self is used when bios are directly queued to this * throtl_grp so that local bios compete fairly with bios * dispatched from children. qnode_on_parent is used when bios are * dispatched from this throtl_grp into its parent and will compete * with the sibling qnode_on_parents and the parent's * qnode_on_self. */ struct throtl_qnode qnode_on_self[2]; struct throtl_qnode qnode_on_parent[2]; /* |
e43473b7f
|
122 123 124 125 126 |
* Dispatch time in jiffies. This is the estimated time when group * will unthrottle and is ready to dispatch more bio. It is used as * key to sort active groups in service tree. */ unsigned long disptime; |
e43473b7f
|
127 |
unsigned int flags; |
693e751e7
|
128 129 |
/* are there any throtl rules between this group and td? */ bool has_rules[2]; |
cd5ab1b0f
|
130 |
/* internally used bytes per second rate limits */ |
9f626e372
|
131 |
uint64_t bps[2][LIMIT_CNT]; |
cd5ab1b0f
|
132 133 |
/* user configured bps limits */ uint64_t bps_conf[2][LIMIT_CNT]; |
e43473b7f
|
134 |
|
cd5ab1b0f
|
135 |
/* internally used IOPS limits */ |
9f626e372
|
136 |
unsigned int iops[2][LIMIT_CNT]; |
cd5ab1b0f
|
137 138 |
/* user configured IOPS limits */ unsigned int iops_conf[2][LIMIT_CNT]; |
8e89d13f4
|
139 |
|
b53b072c4
|
140 |
/* Number of bytes dispatched in current slice */ |
e43473b7f
|
141 |
uint64_t bytes_disp[2]; |
8e89d13f4
|
142 143 |
/* Number of bio's dispatched in current slice */ unsigned int io_disp[2]; |
e43473b7f
|
144 |
|
3f0abd806
|
145 146 147 148 149 150 |
unsigned long last_low_overflow_time[2]; uint64_t last_bytes_disp[2]; unsigned int last_io_disp[2]; unsigned long last_check_time; |
ec80991d6
|
151 |
unsigned long latency_target; /* us */ |
5b81fc3cc
|
152 |
unsigned long latency_target_conf; /* us */ |
e43473b7f
|
153 154 155 |
/* When did we start a new slice */ unsigned long slice_start[2]; unsigned long slice_end[2]; |
9e234eeaf
|
156 157 158 159 160 |
unsigned long last_finish_time; /* ns / 1024 */ unsigned long checked_last_finish_time; /* ns / 1024 */ unsigned long avg_idletime; /* ns / 1024 */ unsigned long idletime_threshold; /* us */ |
5b81fc3cc
|
161 |
unsigned long idletime_threshold_conf; /* us */ |
53696b8d2
|
162 163 164 165 |
unsigned int bio_cnt; /* total bios */ unsigned int bad_bio_cnt; /* bios exceeding latency threshold */ unsigned long bio_cnt_reset_time; |
7ca464383
|
166 |
|
4f1e9630a
|
167 168 |
atomic_t io_split_cnt[2]; atomic_t last_io_split_cnt[2]; |
7ca464383
|
169 170 |
struct blkg_rwstat stat_bytes; struct blkg_rwstat stat_ios; |
e43473b7f
|
171 |
}; |
b9147dd1b
|
172 173 174 175 176 177 178 179 180 181 182 183 |
/* We measure latency for request size from <= 4k to >= 1M */ #define LATENCY_BUCKET_SIZE 9 struct latency_bucket { unsigned long total_latency; /* ns / 1024 */ int samples; }; struct avg_latency_bucket { unsigned long latency; /* ns / 1024 */ bool valid; }; |
e43473b7f
|
184 185 |
struct throtl_data { |
e43473b7f
|
186 |
/* service tree for active throtl groups */ |
c9e0332e8
|
187 |
struct throtl_service_queue service_queue; |
e43473b7f
|
188 |
|
e43473b7f
|
189 190 191 192 |
struct request_queue *queue; /* Total Number of queued bios on READ and WRITE lists */ unsigned int nr_queued[2]; |
297e3d854
|
193 |
unsigned int throtl_slice; |
e43473b7f
|
194 |
/* Work for dispatching throttled bios */ |
69df0ab03
|
195 |
struct work_struct dispatch_work; |
9f626e372
|
196 197 |
unsigned int limit_index; bool limit_valid[LIMIT_CNT]; |
3f0abd806
|
198 199 200 |
unsigned long low_upgrade_time; unsigned long low_downgrade_time; |
7394e31fa
|
201 202 |
unsigned int scale; |
b9147dd1b
|
203 |
|
b889bf66d
|
204 205 206 |
struct latency_bucket tmp_buckets[2][LATENCY_BUCKET_SIZE]; struct avg_latency_bucket avg_buckets[2][LATENCY_BUCKET_SIZE]; struct latency_bucket __percpu *latency_buckets[2]; |
b9147dd1b
|
207 |
unsigned long last_calculate_time; |
6679a90c4
|
208 |
unsigned long filtered_latency; |
b9147dd1b
|
209 210 |
bool track_bio_latency; |
e43473b7f
|
211 |
}; |
e99e88a9d
|
212 |
static void throtl_pending_timer_fn(struct timer_list *t); |
69df0ab03
|
213 |
|
f95a04afa
|
214 215 216 217 |
static inline struct throtl_grp *pd_to_tg(struct blkg_policy_data *pd) { return pd ? container_of(pd, struct throtl_grp, pd) : NULL; } |
3c798398e
|
218 |
static inline struct throtl_grp *blkg_to_tg(struct blkcg_gq *blkg) |
0381411e4
|
219 |
{ |
f95a04afa
|
220 |
return pd_to_tg(blkg_to_pd(blkg, &blkcg_policy_throtl)); |
0381411e4
|
221 |
} |
3c798398e
|
222 |
static inline struct blkcg_gq *tg_to_blkg(struct throtl_grp *tg) |
0381411e4
|
223 |
{ |
f95a04afa
|
224 |
return pd_to_blkg(&tg->pd); |
0381411e4
|
225 |
} |
fda6f272c
|
226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 |
/** * sq_to_tg - return the throl_grp the specified service queue belongs to * @sq: the throtl_service_queue of interest * * Return the throtl_grp @sq belongs to. If @sq is the top-level one * embedded in throtl_data, %NULL is returned. */ static struct throtl_grp *sq_to_tg(struct throtl_service_queue *sq) { if (sq && sq->parent_sq) return container_of(sq, struct throtl_grp, service_queue); else return NULL; } /** * sq_to_td - return throtl_data the specified service queue belongs to * @sq: the throtl_service_queue of interest * |
b43daedc0
|
245 |
* A service_queue can be embedded in either a throtl_grp or throtl_data. |
fda6f272c
|
246 247 248 249 250 251 252 253 254 255 256 |
* Determine the associated throtl_data accordingly and return it. */ static struct throtl_data *sq_to_td(struct throtl_service_queue *sq) { struct throtl_grp *tg = sq_to_tg(sq); if (tg) return tg->td; else return container_of(sq, struct throtl_data, service_queue); } |
7394e31fa
|
257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 |
/* * cgroup's limit in LIMIT_MAX is scaled if low limit is set. This scale is to * make the IO dispatch more smooth. * Scale up: linearly scale up according to lapsed time since upgrade. For * every throtl_slice, the limit scales up 1/2 .low limit till the * limit hits .max limit * Scale down: exponentially scale down if a cgroup doesn't hit its .low limit */ static uint64_t throtl_adjusted_limit(uint64_t low, struct throtl_data *td) { /* arbitrary value to avoid too big scale */ if (td->scale < 4096 && time_after_eq(jiffies, td->low_upgrade_time + td->scale * td->throtl_slice)) td->scale = (jiffies - td->low_upgrade_time) / td->throtl_slice; return low + (low >> 1) * td->scale; } |
9f626e372
|
274 275 |
static uint64_t tg_bps_limit(struct throtl_grp *tg, int rw) { |
b22c417c8
|
276 |
struct blkcg_gq *blkg = tg_to_blkg(tg); |
7394e31fa
|
277 |
struct throtl_data *td; |
b22c417c8
|
278 279 280 281 |
uint64_t ret; if (cgroup_subsys_on_dfl(io_cgrp_subsys) && !blkg->parent) return U64_MAX; |
7394e31fa
|
282 283 284 |
td = tg->td; ret = tg->bps[rw][td->limit_index]; |
9bb67aeb9
|
285 286 287 288 289 290 291 292 |
if (ret == 0 && td->limit_index == LIMIT_LOW) { /* intermediate node or iops isn't 0 */ if (!list_empty(&blkg->blkcg->css.children) || tg->iops[rw][td->limit_index]) return U64_MAX; else return MIN_THROTL_BPS; } |
7394e31fa
|
293 294 295 296 297 298 299 300 |
if (td->limit_index == LIMIT_MAX && tg->bps[rw][LIMIT_LOW] && tg->bps[rw][LIMIT_LOW] != tg->bps[rw][LIMIT_MAX]) { uint64_t adjusted; adjusted = throtl_adjusted_limit(tg->bps[rw][LIMIT_LOW], td); ret = min(tg->bps[rw][LIMIT_MAX], adjusted); } |
b22c417c8
|
301 |
return ret; |
9f626e372
|
302 303 304 305 |
} static unsigned int tg_iops_limit(struct throtl_grp *tg, int rw) { |
b22c417c8
|
306 |
struct blkcg_gq *blkg = tg_to_blkg(tg); |
7394e31fa
|
307 |
struct throtl_data *td; |
b22c417c8
|
308 309 310 311 |
unsigned int ret; if (cgroup_subsys_on_dfl(io_cgrp_subsys) && !blkg->parent) return UINT_MAX; |
9bb67aeb9
|
312 |
|
7394e31fa
|
313 314 |
td = tg->td; ret = tg->iops[rw][td->limit_index]; |
9bb67aeb9
|
315 316 317 318 319 320 321 322 |
if (ret == 0 && tg->td->limit_index == LIMIT_LOW) { /* intermediate node or bps isn't 0 */ if (!list_empty(&blkg->blkcg->css.children) || tg->bps[rw][td->limit_index]) return UINT_MAX; else return MIN_THROTL_IOPS; } |
7394e31fa
|
323 324 325 326 327 328 329 330 331 332 |
if (td->limit_index == LIMIT_MAX && tg->iops[rw][LIMIT_LOW] && tg->iops[rw][LIMIT_LOW] != tg->iops[rw][LIMIT_MAX]) { uint64_t adjusted; adjusted = throtl_adjusted_limit(tg->iops[rw][LIMIT_LOW], td); if (adjusted > UINT_MAX) adjusted = UINT_MAX; ret = min_t(unsigned int, tg->iops[rw][LIMIT_MAX], adjusted); } |
b22c417c8
|
333 |
return ret; |
9f626e372
|
334 |
} |
b9147dd1b
|
335 336 |
#define request_bucket_index(sectors) \ clamp_t(int, order_base_2(sectors) - 3, 0, LATENCY_BUCKET_SIZE - 1) |
fda6f272c
|
337 338 339 340 341 342 343 344 |
/** * throtl_log - log debug message via blktrace * @sq: the service_queue being reported * @fmt: printf format string * @args: printf args * * The messages are prefixed with "throtl BLKG_NAME" if @sq belongs to a * throtl_grp; otherwise, just "throtl". |
fda6f272c
|
345 346 347 348 349 350 |
*/ #define throtl_log(sq, fmt, args...) do { \ struct throtl_grp *__tg = sq_to_tg((sq)); \ struct throtl_data *__td = sq_to_td((sq)); \ \ (void)__td; \ |
59fa0224c
|
351 352 |
if (likely(!blk_trace_note_message_enabled(__td->queue))) \ break; \ |
fda6f272c
|
353 |
if ((__tg)) { \ |
35fe6d763
|
354 355 |
blk_add_cgroup_trace_msg(__td->queue, \ tg_to_blkg(__tg)->blkcg, "throtl " fmt, ##args);\ |
fda6f272c
|
356 357 358 |
} else { \ blk_add_trace_msg(__td->queue, "throtl " fmt, ##args); \ } \ |
54e7ed12b
|
359 |
} while (0) |
e43473b7f
|
360 |
|
ea0ea2bc6
|
361 362 363 364 365 366 367 |
static inline unsigned int throtl_bio_data_size(struct bio *bio) { /* assume it's one sector */ if (unlikely(bio_op(bio) == REQ_OP_DISCARD)) return 512; return bio->bi_iter.bi_size; } |
c5cc2070b
|
368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 |
static void throtl_qnode_init(struct throtl_qnode *qn, struct throtl_grp *tg) { INIT_LIST_HEAD(&qn->node); bio_list_init(&qn->bios); qn->tg = tg; } /** * throtl_qnode_add_bio - add a bio to a throtl_qnode and activate it * @bio: bio being added * @qn: qnode to add bio to * @queued: the service_queue->queued[] list @qn belongs to * * Add @bio to @qn and put @qn on @queued if it's not already on. * @qn->tg's reference count is bumped when @qn is activated. See the * comment on top of throtl_qnode definition for details. */ static void throtl_qnode_add_bio(struct bio *bio, struct throtl_qnode *qn, struct list_head *queued) { bio_list_add(&qn->bios, bio); if (list_empty(&qn->node)) { list_add_tail(&qn->node, queued); blkg_get(tg_to_blkg(qn->tg)); } } /** * throtl_peek_queued - peek the first bio on a qnode list * @queued: the qnode list to peek */ static struct bio *throtl_peek_queued(struct list_head *queued) { |
b7b609de5
|
401 |
struct throtl_qnode *qn; |
c5cc2070b
|
402 403 404 405 |
struct bio *bio; if (list_empty(queued)) return NULL; |
b7b609de5
|
406 |
qn = list_first_entry(queued, struct throtl_qnode, node); |
c5cc2070b
|
407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 |
bio = bio_list_peek(&qn->bios); WARN_ON_ONCE(!bio); return bio; } /** * throtl_pop_queued - pop the first bio form a qnode list * @queued: the qnode list to pop a bio from * @tg_to_put: optional out argument for throtl_grp to put * * Pop the first bio from the qnode list @queued. After popping, the first * qnode is removed from @queued if empty or moved to the end of @queued so * that the popping order is round-robin. * * When the first qnode is removed, its associated throtl_grp should be put * too. If @tg_to_put is NULL, this function automatically puts it; * otherwise, *@tg_to_put is set to the throtl_grp to put and the caller is * responsible for putting it. */ static struct bio *throtl_pop_queued(struct list_head *queued, struct throtl_grp **tg_to_put) { |
b7b609de5
|
429 |
struct throtl_qnode *qn; |
c5cc2070b
|
430 431 432 433 |
struct bio *bio; if (list_empty(queued)) return NULL; |
b7b609de5
|
434 |
qn = list_first_entry(queued, struct throtl_qnode, node); |
c5cc2070b
|
435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 |
bio = bio_list_pop(&qn->bios); WARN_ON_ONCE(!bio); if (bio_list_empty(&qn->bios)) { list_del_init(&qn->node); if (tg_to_put) *tg_to_put = qn->tg; else blkg_put(tg_to_blkg(qn->tg)); } else { list_move_tail(&qn->node, queued); } return bio; } |
49a2f1e3f
|
450 |
/* init a service_queue, assumes the caller zeroed it */ |
b2ce2643c
|
451 |
static void throtl_service_queue_init(struct throtl_service_queue *sq) |
49a2f1e3f
|
452 |
{ |
c5cc2070b
|
453 454 |
INIT_LIST_HEAD(&sq->queued[0]); INIT_LIST_HEAD(&sq->queued[1]); |
9ff01255a
|
455 |
sq->pending_tree = RB_ROOT_CACHED; |
e99e88a9d
|
456 |
timer_setup(&sq->pending_timer, throtl_pending_timer_fn, 0); |
69df0ab03
|
457 |
} |
cf09a8ee1
|
458 459 460 |
static struct blkg_policy_data *throtl_pd_alloc(gfp_t gfp, struct request_queue *q, struct blkcg *blkcg) |
001bea73e
|
461 |
{ |
4fb72036f
|
462 |
struct throtl_grp *tg; |
24bdb8ef0
|
463 |
int rw; |
4fb72036f
|
464 |
|
cf09a8ee1
|
465 |
tg = kzalloc_node(sizeof(*tg), gfp, q->node); |
4fb72036f
|
466 |
if (!tg) |
77ea73388
|
467 |
return NULL; |
4fb72036f
|
468 |
|
7ca464383
|
469 470 471 472 473 |
if (blkg_rwstat_init(&tg->stat_bytes, gfp)) goto err_free_tg; if (blkg_rwstat_init(&tg->stat_ios, gfp)) goto err_exit_stat_bytes; |
b2ce2643c
|
474 475 476 477 478 479 480 481 |
throtl_service_queue_init(&tg->service_queue); for (rw = READ; rw <= WRITE; rw++) { throtl_qnode_init(&tg->qnode_on_self[rw], tg); throtl_qnode_init(&tg->qnode_on_parent[rw], tg); } RB_CLEAR_NODE(&tg->rb_node); |
9f626e372
|
482 483 484 485 |
tg->bps[READ][LIMIT_MAX] = U64_MAX; tg->bps[WRITE][LIMIT_MAX] = U64_MAX; tg->iops[READ][LIMIT_MAX] = UINT_MAX; tg->iops[WRITE][LIMIT_MAX] = UINT_MAX; |
cd5ab1b0f
|
486 487 488 489 490 |
tg->bps_conf[READ][LIMIT_MAX] = U64_MAX; tg->bps_conf[WRITE][LIMIT_MAX] = U64_MAX; tg->iops_conf[READ][LIMIT_MAX] = UINT_MAX; tg->iops_conf[WRITE][LIMIT_MAX] = UINT_MAX; /* LIMIT_LOW will have default value 0 */ |
b2ce2643c
|
491 |
|
ec80991d6
|
492 |
tg->latency_target = DFL_LATENCY_TARGET; |
5b81fc3cc
|
493 |
tg->latency_target_conf = DFL_LATENCY_TARGET; |
b4f428ef2
|
494 495 |
tg->idletime_threshold = DFL_IDLE_THRESHOLD; tg->idletime_threshold_conf = DFL_IDLE_THRESHOLD; |
ec80991d6
|
496 |
|
4fb72036f
|
497 |
return &tg->pd; |
7ca464383
|
498 499 500 501 502 503 |
err_exit_stat_bytes: blkg_rwstat_exit(&tg->stat_bytes); err_free_tg: kfree(tg); return NULL; |
001bea73e
|
504 |
} |
a9520cd6f
|
505 |
static void throtl_pd_init(struct blkg_policy_data *pd) |
a29a171e7
|
506 |
{ |
a9520cd6f
|
507 508 |
struct throtl_grp *tg = pd_to_tg(pd); struct blkcg_gq *blkg = tg_to_blkg(tg); |
77216b048
|
509 |
struct throtl_data *td = blkg->q->td; |
b2ce2643c
|
510 |
struct throtl_service_queue *sq = &tg->service_queue; |
cd1604fab
|
511 |
|
9138125be
|
512 |
/* |
aa6ec29be
|
513 |
* If on the default hierarchy, we switch to properly hierarchical |
9138125be
|
514 515 516 517 518 |
* behavior where limits on a given throtl_grp are applied to the * whole subtree rather than just the group itself. e.g. If 16M * read_bps limit is set on the root group, the whole system can't * exceed 16M for the device. * |
aa6ec29be
|
519 |
* If not on the default hierarchy, the broken flat hierarchy |
9138125be
|
520 521 522 523 524 |
* behavior is retained where all throtl_grps are treated as if * they're all separate root groups right below throtl_data. * Limits of a group don't interact with limits of other groups * regardless of the position of the group in the hierarchy. */ |
b2ce2643c
|
525 |
sq->parent_sq = &td->service_queue; |
9e10a130d
|
526 |
if (cgroup_subsys_on_dfl(io_cgrp_subsys) && blkg->parent) |
b2ce2643c
|
527 |
sq->parent_sq = &blkg_to_tg(blkg->parent)->service_queue; |
77216b048
|
528 |
tg->td = td; |
8a3d26151
|
529 |
} |
693e751e7
|
530 531 532 533 534 535 536 537 |
/* * Set has_rules[] if @tg or any of its parents have limits configured. * This doesn't require walking up to the top of the hierarchy as the * parent's has_rules[] is guaranteed to be correct. */ static void tg_update_has_rules(struct throtl_grp *tg) { struct throtl_grp *parent_tg = sq_to_tg(tg->service_queue.parent_sq); |
9f626e372
|
538 |
struct throtl_data *td = tg->td; |
693e751e7
|
539 540 541 542 |
int rw; for (rw = READ; rw <= WRITE; rw++) tg->has_rules[rw] = (parent_tg && parent_tg->has_rules[rw]) || |
9f626e372
|
543 544 545 |
(td->limit_valid[td->limit_index] && (tg_bps_limit(tg, rw) != U64_MAX || tg_iops_limit(tg, rw) != UINT_MAX)); |
693e751e7
|
546 |
} |
a9520cd6f
|
547 |
static void throtl_pd_online(struct blkg_policy_data *pd) |
693e751e7
|
548 |
{ |
aec242468
|
549 |
struct throtl_grp *tg = pd_to_tg(pd); |
693e751e7
|
550 551 552 553 |
/* * We don't want new groups to escape the limits of its ancestors. * Update has_rules[] after a new group is brought online. */ |
aec242468
|
554 |
tg_update_has_rules(tg); |
693e751e7
|
555 |
} |
acaf523a7
|
556 |
#ifdef CONFIG_BLK_DEV_THROTTLING_LOW |
cd5ab1b0f
|
557 558 559 560 561 562 563 564 565 566 567 |
static void blk_throtl_update_limit_valid(struct throtl_data *td) { struct cgroup_subsys_state *pos_css; struct blkcg_gq *blkg; bool low_valid = false; rcu_read_lock(); blkg_for_each_descendant_post(blkg, pos_css, td->queue->root_blkg) { struct throtl_grp *tg = blkg_to_tg(blkg); if (tg->bps[READ][LIMIT_LOW] || tg->bps[WRITE][LIMIT_LOW] || |
43ada7878
|
568 |
tg->iops[READ][LIMIT_LOW] || tg->iops[WRITE][LIMIT_LOW]) { |
cd5ab1b0f
|
569 |
low_valid = true; |
43ada7878
|
570 571 |
break; } |
cd5ab1b0f
|
572 573 574 575 576 |
} rcu_read_unlock(); td->limit_valid[LIMIT_LOW] = low_valid; } |
acaf523a7
|
577 578 579 580 581 |
#else static inline void blk_throtl_update_limit_valid(struct throtl_data *td) { } #endif |
cd5ab1b0f
|
582 |
|
c79892c55
|
583 |
static void throtl_upgrade_state(struct throtl_data *td); |
cd5ab1b0f
|
584 585 586 587 588 589 590 591 592 593 |
static void throtl_pd_offline(struct blkg_policy_data *pd) { struct throtl_grp *tg = pd_to_tg(pd); tg->bps[READ][LIMIT_LOW] = 0; tg->bps[WRITE][LIMIT_LOW] = 0; tg->iops[READ][LIMIT_LOW] = 0; tg->iops[WRITE][LIMIT_LOW] = 0; blk_throtl_update_limit_valid(tg->td); |
c79892c55
|
594 595 |
if (!tg->td->limit_valid[tg->td->limit_index]) throtl_upgrade_state(tg->td); |
cd5ab1b0f
|
596 |
} |
001bea73e
|
597 598 |
static void throtl_pd_free(struct blkg_policy_data *pd) { |
4fb72036f
|
599 |
struct throtl_grp *tg = pd_to_tg(pd); |
b2ce2643c
|
600 |
del_timer_sync(&tg->service_queue.pending_timer); |
7ca464383
|
601 602 |
blkg_rwstat_exit(&tg->stat_bytes); blkg_rwstat_exit(&tg->stat_ios); |
4fb72036f
|
603 |
kfree(tg); |
001bea73e
|
604 |
} |
0049af73b
|
605 606 |
static struct throtl_grp * throtl_rb_first(struct throtl_service_queue *parent_sq) |
e43473b7f
|
607 |
{ |
9ff01255a
|
608 |
struct rb_node *n; |
e43473b7f
|
609 |
|
9ff01255a
|
610 611 612 613 614 |
n = rb_first_cached(&parent_sq->pending_tree); WARN_ON_ONCE(!n); if (!n) return NULL; return rb_entry_tg(n); |
e43473b7f
|
615 |
} |
0049af73b
|
616 617 |
static void throtl_rb_erase(struct rb_node *n, struct throtl_service_queue *parent_sq) |
e43473b7f
|
618 |
{ |
9ff01255a
|
619 620 |
rb_erase_cached(n, &parent_sq->pending_tree); RB_CLEAR_NODE(n); |
0049af73b
|
621 |
--parent_sq->nr_pending; |
e43473b7f
|
622 |
} |
0049af73b
|
623 |
static void update_min_dispatch_time(struct throtl_service_queue *parent_sq) |
e43473b7f
|
624 625 |
{ struct throtl_grp *tg; |
0049af73b
|
626 |
tg = throtl_rb_first(parent_sq); |
e43473b7f
|
627 628 |
if (!tg) return; |
0049af73b
|
629 |
parent_sq->first_pending_disptime = tg->disptime; |
e43473b7f
|
630 |
} |
77216b048
|
631 |
static void tg_service_queue_add(struct throtl_grp *tg) |
e43473b7f
|
632 |
{ |
77216b048
|
633 |
struct throtl_service_queue *parent_sq = tg->service_queue.parent_sq; |
9ff01255a
|
634 |
struct rb_node **node = &parent_sq->pending_tree.rb_root.rb_node; |
e43473b7f
|
635 636 637 |
struct rb_node *parent = NULL; struct throtl_grp *__tg; unsigned long key = tg->disptime; |
9ff01255a
|
638 |
bool leftmost = true; |
e43473b7f
|
639 640 641 642 643 644 645 646 647 |
while (*node != NULL) { parent = *node; __tg = rb_entry_tg(parent); if (time_before(key, __tg->disptime)) node = &parent->rb_left; else { node = &parent->rb_right; |
9ff01255a
|
648 |
leftmost = false; |
e43473b7f
|
649 650 |
} } |
e43473b7f
|
651 |
rb_link_node(&tg->rb_node, parent, node); |
9ff01255a
|
652 653 |
rb_insert_color_cached(&tg->rb_node, &parent_sq->pending_tree, leftmost); |
e43473b7f
|
654 |
} |
77216b048
|
655 |
static void throtl_enqueue_tg(struct throtl_grp *tg) |
e43473b7f
|
656 |
{ |
29379674b
|
657 658 659 660 661 |
if (!(tg->flags & THROTL_TG_PENDING)) { tg_service_queue_add(tg); tg->flags |= THROTL_TG_PENDING; tg->service_queue.parent_sq->nr_pending++; } |
e43473b7f
|
662 |
} |
77216b048
|
663 |
static void throtl_dequeue_tg(struct throtl_grp *tg) |
e43473b7f
|
664 |
{ |
29379674b
|
665 666 667 668 |
if (tg->flags & THROTL_TG_PENDING) { throtl_rb_erase(&tg->rb_node, tg->service_queue.parent_sq); tg->flags &= ~THROTL_TG_PENDING; } |
e43473b7f
|
669 |
} |
a9131a27e
|
670 |
/* Call with queue lock held */ |
69df0ab03
|
671 672 |
static void throtl_schedule_pending_timer(struct throtl_service_queue *sq, unsigned long expires) |
a9131a27e
|
673 |
{ |
a41b816c1
|
674 |
unsigned long max_expire = jiffies + 8 * sq_to_td(sq)->throtl_slice; |
06cceedcc
|
675 676 677 678 679 680 681 682 683 684 |
/* * Since we are adjusting the throttle limit dynamically, the sleep * time calculated according to previous limit might be invalid. It's * possible the cgroup sleep time is very long and no other cgroups * have IO running so notify the limit changes. Make sure the cgroup * doesn't sleep too long to avoid the missed notification. */ if (time_after(expires, max_expire)) expires = max_expire; |
69df0ab03
|
685 686 687 |
mod_timer(&sq->pending_timer, expires); throtl_log(sq, "schedule timer. delay=%lu jiffies=%lu", expires - jiffies, jiffies); |
a9131a27e
|
688 |
} |
7f52f98c2
|
689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 |
/** * throtl_schedule_next_dispatch - schedule the next dispatch cycle * @sq: the service_queue to schedule dispatch for * @force: force scheduling * * Arm @sq->pending_timer so that the next dispatch cycle starts on the * dispatch time of the first pending child. Returns %true if either timer * is armed or there's no pending child left. %false if the current * dispatch window is still open and the caller should continue * dispatching. * * If @force is %true, the dispatch timer is always scheduled and this * function is guaranteed to return %true. This is to be used when the * caller can't dispatch itself and needs to invoke pending_timer * unconditionally. Note that forced scheduling is likely to induce short * delay before dispatch starts even if @sq->first_pending_disptime is not * in the future and thus shouldn't be used in hot paths. */ static bool throtl_schedule_next_dispatch(struct throtl_service_queue *sq, bool force) |
e43473b7f
|
709 |
{ |
6a525600f
|
710 |
/* any pending children left? */ |
c9e0332e8
|
711 |
if (!sq->nr_pending) |
7f52f98c2
|
712 |
return true; |
e43473b7f
|
713 |
|
c9e0332e8
|
714 |
update_min_dispatch_time(sq); |
e43473b7f
|
715 |
|
69df0ab03
|
716 |
/* is the next dispatch time in the future? */ |
7f52f98c2
|
717 |
if (force || time_after(sq->first_pending_disptime, jiffies)) { |
69df0ab03
|
718 |
throtl_schedule_pending_timer(sq, sq->first_pending_disptime); |
7f52f98c2
|
719 |
return true; |
69df0ab03
|
720 |
} |
7f52f98c2
|
721 722 |
/* tell the caller to continue dispatching */ return false; |
e43473b7f
|
723 |
} |
32ee5bc47
|
724 725 726 727 728 |
static inline void throtl_start_new_slice_with_credit(struct throtl_grp *tg, bool rw, unsigned long start) { tg->bytes_disp[rw] = 0; tg->io_disp[rw] = 0; |
4f1e9630a
|
729 |
atomic_set(&tg->io_split_cnt[rw], 0); |
32ee5bc47
|
730 731 732 733 734 735 736 737 |
/* * Previous slice has expired. We must have trimmed it after last * bio dispatch. That means since start of last slice, we never used * that bandwidth. Do try to make use of that bandwidth while giving * credit. */ if (time_after_eq(start, tg->slice_start[rw])) tg->slice_start[rw] = start; |
297e3d854
|
738 |
tg->slice_end[rw] = jiffies + tg->td->throtl_slice; |
32ee5bc47
|
739 740 741 742 743 |
throtl_log(&tg->service_queue, "[%c] new slice with credit start=%lu end=%lu jiffies=%lu", rw == READ ? 'R' : 'W', tg->slice_start[rw], tg->slice_end[rw], jiffies); } |
0f3457f60
|
744 |
static inline void throtl_start_new_slice(struct throtl_grp *tg, bool rw) |
e43473b7f
|
745 746 |
{ tg->bytes_disp[rw] = 0; |
8e89d13f4
|
747 |
tg->io_disp[rw] = 0; |
e43473b7f
|
748 |
tg->slice_start[rw] = jiffies; |
297e3d854
|
749 |
tg->slice_end[rw] = jiffies + tg->td->throtl_slice; |
4f1e9630a
|
750 751 |
atomic_set(&tg->io_split_cnt[rw], 0); |
fda6f272c
|
752 753 754 755 |
throtl_log(&tg->service_queue, "[%c] new slice start=%lu end=%lu jiffies=%lu", rw == READ ? 'R' : 'W', tg->slice_start[rw], tg->slice_end[rw], jiffies); |
e43473b7f
|
756 |
} |
0f3457f60
|
757 758 |
static inline void throtl_set_slice_end(struct throtl_grp *tg, bool rw, unsigned long jiffy_end) |
d1ae8ffdf
|
759 |
{ |
297e3d854
|
760 |
tg->slice_end[rw] = roundup(jiffy_end, tg->td->throtl_slice); |
d1ae8ffdf
|
761 |
} |
0f3457f60
|
762 763 |
static inline void throtl_extend_slice(struct throtl_grp *tg, bool rw, unsigned long jiffy_end) |
e43473b7f
|
764 |
{ |
1da30f952
|
765 |
throtl_set_slice_end(tg, rw, jiffy_end); |
fda6f272c
|
766 767 768 769 |
throtl_log(&tg->service_queue, "[%c] extend slice start=%lu end=%lu jiffies=%lu", rw == READ ? 'R' : 'W', tg->slice_start[rw], tg->slice_end[rw], jiffies); |
e43473b7f
|
770 771 772 |
} /* Determine if previously allocated or extended slice is complete or not */ |
0f3457f60
|
773 |
static bool throtl_slice_used(struct throtl_grp *tg, bool rw) |
e43473b7f
|
774 775 |
{ if (time_in_range(jiffies, tg->slice_start[rw], tg->slice_end[rw])) |
5cf8c2277
|
776 |
return false; |
e43473b7f
|
777 |
|
0b6bad7d6
|
778 |
return true; |
e43473b7f
|
779 780 781 |
} /* Trim the used slices and adjust slice start accordingly */ |
0f3457f60
|
782 |
static inline void throtl_trim_slice(struct throtl_grp *tg, bool rw) |
e43473b7f
|
783 |
{ |
3aad5d3ee
|
784 785 |
unsigned long nr_slices, time_elapsed, io_trim; u64 bytes_trim, tmp; |
e43473b7f
|
786 787 788 789 790 791 792 793 |
BUG_ON(time_before(tg->slice_end[rw], tg->slice_start[rw])); /* * If bps are unlimited (-1), then time slice don't get * renewed. Don't try to trim the slice if slice is used. A new * slice will start when appropriate. */ |
0f3457f60
|
794 |
if (throtl_slice_used(tg, rw)) |
e43473b7f
|
795 |
return; |
d1ae8ffdf
|
796 797 798 |
/* * A bio has been dispatched. Also adjust slice_end. It might happen * that initially cgroup limit was very low resulting in high |
b53b072c4
|
799 |
* slice_end, but later limit was bumped up and bio was dispatched |
d1ae8ffdf
|
800 801 802 |
* sooner, then we need to reduce slice_end. A high bogus slice_end * is bad because it does not allow new slice to start. */ |
297e3d854
|
803 |
throtl_set_slice_end(tg, rw, jiffies + tg->td->throtl_slice); |
d1ae8ffdf
|
804 |
|
e43473b7f
|
805 |
time_elapsed = jiffies - tg->slice_start[rw]; |
297e3d854
|
806 |
nr_slices = time_elapsed / tg->td->throtl_slice; |
e43473b7f
|
807 808 809 |
if (!nr_slices) return; |
297e3d854
|
810 |
tmp = tg_bps_limit(tg, rw) * tg->td->throtl_slice * nr_slices; |
3aad5d3ee
|
811 812 |
do_div(tmp, HZ); bytes_trim = tmp; |
e43473b7f
|
813 |
|
297e3d854
|
814 815 |
io_trim = (tg_iops_limit(tg, rw) * tg->td->throtl_slice * nr_slices) / HZ; |
e43473b7f
|
816 |
|
8e89d13f4
|
817 |
if (!bytes_trim && !io_trim) |
e43473b7f
|
818 819 820 821 822 823 |
return; if (tg->bytes_disp[rw] >= bytes_trim) tg->bytes_disp[rw] -= bytes_trim; else tg->bytes_disp[rw] = 0; |
8e89d13f4
|
824 825 826 827 |
if (tg->io_disp[rw] >= io_trim) tg->io_disp[rw] -= io_trim; else tg->io_disp[rw] = 0; |
297e3d854
|
828 |
tg->slice_start[rw] += nr_slices * tg->td->throtl_slice; |
e43473b7f
|
829 |
|
fda6f272c
|
830 831 832 833 |
throtl_log(&tg->service_queue, "[%c] trim slice nr=%lu bytes=%llu io=%lu start=%lu end=%lu jiffies=%lu", rw == READ ? 'R' : 'W', nr_slices, bytes_trim, io_trim, tg->slice_start[rw], tg->slice_end[rw], jiffies); |
e43473b7f
|
834 |
} |
0f3457f60
|
835 |
static bool tg_with_in_iops_limit(struct throtl_grp *tg, struct bio *bio, |
4599ea49d
|
836 |
u32 iops_limit, unsigned long *wait) |
e43473b7f
|
837 838 |
{ bool rw = bio_data_dir(bio); |
8e89d13f4
|
839 |
unsigned int io_allowed; |
e43473b7f
|
840 |
unsigned long jiffy_elapsed, jiffy_wait, jiffy_elapsed_rnd; |
c49c06e49
|
841 |
u64 tmp; |
e43473b7f
|
842 |
|
87fbeb881
|
843 844 845 846 847 |
if (iops_limit == UINT_MAX) { if (wait) *wait = 0; return true; } |
3a10f999f
|
848 |
jiffy_elapsed = jiffies - tg->slice_start[rw]; |
e43473b7f
|
849 |
|
3a10f999f
|
850 851 |
/* Round up to the next throttle slice, wait time must be nonzero */ jiffy_elapsed_rnd = roundup(jiffy_elapsed + 1, tg->td->throtl_slice); |
8e89d13f4
|
852 |
|
c49c06e49
|
853 854 855 856 857 858 |
/* * jiffy_elapsed_rnd should not be a big value as minimum iops can be * 1 then at max jiffy elapsed should be equivalent of 1 second as we * will allow dispatch after 1 second and after that slice should * have been trimmed. */ |
4599ea49d
|
859 |
tmp = (u64)iops_limit * jiffy_elapsed_rnd; |
c49c06e49
|
860 861 862 863 864 865 |
do_div(tmp, HZ); if (tmp > UINT_MAX) io_allowed = UINT_MAX; else io_allowed = tmp; |
8e89d13f4
|
866 867 |
if (tg->io_disp[rw] + 1 <= io_allowed) { |
e43473b7f
|
868 869 |
if (wait) *wait = 0; |
5cf8c2277
|
870 |
return true; |
e43473b7f
|
871 |
} |
8e89d13f4
|
872 |
/* Calc approx time to dispatch */ |
991f61fe7
|
873 |
jiffy_wait = jiffy_elapsed_rnd - jiffy_elapsed; |
8e89d13f4
|
874 875 876 |
if (wait) *wait = jiffy_wait; |
0b6bad7d6
|
877 |
return false; |
8e89d13f4
|
878 |
} |
0f3457f60
|
879 |
static bool tg_with_in_bps_limit(struct throtl_grp *tg, struct bio *bio, |
4599ea49d
|
880 |
u64 bps_limit, unsigned long *wait) |
8e89d13f4
|
881 882 |
{ bool rw = bio_data_dir(bio); |
3aad5d3ee
|
883 |
u64 bytes_allowed, extra_bytes, tmp; |
8e89d13f4
|
884 |
unsigned long jiffy_elapsed, jiffy_wait, jiffy_elapsed_rnd; |
ea0ea2bc6
|
885 |
unsigned int bio_size = throtl_bio_data_size(bio); |
e43473b7f
|
886 |
|
87fbeb881
|
887 888 889 890 891 |
if (bps_limit == U64_MAX) { if (wait) *wait = 0; return true; } |
e43473b7f
|
892 893 894 895 |
jiffy_elapsed = jiffy_elapsed_rnd = jiffies - tg->slice_start[rw]; /* Slice has just started. Consider one slice interval */ if (!jiffy_elapsed) |
297e3d854
|
896 |
jiffy_elapsed_rnd = tg->td->throtl_slice; |
e43473b7f
|
897 |
|
297e3d854
|
898 |
jiffy_elapsed_rnd = roundup(jiffy_elapsed_rnd, tg->td->throtl_slice); |
e43473b7f
|
899 |
|
4599ea49d
|
900 |
tmp = bps_limit * jiffy_elapsed_rnd; |
5e901a2b9
|
901 |
do_div(tmp, HZ); |
3aad5d3ee
|
902 |
bytes_allowed = tmp; |
e43473b7f
|
903 |
|
ea0ea2bc6
|
904 |
if (tg->bytes_disp[rw] + bio_size <= bytes_allowed) { |
e43473b7f
|
905 906 |
if (wait) *wait = 0; |
5cf8c2277
|
907 |
return true; |
e43473b7f
|
908 909 910 |
} /* Calc approx time to dispatch */ |
ea0ea2bc6
|
911 |
extra_bytes = tg->bytes_disp[rw] + bio_size - bytes_allowed; |
4599ea49d
|
912 |
jiffy_wait = div64_u64(extra_bytes * HZ, bps_limit); |
e43473b7f
|
913 914 915 916 917 918 919 920 921 |
if (!jiffy_wait) jiffy_wait = 1; /* * This wait time is without taking into consideration the rounding * up we did. Add that time also. */ jiffy_wait = jiffy_wait + (jiffy_elapsed_rnd - jiffy_elapsed); |
e43473b7f
|
922 923 |
if (wait) *wait = jiffy_wait; |
0b6bad7d6
|
924 |
return false; |
8e89d13f4
|
925 926 927 928 929 930 |
} /* * Returns whether one can dispatch a bio or not. Also returns approx number * of jiffies to wait before this bio is with-in IO rate and can be dispatched */ |
0f3457f60
|
931 932 |
static bool tg_may_dispatch(struct throtl_grp *tg, struct bio *bio, unsigned long *wait) |
8e89d13f4
|
933 934 935 |
{ bool rw = bio_data_dir(bio); unsigned long bps_wait = 0, iops_wait = 0, max_wait = 0; |
4599ea49d
|
936 937 |
u64 bps_limit = tg_bps_limit(tg, rw); u32 iops_limit = tg_iops_limit(tg, rw); |
8e89d13f4
|
938 939 940 941 942 943 944 |
/* * Currently whole state machine of group depends on first bio * queued in the group bio list. So one should not be calling * this function with a different bio if there are other bios * queued. */ |
73f0d49a9
|
945 |
BUG_ON(tg->service_queue.nr_queued[rw] && |
c5cc2070b
|
946 |
bio != throtl_peek_queued(&tg->service_queue.queued[rw])); |
e43473b7f
|
947 |
|
8e89d13f4
|
948 |
/* If tg->bps = -1, then BW is unlimited */ |
4599ea49d
|
949 |
if (bps_limit == U64_MAX && iops_limit == UINT_MAX) { |
8e89d13f4
|
950 951 |
if (wait) *wait = 0; |
5cf8c2277
|
952 |
return true; |
8e89d13f4
|
953 954 955 956 957 |
} /* * If previous slice expired, start a new one otherwise renew/extend * existing slice to make sure it is at least throtl_slice interval |
164c80ed8
|
958 959 960 |
* long since now. New slice is started only for empty throttle group. * If there is queued bio, that means there should be an active * slice and it should be extended instead. |
8e89d13f4
|
961 |
*/ |
164c80ed8
|
962 |
if (throtl_slice_used(tg, rw) && !(tg->service_queue.nr_queued[rw])) |
0f3457f60
|
963 |
throtl_start_new_slice(tg, rw); |
8e89d13f4
|
964 |
else { |
297e3d854
|
965 966 967 968 |
if (time_before(tg->slice_end[rw], jiffies + tg->td->throtl_slice)) throtl_extend_slice(tg, rw, jiffies + tg->td->throtl_slice); |
8e89d13f4
|
969 |
} |
4f1e9630a
|
970 971 |
if (iops_limit != UINT_MAX) tg->io_disp[rw] += atomic_xchg(&tg->io_split_cnt[rw], 0); |
4599ea49d
|
972 973 |
if (tg_with_in_bps_limit(tg, bio, bps_limit, &bps_wait) && tg_with_in_iops_limit(tg, bio, iops_limit, &iops_wait)) { |
8e89d13f4
|
974 975 |
if (wait) *wait = 0; |
0b6bad7d6
|
976 |
return true; |
8e89d13f4
|
977 978 979 980 981 982 983 984 |
} max_wait = max(bps_wait, iops_wait); if (wait) *wait = max_wait; if (time_before(tg->slice_end[rw], jiffies + max_wait)) |
0f3457f60
|
985 |
throtl_extend_slice(tg, rw, jiffies + max_wait); |
e43473b7f
|
986 |
|
0b6bad7d6
|
987 |
return false; |
e43473b7f
|
988 989 990 991 992 |
} static void throtl_charge_bio(struct throtl_grp *tg, struct bio *bio) { bool rw = bio_data_dir(bio); |
ea0ea2bc6
|
993 |
unsigned int bio_size = throtl_bio_data_size(bio); |
e43473b7f
|
994 995 |
/* Charge the bio to the group */ |
ea0ea2bc6
|
996 |
tg->bytes_disp[rw] += bio_size; |
8e89d13f4
|
997 |
tg->io_disp[rw]++; |
ea0ea2bc6
|
998 |
tg->last_bytes_disp[rw] += bio_size; |
3f0abd806
|
999 |
tg->last_io_disp[rw]++; |
e43473b7f
|
1000 |
|
2a0f61e6e
|
1001 |
/* |
8d2bbd4c8
|
1002 |
* BIO_THROTTLED is used to prevent the same bio to be throttled |
2a0f61e6e
|
1003 1004 1005 |
* more than once as a throttled bio will go through blk-throtl the * second time when it eventually gets issued. Set it when a bio * is being charged to a tg. |
2a0f61e6e
|
1006 |
*/ |
8d2bbd4c8
|
1007 1008 |
if (!bio_flagged(bio, BIO_THROTTLED)) bio_set_flag(bio, BIO_THROTTLED); |
e43473b7f
|
1009 |
} |
c5cc2070b
|
1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 |
/** * throtl_add_bio_tg - add a bio to the specified throtl_grp * @bio: bio to add * @qn: qnode to use * @tg: the target throtl_grp * * Add @bio to @tg's service_queue using @qn. If @qn is not specified, * tg->qnode_on_self[] is used. */ static void throtl_add_bio_tg(struct bio *bio, struct throtl_qnode *qn, struct throtl_grp *tg) |
e43473b7f
|
1021 |
{ |
73f0d49a9
|
1022 |
struct throtl_service_queue *sq = &tg->service_queue; |
e43473b7f
|
1023 |
bool rw = bio_data_dir(bio); |
c5cc2070b
|
1024 1025 |
if (!qn) qn = &tg->qnode_on_self[rw]; |
0e9f4164b
|
1026 1027 1028 1029 |
/* * If @tg doesn't currently have any bios queued in the same * direction, queueing @bio can change when @tg should be * dispatched. Mark that @tg was empty. This is automatically |
b53b072c4
|
1030 |
* cleared on the next tg_update_disptime(). |
0e9f4164b
|
1031 1032 1033 |
*/ if (!sq->nr_queued[rw]) tg->flags |= THROTL_TG_WAS_EMPTY; |
c5cc2070b
|
1034 |
throtl_qnode_add_bio(bio, qn, &sq->queued[rw]); |
73f0d49a9
|
1035 |
sq->nr_queued[rw]++; |
77216b048
|
1036 |
throtl_enqueue_tg(tg); |
e43473b7f
|
1037 |
} |
77216b048
|
1038 |
static void tg_update_disptime(struct throtl_grp *tg) |
e43473b7f
|
1039 |
{ |
73f0d49a9
|
1040 |
struct throtl_service_queue *sq = &tg->service_queue; |
e43473b7f
|
1041 1042 |
unsigned long read_wait = -1, write_wait = -1, min_wait = -1, disptime; struct bio *bio; |
d609af3a1
|
1043 1044 |
bio = throtl_peek_queued(&sq->queued[READ]); if (bio) |
0f3457f60
|
1045 |
tg_may_dispatch(tg, bio, &read_wait); |
e43473b7f
|
1046 |
|
d609af3a1
|
1047 1048 |
bio = throtl_peek_queued(&sq->queued[WRITE]); if (bio) |
0f3457f60
|
1049 |
tg_may_dispatch(tg, bio, &write_wait); |
e43473b7f
|
1050 1051 1052 |
min_wait = min(read_wait, write_wait); disptime = jiffies + min_wait; |
e43473b7f
|
1053 |
/* Update dispatch time */ |
77216b048
|
1054 |
throtl_dequeue_tg(tg); |
e43473b7f
|
1055 |
tg->disptime = disptime; |
77216b048
|
1056 |
throtl_enqueue_tg(tg); |
0e9f4164b
|
1057 1058 1059 |
/* see throtl_add_bio_tg() */ tg->flags &= ~THROTL_TG_WAS_EMPTY; |
e43473b7f
|
1060 |
} |
32ee5bc47
|
1061 1062 1063 1064 1065 1066 1067 1068 1069 |
static void start_parent_slice_with_credit(struct throtl_grp *child_tg, struct throtl_grp *parent_tg, bool rw) { if (throtl_slice_used(parent_tg, rw)) { throtl_start_new_slice_with_credit(parent_tg, rw, child_tg->slice_start[rw]); } } |
77216b048
|
1070 |
static void tg_dispatch_one_bio(struct throtl_grp *tg, bool rw) |
e43473b7f
|
1071 |
{ |
73f0d49a9
|
1072 |
struct throtl_service_queue *sq = &tg->service_queue; |
6bc9c2b46
|
1073 1074 |
struct throtl_service_queue *parent_sq = sq->parent_sq; struct throtl_grp *parent_tg = sq_to_tg(parent_sq); |
c5cc2070b
|
1075 |
struct throtl_grp *tg_to_put = NULL; |
e43473b7f
|
1076 |
struct bio *bio; |
c5cc2070b
|
1077 1078 1079 1080 1081 1082 1083 |
/* * @bio is being transferred from @tg to @parent_sq. Popping a bio * from @tg may put its reference and @parent_sq might end up * getting released prematurely. Remember the tg to put and put it * after @bio is transferred to @parent_sq. */ bio = throtl_pop_queued(&sq->queued[rw], &tg_to_put); |
73f0d49a9
|
1084 |
sq->nr_queued[rw]--; |
e43473b7f
|
1085 1086 |
throtl_charge_bio(tg, bio); |
6bc9c2b46
|
1087 1088 1089 1090 1091 1092 1093 1094 1095 |
/* * If our parent is another tg, we just need to transfer @bio to * the parent using throtl_add_bio_tg(). If our parent is * @td->service_queue, @bio is ready to be issued. Put it on its * bio_lists[] and decrease total number queued. The caller is * responsible for issuing these bios. */ if (parent_tg) { |
c5cc2070b
|
1096 |
throtl_add_bio_tg(bio, &tg->qnode_on_parent[rw], parent_tg); |
32ee5bc47
|
1097 |
start_parent_slice_with_credit(tg, parent_tg, rw); |
6bc9c2b46
|
1098 |
} else { |
c5cc2070b
|
1099 1100 |
throtl_qnode_add_bio(bio, &tg->qnode_on_parent[rw], &parent_sq->queued[rw]); |
6bc9c2b46
|
1101 1102 1103 |
BUG_ON(tg->td->nr_queued[rw] <= 0); tg->td->nr_queued[rw]--; } |
e43473b7f
|
1104 |
|
0f3457f60
|
1105 |
throtl_trim_slice(tg, rw); |
6bc9c2b46
|
1106 |
|
c5cc2070b
|
1107 1108 |
if (tg_to_put) blkg_put(tg_to_blkg(tg_to_put)); |
e43473b7f
|
1109 |
} |
77216b048
|
1110 |
static int throtl_dispatch_tg(struct throtl_grp *tg) |
e43473b7f
|
1111 |
{ |
73f0d49a9
|
1112 |
struct throtl_service_queue *sq = &tg->service_queue; |
e43473b7f
|
1113 |
unsigned int nr_reads = 0, nr_writes = 0; |
e675df2ad
|
1114 1115 |
unsigned int max_nr_reads = THROTL_GRP_QUANTUM * 3 / 4; unsigned int max_nr_writes = THROTL_GRP_QUANTUM - max_nr_reads; |
e43473b7f
|
1116 1117 1118 |
struct bio *bio; /* Try to dispatch 75% READS and 25% WRITES */ |
c5cc2070b
|
1119 |
while ((bio = throtl_peek_queued(&sq->queued[READ])) && |
0f3457f60
|
1120 |
tg_may_dispatch(tg, bio, NULL)) { |
e43473b7f
|
1121 |
|
77216b048
|
1122 |
tg_dispatch_one_bio(tg, bio_data_dir(bio)); |
e43473b7f
|
1123 1124 1125 1126 1127 |
nr_reads++; if (nr_reads >= max_nr_reads) break; } |
c5cc2070b
|
1128 |
while ((bio = throtl_peek_queued(&sq->queued[WRITE])) && |
0f3457f60
|
1129 |
tg_may_dispatch(tg, bio, NULL)) { |
e43473b7f
|
1130 |
|
77216b048
|
1131 |
tg_dispatch_one_bio(tg, bio_data_dir(bio)); |
e43473b7f
|
1132 1133 1134 1135 1136 1137 1138 1139 |
nr_writes++; if (nr_writes >= max_nr_writes) break; } return nr_reads + nr_writes; } |
651930bc1
|
1140 |
static int throtl_select_dispatch(struct throtl_service_queue *parent_sq) |
e43473b7f
|
1141 1142 |
{ unsigned int nr_disp = 0; |
e43473b7f
|
1143 1144 |
while (1) { |
2397611ac
|
1145 |
struct throtl_grp *tg; |
2ab74cd29
|
1146 |
struct throtl_service_queue *sq; |
e43473b7f
|
1147 |
|
2397611ac
|
1148 1149 1150 1151 |
if (!parent_sq->nr_pending) break; tg = throtl_rb_first(parent_sq); |
e43473b7f
|
1152 1153 1154 1155 1156 |
if (!tg) break; if (time_before(jiffies, tg->disptime)) break; |
77216b048
|
1157 |
throtl_dequeue_tg(tg); |
e43473b7f
|
1158 |
|
77216b048
|
1159 |
nr_disp += throtl_dispatch_tg(tg); |
e43473b7f
|
1160 |
|
2ab74cd29
|
1161 |
sq = &tg->service_queue; |
73f0d49a9
|
1162 |
if (sq->nr_queued[0] || sq->nr_queued[1]) |
77216b048
|
1163 |
tg_update_disptime(tg); |
e43473b7f
|
1164 |
|
e675df2ad
|
1165 |
if (nr_disp >= THROTL_QUANTUM) |
e43473b7f
|
1166 1167 1168 1169 1170 |
break; } return nr_disp; } |
c79892c55
|
1171 1172 |
static bool throtl_can_upgrade(struct throtl_data *td, struct throtl_grp *this_tg); |
6e1a5704c
|
1173 1174 |
/** * throtl_pending_timer_fn - timer function for service_queue->pending_timer |
216382dcc
|
1175 |
* @t: the pending_timer member of the throtl_service_queue being serviced |
6e1a5704c
|
1176 1177 1178 1179 |
* * This timer is armed when a child throtl_grp with active bio's become * pending and queued on the service_queue's pending_tree and expires when * the first child throtl_grp should be dispatched. This function |
2e48a530a
|
1180 1181 1182 1183 1184 1185 1186 |
* dispatches bio's from the children throtl_grps to the parent * service_queue. * * If the parent's parent is another throtl_grp, dispatching is propagated * by either arming its pending_timer or repeating dispatch directly. If * the top-level service_tree is reached, throtl_data->dispatch_work is * kicked so that the ready bio's are issued. |
6e1a5704c
|
1187 |
*/ |
e99e88a9d
|
1188 |
static void throtl_pending_timer_fn(struct timer_list *t) |
69df0ab03
|
1189 |
{ |
e99e88a9d
|
1190 |
struct throtl_service_queue *sq = from_timer(sq, t, pending_timer); |
2e48a530a
|
1191 |
struct throtl_grp *tg = sq_to_tg(sq); |
69df0ab03
|
1192 |
struct throtl_data *td = sq_to_td(sq); |
cb76199c3
|
1193 |
struct request_queue *q = td->queue; |
2e48a530a
|
1194 1195 |
struct throtl_service_queue *parent_sq; bool dispatched; |
6e1a5704c
|
1196 |
int ret; |
e43473b7f
|
1197 |
|
0d945c1f9
|
1198 |
spin_lock_irq(&q->queue_lock); |
c79892c55
|
1199 1200 |
if (throtl_can_upgrade(td, NULL)) throtl_upgrade_state(td); |
2e48a530a
|
1201 1202 1203 |
again: parent_sq = sq->parent_sq; dispatched = false; |
e43473b7f
|
1204 |
|
7f52f98c2
|
1205 1206 |
while (true) { throtl_log(sq, "dispatch nr_queued=%u read=%u write=%u", |
2e48a530a
|
1207 1208 |
sq->nr_queued[READ] + sq->nr_queued[WRITE], sq->nr_queued[READ], sq->nr_queued[WRITE]); |
7f52f98c2
|
1209 1210 1211 |
ret = throtl_select_dispatch(sq); if (ret) { |
7f52f98c2
|
1212 1213 1214 |
throtl_log(sq, "bios disp=%u", ret); dispatched = true; } |
e43473b7f
|
1215 |
|
7f52f98c2
|
1216 1217 |
if (throtl_schedule_next_dispatch(sq, false)) break; |
e43473b7f
|
1218 |
|
7f52f98c2
|
1219 |
/* this dispatch windows is still open, relax and repeat */ |
0d945c1f9
|
1220 |
spin_unlock_irq(&q->queue_lock); |
7f52f98c2
|
1221 |
cpu_relax(); |
0d945c1f9
|
1222 |
spin_lock_irq(&q->queue_lock); |
651930bc1
|
1223 |
} |
e43473b7f
|
1224 |
|
2e48a530a
|
1225 1226 |
if (!dispatched) goto out_unlock; |
6e1a5704c
|
1227 |
|
2e48a530a
|
1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 |
if (parent_sq) { /* @parent_sq is another throl_grp, propagate dispatch */ if (tg->flags & THROTL_TG_WAS_EMPTY) { tg_update_disptime(tg); if (!throtl_schedule_next_dispatch(parent_sq, false)) { /* window is already open, repeat dispatching */ sq = parent_sq; tg = sq_to_tg(sq); goto again; } } } else { |
b53b072c4
|
1240 |
/* reached the top-level, queue issuing */ |
2e48a530a
|
1241 1242 1243 |
queue_work(kthrotld_workqueue, &td->dispatch_work); } out_unlock: |
0d945c1f9
|
1244 |
spin_unlock_irq(&q->queue_lock); |
6e1a5704c
|
1245 |
} |
e43473b7f
|
1246 |
|
6e1a5704c
|
1247 1248 1249 1250 |
/** * blk_throtl_dispatch_work_fn - work function for throtl_data->dispatch_work * @work: work item being executed * |
b53b072c4
|
1251 1252 |
* This function is queued for execution when bios reach the bio_lists[] * of throtl_data->service_queue. Those bios are ready and issued by this |
6e1a5704c
|
1253 1254 |
* function. */ |
8876e140e
|
1255 |
static void blk_throtl_dispatch_work_fn(struct work_struct *work) |
6e1a5704c
|
1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 |
{ struct throtl_data *td = container_of(work, struct throtl_data, dispatch_work); struct throtl_service_queue *td_sq = &td->service_queue; struct request_queue *q = td->queue; struct bio_list bio_list_on_stack; struct bio *bio; struct blk_plug plug; int rw; bio_list_init(&bio_list_on_stack); |
0d945c1f9
|
1267 |
spin_lock_irq(&q->queue_lock); |
c5cc2070b
|
1268 1269 1270 |
for (rw = READ; rw <= WRITE; rw++) while ((bio = throtl_pop_queued(&td_sq->queued[rw], NULL))) bio_list_add(&bio_list_on_stack, bio); |
0d945c1f9
|
1271 |
spin_unlock_irq(&q->queue_lock); |
6e1a5704c
|
1272 1273 |
if (!bio_list_empty(&bio_list_on_stack)) { |
69d60eb96
|
1274 |
blk_start_plug(&plug); |
ed00aabd5
|
1275 1276 |
while ((bio = bio_list_pop(&bio_list_on_stack))) submit_bio_noacct(bio); |
69d60eb96
|
1277 |
blk_finish_plug(&plug); |
e43473b7f
|
1278 |
} |
e43473b7f
|
1279 |
} |
f95a04afa
|
1280 1281 |
static u64 tg_prfill_conf_u64(struct seq_file *sf, struct blkg_policy_data *pd, int off) |
60c2bc2d5
|
1282 |
{ |
f95a04afa
|
1283 1284 |
struct throtl_grp *tg = pd_to_tg(pd); u64 v = *(u64 *)((void *)tg + off); |
60c2bc2d5
|
1285 |
|
2ab5492de
|
1286 |
if (v == U64_MAX) |
60c2bc2d5
|
1287 |
return 0; |
f95a04afa
|
1288 |
return __blkg_prfill_u64(sf, pd, v); |
60c2bc2d5
|
1289 |
} |
f95a04afa
|
1290 1291 |
static u64 tg_prfill_conf_uint(struct seq_file *sf, struct blkg_policy_data *pd, int off) |
e43473b7f
|
1292 |
{ |
f95a04afa
|
1293 1294 |
struct throtl_grp *tg = pd_to_tg(pd); unsigned int v = *(unsigned int *)((void *)tg + off); |
fe0714377
|
1295 |
|
2ab5492de
|
1296 |
if (v == UINT_MAX) |
af133ceb2
|
1297 |
return 0; |
f95a04afa
|
1298 |
return __blkg_prfill_u64(sf, pd, v); |
e43473b7f
|
1299 |
} |
2da8ca822
|
1300 |
static int tg_print_conf_u64(struct seq_file *sf, void *v) |
8e89d13f4
|
1301 |
{ |
2da8ca822
|
1302 1303 |
blkcg_print_blkgs(sf, css_to_blkcg(seq_css(sf)), tg_prfill_conf_u64, &blkcg_policy_throtl, seq_cft(sf)->private, false); |
af133ceb2
|
1304 |
return 0; |
8e89d13f4
|
1305 |
} |
2da8ca822
|
1306 |
static int tg_print_conf_uint(struct seq_file *sf, void *v) |
8e89d13f4
|
1307 |
{ |
2da8ca822
|
1308 1309 |
blkcg_print_blkgs(sf, css_to_blkcg(seq_css(sf)), tg_prfill_conf_uint, &blkcg_policy_throtl, seq_cft(sf)->private, false); |
af133ceb2
|
1310 |
return 0; |
60c2bc2d5
|
1311 |
} |
9bb67aeb9
|
1312 |
static void tg_conf_updated(struct throtl_grp *tg, bool global) |
60c2bc2d5
|
1313 |
{ |
69948b070
|
1314 |
struct throtl_service_queue *sq = &tg->service_queue; |
492eb21b9
|
1315 |
struct cgroup_subsys_state *pos_css; |
69948b070
|
1316 |
struct blkcg_gq *blkg; |
af133ceb2
|
1317 |
|
fda6f272c
|
1318 1319 |
throtl_log(&tg->service_queue, "limit change rbps=%llu wbps=%llu riops=%u wiops=%u", |
9f626e372
|
1320 1321 |
tg_bps_limit(tg, READ), tg_bps_limit(tg, WRITE), tg_iops_limit(tg, READ), tg_iops_limit(tg, WRITE)); |
632b44935
|
1322 1323 |
/* |
693e751e7
|
1324 1325 1326 1327 1328 1329 |
* Update has_rules[] flags for the updated tg's subtree. A tg is * considered to have rules if either the tg itself or any of its * ancestors has rules. This identifies groups without any * restrictions in the whole hierarchy and allows them to bypass * blk-throttle. */ |
9bb67aeb9
|
1330 1331 |
blkg_for_each_descendant_pre(blkg, pos_css, global ? tg->td->queue->root_blkg : tg_to_blkg(tg)) { |
5b81fc3cc
|
1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 |
struct throtl_grp *this_tg = blkg_to_tg(blkg); struct throtl_grp *parent_tg; tg_update_has_rules(this_tg); /* ignore root/second level */ if (!cgroup_subsys_on_dfl(io_cgrp_subsys) || !blkg->parent || !blkg->parent->parent) continue; parent_tg = blkg_to_tg(blkg->parent); /* * make sure all children has lower idle time threshold and * higher latency target */ this_tg->idletime_threshold = min(this_tg->idletime_threshold, parent_tg->idletime_threshold); this_tg->latency_target = max(this_tg->latency_target, parent_tg->latency_target); } |
693e751e7
|
1350 1351 |
/* |
632b44935
|
1352 1353 1354 1355 1356 1357 1358 |
* We're already holding queue_lock and know @tg is valid. Let's * apply the new config directly. * * Restart the slices for both READ and WRITES. It might happen * that a group's limit are dropped suddenly and we don't want to * account recently dispatched IO with new low rate. */ |
ff8b22c0f
|
1359 1360 |
throtl_start_new_slice(tg, READ); throtl_start_new_slice(tg, WRITE); |
632b44935
|
1361 |
|
5b2c16aae
|
1362 |
if (tg->flags & THROTL_TG_PENDING) { |
77216b048
|
1363 |
tg_update_disptime(tg); |
7f52f98c2
|
1364 |
throtl_schedule_next_dispatch(sq->parent_sq, true); |
632b44935
|
1365 |
} |
69948b070
|
1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 |
} static ssize_t tg_set_conf(struct kernfs_open_file *of, char *buf, size_t nbytes, loff_t off, bool is_u64) { struct blkcg *blkcg = css_to_blkcg(of_css(of)); struct blkg_conf_ctx ctx; struct throtl_grp *tg; int ret; u64 v; ret = blkg_conf_prep(blkcg, &blkcg_policy_throtl, buf, &ctx); if (ret) return ret; ret = -EINVAL; if (sscanf(ctx.body, "%llu", &v) != 1) goto out_finish; if (!v) |
2ab5492de
|
1385 |
v = U64_MAX; |
69948b070
|
1386 1387 1388 1389 1390 1391 1392 |
tg = blkg_to_tg(ctx.blkg); if (is_u64) *(u64 *)((void *)tg + of_cft(of)->private) = v; else *(unsigned int *)((void *)tg + of_cft(of)->private) = v; |
60c2bc2d5
|
1393 |
|
9bb67aeb9
|
1394 |
tg_conf_updated(tg, false); |
36aa9e5f5
|
1395 1396 |
ret = 0; out_finish: |
60c2bc2d5
|
1397 |
blkg_conf_finish(&ctx); |
36aa9e5f5
|
1398 |
return ret ?: nbytes; |
8e89d13f4
|
1399 |
} |
451af504d
|
1400 1401 |
static ssize_t tg_set_conf_u64(struct kernfs_open_file *of, char *buf, size_t nbytes, loff_t off) |
60c2bc2d5
|
1402 |
{ |
451af504d
|
1403 |
return tg_set_conf(of, buf, nbytes, off, true); |
60c2bc2d5
|
1404 |
} |
451af504d
|
1405 1406 |
static ssize_t tg_set_conf_uint(struct kernfs_open_file *of, char *buf, size_t nbytes, loff_t off) |
60c2bc2d5
|
1407 |
{ |
451af504d
|
1408 |
return tg_set_conf(of, buf, nbytes, off, false); |
60c2bc2d5
|
1409 |
} |
7ca464383
|
1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 |
static int tg_print_rwstat(struct seq_file *sf, void *v) { blkcg_print_blkgs(sf, css_to_blkcg(seq_css(sf)), blkg_prfill_rwstat, &blkcg_policy_throtl, seq_cft(sf)->private, true); return 0; } static u64 tg_prfill_rwstat_recursive(struct seq_file *sf, struct blkg_policy_data *pd, int off) { struct blkg_rwstat_sample sum; blkg_rwstat_recursive_sum(pd_to_blkg(pd), &blkcg_policy_throtl, off, &sum); return __blkg_prfill_rwstat(sf, pd, &sum); } static int tg_print_rwstat_recursive(struct seq_file *sf, void *v) { blkcg_print_blkgs(sf, css_to_blkcg(seq_css(sf)), tg_prfill_rwstat_recursive, &blkcg_policy_throtl, seq_cft(sf)->private, true); return 0; } |
880f50e22
|
1435 |
static struct cftype throtl_legacy_files[] = { |
60c2bc2d5
|
1436 1437 |
{ .name = "throttle.read_bps_device", |
9f626e372
|
1438 |
.private = offsetof(struct throtl_grp, bps[READ][LIMIT_MAX]), |
2da8ca822
|
1439 |
.seq_show = tg_print_conf_u64, |
451af504d
|
1440 |
.write = tg_set_conf_u64, |
60c2bc2d5
|
1441 1442 1443 |
}, { .name = "throttle.write_bps_device", |
9f626e372
|
1444 |
.private = offsetof(struct throtl_grp, bps[WRITE][LIMIT_MAX]), |
2da8ca822
|
1445 |
.seq_show = tg_print_conf_u64, |
451af504d
|
1446 |
.write = tg_set_conf_u64, |
60c2bc2d5
|
1447 1448 1449 |
}, { .name = "throttle.read_iops_device", |
9f626e372
|
1450 |
.private = offsetof(struct throtl_grp, iops[READ][LIMIT_MAX]), |
2da8ca822
|
1451 |
.seq_show = tg_print_conf_uint, |
451af504d
|
1452 |
.write = tg_set_conf_uint, |
60c2bc2d5
|
1453 1454 1455 |
}, { .name = "throttle.write_iops_device", |
9f626e372
|
1456 |
.private = offsetof(struct throtl_grp, iops[WRITE][LIMIT_MAX]), |
2da8ca822
|
1457 |
.seq_show = tg_print_conf_uint, |
451af504d
|
1458 |
.write = tg_set_conf_uint, |
60c2bc2d5
|
1459 1460 1461 |
}, { .name = "throttle.io_service_bytes", |
7ca464383
|
1462 1463 |
.private = offsetof(struct throtl_grp, stat_bytes), .seq_show = tg_print_rwstat, |
60c2bc2d5
|
1464 1465 |
}, { |
17534c6f2
|
1466 |
.name = "throttle.io_service_bytes_recursive", |
7ca464383
|
1467 1468 |
.private = offsetof(struct throtl_grp, stat_bytes), .seq_show = tg_print_rwstat_recursive, |
17534c6f2
|
1469 1470 |
}, { |
60c2bc2d5
|
1471 |
.name = "throttle.io_serviced", |
7ca464383
|
1472 1473 |
.private = offsetof(struct throtl_grp, stat_ios), .seq_show = tg_print_rwstat, |
60c2bc2d5
|
1474 |
}, |
17534c6f2
|
1475 1476 |
{ .name = "throttle.io_serviced_recursive", |
7ca464383
|
1477 1478 |
.private = offsetof(struct throtl_grp, stat_ios), .seq_show = tg_print_rwstat_recursive, |
17534c6f2
|
1479 |
}, |
60c2bc2d5
|
1480 1481 |
{ } /* terminate */ }; |
cd5ab1b0f
|
1482 |
static u64 tg_prfill_limit(struct seq_file *sf, struct blkg_policy_data *pd, |
2ee867dcf
|
1483 1484 1485 1486 1487 |
int off) { struct throtl_grp *tg = pd_to_tg(pd); const char *dname = blkg_dev_name(pd->blkg); char bufs[4][21] = { "max", "max", "max", "max" }; |
cd5ab1b0f
|
1488 1489 |
u64 bps_dft; unsigned int iops_dft; |
ada75b6e5
|
1490 |
char idle_time[26] = ""; |
ec80991d6
|
1491 |
char latency_time[26] = ""; |
2ee867dcf
|
1492 1493 1494 |
if (!dname) return 0; |
9f626e372
|
1495 |
|
cd5ab1b0f
|
1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 |
if (off == LIMIT_LOW) { bps_dft = 0; iops_dft = 0; } else { bps_dft = U64_MAX; iops_dft = UINT_MAX; } if (tg->bps_conf[READ][off] == bps_dft && tg->bps_conf[WRITE][off] == bps_dft && tg->iops_conf[READ][off] == iops_dft && |
ada75b6e5
|
1507 |
tg->iops_conf[WRITE][off] == iops_dft && |
ec80991d6
|
1508 |
(off != LIMIT_LOW || |
b4f428ef2
|
1509 |
(tg->idletime_threshold_conf == DFL_IDLE_THRESHOLD && |
5b81fc3cc
|
1510 |
tg->latency_target_conf == DFL_LATENCY_TARGET))) |
2ee867dcf
|
1511 |
return 0; |
9bb67aeb9
|
1512 |
if (tg->bps_conf[READ][off] != U64_MAX) |
9f626e372
|
1513 |
snprintf(bufs[0], sizeof(bufs[0]), "%llu", |
cd5ab1b0f
|
1514 |
tg->bps_conf[READ][off]); |
9bb67aeb9
|
1515 |
if (tg->bps_conf[WRITE][off] != U64_MAX) |
9f626e372
|
1516 |
snprintf(bufs[1], sizeof(bufs[1]), "%llu", |
cd5ab1b0f
|
1517 |
tg->bps_conf[WRITE][off]); |
9bb67aeb9
|
1518 |
if (tg->iops_conf[READ][off] != UINT_MAX) |
9f626e372
|
1519 |
snprintf(bufs[2], sizeof(bufs[2]), "%u", |
cd5ab1b0f
|
1520 |
tg->iops_conf[READ][off]); |
9bb67aeb9
|
1521 |
if (tg->iops_conf[WRITE][off] != UINT_MAX) |
9f626e372
|
1522 |
snprintf(bufs[3], sizeof(bufs[3]), "%u", |
cd5ab1b0f
|
1523 |
tg->iops_conf[WRITE][off]); |
ada75b6e5
|
1524 |
if (off == LIMIT_LOW) { |
5b81fc3cc
|
1525 |
if (tg->idletime_threshold_conf == ULONG_MAX) |
ada75b6e5
|
1526 1527 1528 |
strcpy(idle_time, " idle=max"); else snprintf(idle_time, sizeof(idle_time), " idle=%lu", |
5b81fc3cc
|
1529 |
tg->idletime_threshold_conf); |
ec80991d6
|
1530 |
|
5b81fc3cc
|
1531 |
if (tg->latency_target_conf == ULONG_MAX) |
ec80991d6
|
1532 1533 1534 |
strcpy(latency_time, " latency=max"); else snprintf(latency_time, sizeof(latency_time), |
5b81fc3cc
|
1535 |
" latency=%lu", tg->latency_target_conf); |
ada75b6e5
|
1536 |
} |
2ee867dcf
|
1537 |
|
ec80991d6
|
1538 1539 1540 1541 |
seq_printf(sf, "%s rbps=%s wbps=%s riops=%s wiops=%s%s%s ", dname, bufs[0], bufs[1], bufs[2], bufs[3], idle_time, latency_time); |
2ee867dcf
|
1542 1543 |
return 0; } |
cd5ab1b0f
|
1544 |
static int tg_print_limit(struct seq_file *sf, void *v) |
2ee867dcf
|
1545 |
{ |
cd5ab1b0f
|
1546 |
blkcg_print_blkgs(sf, css_to_blkcg(seq_css(sf)), tg_prfill_limit, |
2ee867dcf
|
1547 1548 1549 |
&blkcg_policy_throtl, seq_cft(sf)->private, false); return 0; } |
cd5ab1b0f
|
1550 |
static ssize_t tg_set_limit(struct kernfs_open_file *of, |
2ee867dcf
|
1551 1552 1553 1554 1555 1556 |
char *buf, size_t nbytes, loff_t off) { struct blkcg *blkcg = css_to_blkcg(of_css(of)); struct blkg_conf_ctx ctx; struct throtl_grp *tg; u64 v[4]; |
ada75b6e5
|
1557 |
unsigned long idle_time; |
ec80991d6
|
1558 |
unsigned long latency_time; |
2ee867dcf
|
1559 |
int ret; |
cd5ab1b0f
|
1560 |
int index = of_cft(of)->private; |
2ee867dcf
|
1561 1562 1563 1564 1565 1566 |
ret = blkg_conf_prep(blkcg, &blkcg_policy_throtl, buf, &ctx); if (ret) return ret; tg = blkg_to_tg(ctx.blkg); |
cd5ab1b0f
|
1567 1568 1569 1570 |
v[0] = tg->bps_conf[READ][index]; v[1] = tg->bps_conf[WRITE][index]; v[2] = tg->iops_conf[READ][index]; v[3] = tg->iops_conf[WRITE][index]; |
2ee867dcf
|
1571 |
|
5b81fc3cc
|
1572 1573 |
idle_time = tg->idletime_threshold_conf; latency_time = tg->latency_target_conf; |
2ee867dcf
|
1574 1575 1576 |
while (true) { char tok[27]; /* wiops=18446744073709551616 */ char *p; |
2ab5492de
|
1577 |
u64 val = U64_MAX; |
2ee867dcf
|
1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 |
int len; if (sscanf(ctx.body, "%26s%n", tok, &len) != 1) break; if (tok[0] == '\0') break; ctx.body += len; ret = -EINVAL; p = tok; strsep(&p, "="); if (!p || (sscanf(p, "%llu", &val) != 1 && strcmp(p, "max"))) goto out_finish; ret = -ERANGE; if (!val) goto out_finish; ret = -EINVAL; |
5b7048b89
|
1597 |
if (!strcmp(tok, "rbps") && val > 1) |
2ee867dcf
|
1598 |
v[0] = val; |
5b7048b89
|
1599 |
else if (!strcmp(tok, "wbps") && val > 1) |
2ee867dcf
|
1600 |
v[1] = val; |
5b7048b89
|
1601 |
else if (!strcmp(tok, "riops") && val > 1) |
2ee867dcf
|
1602 |
v[2] = min_t(u64, val, UINT_MAX); |
5b7048b89
|
1603 |
else if (!strcmp(tok, "wiops") && val > 1) |
2ee867dcf
|
1604 |
v[3] = min_t(u64, val, UINT_MAX); |
ada75b6e5
|
1605 1606 |
else if (off == LIMIT_LOW && !strcmp(tok, "idle")) idle_time = val; |
ec80991d6
|
1607 1608 |
else if (off == LIMIT_LOW && !strcmp(tok, "latency")) latency_time = val; |
2ee867dcf
|
1609 1610 1611 |
else goto out_finish; } |
cd5ab1b0f
|
1612 1613 1614 1615 |
tg->bps_conf[READ][index] = v[0]; tg->bps_conf[WRITE][index] = v[1]; tg->iops_conf[READ][index] = v[2]; tg->iops_conf[WRITE][index] = v[3]; |
2ee867dcf
|
1616 |
|
cd5ab1b0f
|
1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 |
if (index == LIMIT_MAX) { tg->bps[READ][index] = v[0]; tg->bps[WRITE][index] = v[1]; tg->iops[READ][index] = v[2]; tg->iops[WRITE][index] = v[3]; } tg->bps[READ][LIMIT_LOW] = min(tg->bps_conf[READ][LIMIT_LOW], tg->bps_conf[READ][LIMIT_MAX]); tg->bps[WRITE][LIMIT_LOW] = min(tg->bps_conf[WRITE][LIMIT_LOW], tg->bps_conf[WRITE][LIMIT_MAX]); tg->iops[READ][LIMIT_LOW] = min(tg->iops_conf[READ][LIMIT_LOW], tg->iops_conf[READ][LIMIT_MAX]); tg->iops[WRITE][LIMIT_LOW] = min(tg->iops_conf[WRITE][LIMIT_LOW], tg->iops_conf[WRITE][LIMIT_MAX]); |
b4f428ef2
|
1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 |
tg->idletime_threshold_conf = idle_time; tg->latency_target_conf = latency_time; /* force user to configure all settings for low limit */ if (!(tg->bps[READ][LIMIT_LOW] || tg->iops[READ][LIMIT_LOW] || tg->bps[WRITE][LIMIT_LOW] || tg->iops[WRITE][LIMIT_LOW]) || tg->idletime_threshold_conf == DFL_IDLE_THRESHOLD || tg->latency_target_conf == DFL_LATENCY_TARGET) { tg->bps[READ][LIMIT_LOW] = 0; tg->bps[WRITE][LIMIT_LOW] = 0; tg->iops[READ][LIMIT_LOW] = 0; tg->iops[WRITE][LIMIT_LOW] = 0; tg->idletime_threshold = DFL_IDLE_THRESHOLD; tg->latency_target = DFL_LATENCY_TARGET; } else if (index == LIMIT_LOW) { |
5b81fc3cc
|
1646 |
tg->idletime_threshold = tg->idletime_threshold_conf; |
5b81fc3cc
|
1647 |
tg->latency_target = tg->latency_target_conf; |
cd5ab1b0f
|
1648 |
} |
b4f428ef2
|
1649 1650 1651 1652 1653 1654 1655 |
blk_throtl_update_limit_valid(tg->td); if (tg->td->limit_valid[LIMIT_LOW]) { if (index == LIMIT_LOW) tg->td->limit_index = LIMIT_LOW; } else tg->td->limit_index = LIMIT_MAX; |
9bb67aeb9
|
1656 1657 |
tg_conf_updated(tg, index == LIMIT_LOW && tg->td->limit_valid[LIMIT_LOW]); |
2ee867dcf
|
1658 1659 1660 1661 1662 1663 1664 |
ret = 0; out_finish: blkg_conf_finish(&ctx); return ret ?: nbytes; } static struct cftype throtl_files[] = { |
cd5ab1b0f
|
1665 1666 1667 1668 1669 1670 1671 1672 1673 |
#ifdef CONFIG_BLK_DEV_THROTTLING_LOW { .name = "low", .flags = CFTYPE_NOT_ON_ROOT, .seq_show = tg_print_limit, .write = tg_set_limit, .private = LIMIT_LOW, }, #endif |
2ee867dcf
|
1674 1675 1676 |
{ .name = "max", .flags = CFTYPE_NOT_ON_ROOT, |
cd5ab1b0f
|
1677 1678 1679 |
.seq_show = tg_print_limit, .write = tg_set_limit, .private = LIMIT_MAX, |
2ee867dcf
|
1680 1681 1682 |
}, { } /* terminate */ }; |
da5277700
|
1683 |
static void throtl_shutdown_wq(struct request_queue *q) |
e43473b7f
|
1684 1685 |
{ struct throtl_data *td = q->td; |
69df0ab03
|
1686 |
cancel_work_sync(&td->dispatch_work); |
e43473b7f
|
1687 |
} |
3c798398e
|
1688 |
static struct blkcg_policy blkcg_policy_throtl = { |
2ee867dcf
|
1689 |
.dfl_cftypes = throtl_files, |
880f50e22
|
1690 |
.legacy_cftypes = throtl_legacy_files, |
f9fcc2d39
|
1691 |
|
001bea73e
|
1692 |
.pd_alloc_fn = throtl_pd_alloc, |
f9fcc2d39
|
1693 |
.pd_init_fn = throtl_pd_init, |
693e751e7
|
1694 |
.pd_online_fn = throtl_pd_online, |
cd5ab1b0f
|
1695 |
.pd_offline_fn = throtl_pd_offline, |
001bea73e
|
1696 |
.pd_free_fn = throtl_pd_free, |
e43473b7f
|
1697 |
}; |
3f0abd806
|
1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 |
static unsigned long __tg_last_low_overflow_time(struct throtl_grp *tg) { unsigned long rtime = jiffies, wtime = jiffies; if (tg->bps[READ][LIMIT_LOW] || tg->iops[READ][LIMIT_LOW]) rtime = tg->last_low_overflow_time[READ]; if (tg->bps[WRITE][LIMIT_LOW] || tg->iops[WRITE][LIMIT_LOW]) wtime = tg->last_low_overflow_time[WRITE]; return min(rtime, wtime); } /* tg should not be an intermediate node */ static unsigned long tg_last_low_overflow_time(struct throtl_grp *tg) { struct throtl_service_queue *parent_sq; struct throtl_grp *parent = tg; unsigned long ret = __tg_last_low_overflow_time(tg); while (true) { parent_sq = parent->service_queue.parent_sq; parent = sq_to_tg(parent_sq); if (!parent) break; /* * The parent doesn't have low limit, it always reaches low * limit. Its overflow time is useless for children */ if (!parent->bps[READ][LIMIT_LOW] && !parent->iops[READ][LIMIT_LOW] && !parent->bps[WRITE][LIMIT_LOW] && !parent->iops[WRITE][LIMIT_LOW]) continue; if (time_after(__tg_last_low_overflow_time(parent), ret)) ret = __tg_last_low_overflow_time(parent); } return ret; } |
9e234eeaf
|
1736 1737 1738 1739 1740 |
static bool throtl_tg_is_idle(struct throtl_grp *tg) { /* * cgroup is idle if: * - single idle is too long, longer than a fixed value (in case user |
b4f428ef2
|
1741 |
* configure a too big threshold) or 4 times of idletime threshold |
9e234eeaf
|
1742 |
* - average think time is more than threshold |
53696b8d2
|
1743 |
* - IO latency is largely below threshold |
9e234eeaf
|
1744 |
*/ |
b4f428ef2
|
1745 |
unsigned long time; |
4cff729f6
|
1746 |
bool ret; |
9e234eeaf
|
1747 |
|
b4f428ef2
|
1748 1749 1750 1751 1752 1753 |
time = min_t(unsigned long, MAX_IDLE_TIME, 4 * tg->idletime_threshold); ret = tg->latency_target == DFL_LATENCY_TARGET || tg->idletime_threshold == DFL_IDLE_THRESHOLD || (ktime_get_ns() >> 10) - tg->last_finish_time > time || tg->avg_idletime > tg->idletime_threshold || (tg->latency_target && tg->bio_cnt && |
53696b8d2
|
1754 |
tg->bad_bio_cnt * 5 < tg->bio_cnt); |
4cff729f6
|
1755 1756 1757 1758 1759 |
throtl_log(&tg->service_queue, "avg_idle=%ld, idle_threshold=%ld, bad_bio=%d, total_bio=%d, is_idle=%d, scale=%d", tg->avg_idletime, tg->idletime_threshold, tg->bad_bio_cnt, tg->bio_cnt, ret, tg->td->scale); return ret; |
9e234eeaf
|
1760 |
} |
c79892c55
|
1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 |
static bool throtl_tg_can_upgrade(struct throtl_grp *tg) { struct throtl_service_queue *sq = &tg->service_queue; bool read_limit, write_limit; /* * if cgroup reaches low limit (if low limit is 0, the cgroup always * reaches), it's ok to upgrade to next limit */ read_limit = tg->bps[READ][LIMIT_LOW] || tg->iops[READ][LIMIT_LOW]; write_limit = tg->bps[WRITE][LIMIT_LOW] || tg->iops[WRITE][LIMIT_LOW]; if (!read_limit && !write_limit) return true; if (read_limit && sq->nr_queued[READ] && (!write_limit || sq->nr_queued[WRITE])) return true; if (write_limit && sq->nr_queued[WRITE] && (!read_limit || sq->nr_queued[READ])) return true; |
aec242468
|
1780 1781 |
if (time_after_eq(jiffies, |
fa6fb5aab
|
1782 1783 |
tg_last_low_overflow_time(tg) + tg->td->throtl_slice) && throtl_tg_is_idle(tg)) |
aec242468
|
1784 |
return true; |
c79892c55
|
1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 |
return false; } static bool throtl_hierarchy_can_upgrade(struct throtl_grp *tg) { while (true) { if (throtl_tg_can_upgrade(tg)) return true; tg = sq_to_tg(tg->service_queue.parent_sq); if (!tg || !tg_to_blkg(tg)->parent) return false; } return false; } static bool throtl_can_upgrade(struct throtl_data *td, struct throtl_grp *this_tg) { struct cgroup_subsys_state *pos_css; struct blkcg_gq *blkg; if (td->limit_index != LIMIT_LOW) return false; |
297e3d854
|
1808 |
if (time_before(jiffies, td->low_downgrade_time + td->throtl_slice)) |
3f0abd806
|
1809 |
return false; |
c79892c55
|
1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 |
rcu_read_lock(); blkg_for_each_descendant_post(blkg, pos_css, td->queue->root_blkg) { struct throtl_grp *tg = blkg_to_tg(blkg); if (tg == this_tg) continue; if (!list_empty(&tg_to_blkg(tg)->blkcg->css.children)) continue; if (!throtl_hierarchy_can_upgrade(tg)) { rcu_read_unlock(); return false; } } rcu_read_unlock(); return true; } |
fa6fb5aab
|
1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 |
static void throtl_upgrade_check(struct throtl_grp *tg) { unsigned long now = jiffies; if (tg->td->limit_index != LIMIT_LOW) return; if (time_after(tg->last_check_time + tg->td->throtl_slice, now)) return; tg->last_check_time = now; if (!time_after_eq(now, __tg_last_low_overflow_time(tg) + tg->td->throtl_slice)) return; if (throtl_can_upgrade(tg->td, NULL)) throtl_upgrade_state(tg->td); } |
c79892c55
|
1845 1846 1847 1848 |
static void throtl_upgrade_state(struct throtl_data *td) { struct cgroup_subsys_state *pos_css; struct blkcg_gq *blkg; |
4cff729f6
|
1849 |
throtl_log(&td->service_queue, "upgrade to max"); |
c79892c55
|
1850 |
td->limit_index = LIMIT_MAX; |
3f0abd806
|
1851 |
td->low_upgrade_time = jiffies; |
7394e31fa
|
1852 |
td->scale = 0; |
c79892c55
|
1853 1854 1855 1856 1857 1858 1859 |
rcu_read_lock(); blkg_for_each_descendant_post(blkg, pos_css, td->queue->root_blkg) { struct throtl_grp *tg = blkg_to_tg(blkg); struct throtl_service_queue *sq = &tg->service_queue; tg->disptime = jiffies - 1; throtl_select_dispatch(sq); |
4f02fb761
|
1860 |
throtl_schedule_next_dispatch(sq, true); |
c79892c55
|
1861 1862 1863 |
} rcu_read_unlock(); throtl_select_dispatch(&td->service_queue); |
4f02fb761
|
1864 |
throtl_schedule_next_dispatch(&td->service_queue, true); |
c79892c55
|
1865 1866 |
queue_work(kthrotld_workqueue, &td->dispatch_work); } |
4247d9c8b
|
1867 |
static void throtl_downgrade_state(struct throtl_data *td) |
3f0abd806
|
1868 |
{ |
7394e31fa
|
1869 |
td->scale /= 2; |
4cff729f6
|
1870 |
throtl_log(&td->service_queue, "downgrade, scale %d", td->scale); |
7394e31fa
|
1871 1872 1873 1874 |
if (td->scale) { td->low_upgrade_time = jiffies - td->scale * td->throtl_slice; return; } |
4247d9c8b
|
1875 |
td->limit_index = LIMIT_LOW; |
3f0abd806
|
1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 |
td->low_downgrade_time = jiffies; } static bool throtl_tg_can_downgrade(struct throtl_grp *tg) { struct throtl_data *td = tg->td; unsigned long now = jiffies; /* * If cgroup is below low limit, consider downgrade and throttle other * cgroups */ |
297e3d854
|
1888 1889 |
if (time_after_eq(now, td->low_upgrade_time + td->throtl_slice) && time_after_eq(now, tg_last_low_overflow_time(tg) + |
fa6fb5aab
|
1890 1891 1892 |
td->throtl_slice) && (!throtl_tg_is_idle(tg) || !list_empty(&tg_to_blkg(tg)->blkcg->css.children))) |
3f0abd806
|
1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 |
return true; return false; } static bool throtl_hierarchy_can_downgrade(struct throtl_grp *tg) { while (true) { if (!throtl_tg_can_downgrade(tg)) return false; tg = sq_to_tg(tg->service_queue.parent_sq); if (!tg || !tg_to_blkg(tg)->parent) break; } return true; } static void throtl_downgrade_check(struct throtl_grp *tg) { uint64_t bps; unsigned int iops; unsigned long elapsed_time; unsigned long now = jiffies; if (tg->td->limit_index != LIMIT_MAX || !tg->td->limit_valid[LIMIT_LOW]) return; if (!list_empty(&tg_to_blkg(tg)->blkcg->css.children)) return; |
297e3d854
|
1921 |
if (time_after(tg->last_check_time + tg->td->throtl_slice, now)) |
3f0abd806
|
1922 1923 1924 1925 |
return; elapsed_time = now - tg->last_check_time; tg->last_check_time = now; |
297e3d854
|
1926 1927 |
if (time_before(now, tg_last_low_overflow_time(tg) + tg->td->throtl_slice)) |
3f0abd806
|
1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 |
return; if (tg->bps[READ][LIMIT_LOW]) { bps = tg->last_bytes_disp[READ] * HZ; do_div(bps, elapsed_time); if (bps >= tg->bps[READ][LIMIT_LOW]) tg->last_low_overflow_time[READ] = now; } if (tg->bps[WRITE][LIMIT_LOW]) { bps = tg->last_bytes_disp[WRITE] * HZ; do_div(bps, elapsed_time); if (bps >= tg->bps[WRITE][LIMIT_LOW]) tg->last_low_overflow_time[WRITE] = now; } if (tg->iops[READ][LIMIT_LOW]) { |
4f1e9630a
|
1945 |
tg->last_io_disp[READ] += atomic_xchg(&tg->last_io_split_cnt[READ], 0); |
3f0abd806
|
1946 1947 1948 1949 1950 1951 |
iops = tg->last_io_disp[READ] * HZ / elapsed_time; if (iops >= tg->iops[READ][LIMIT_LOW]) tg->last_low_overflow_time[READ] = now; } if (tg->iops[WRITE][LIMIT_LOW]) { |
4f1e9630a
|
1952 |
tg->last_io_disp[WRITE] += atomic_xchg(&tg->last_io_split_cnt[WRITE], 0); |
3f0abd806
|
1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 |
iops = tg->last_io_disp[WRITE] * HZ / elapsed_time; if (iops >= tg->iops[WRITE][LIMIT_LOW]) tg->last_low_overflow_time[WRITE] = now; } /* * If cgroup is below low limit, consider downgrade and throttle other * cgroups */ if (throtl_hierarchy_can_downgrade(tg)) |
4247d9c8b
|
1963 |
throtl_downgrade_state(tg->td); |
3f0abd806
|
1964 1965 1966 1967 1968 1969 |
tg->last_bytes_disp[READ] = 0; tg->last_bytes_disp[WRITE] = 0; tg->last_io_disp[READ] = 0; tg->last_io_disp[WRITE] = 0; } |
9e234eeaf
|
1970 1971 |
static void blk_throtl_update_idletime(struct throtl_grp *tg) { |
7901601ae
|
1972 |
unsigned long now; |
9e234eeaf
|
1973 |
unsigned long last_finish_time = tg->last_finish_time; |
7901601ae
|
1974 1975 1976 1977 1978 |
if (last_finish_time == 0) return; now = ktime_get_ns() >> 10; if (now <= last_finish_time || |
9e234eeaf
|
1979 1980 1981 1982 1983 1984 |
last_finish_time == tg->checked_last_finish_time) return; tg->avg_idletime = (tg->avg_idletime * 7 + now - last_finish_time) >> 3; tg->checked_last_finish_time = last_finish_time; } |
b9147dd1b
|
1985 1986 1987 |
#ifdef CONFIG_BLK_DEV_THROTTLING_LOW static void throtl_update_latency_buckets(struct throtl_data *td) { |
b889bf66d
|
1988 1989 1990 1991 |
struct avg_latency_bucket avg_latency[2][LATENCY_BUCKET_SIZE]; int i, cpu, rw; unsigned long last_latency[2] = { 0 }; unsigned long latency[2]; |
b9147dd1b
|
1992 |
|
b185efa78
|
1993 |
if (!blk_queue_nonrot(td->queue) || !td->limit_valid[LIMIT_LOW]) |
b9147dd1b
|
1994 1995 1996 1997 1998 1999 |
return; if (time_before(jiffies, td->last_calculate_time + HZ)) return; td->last_calculate_time = jiffies; memset(avg_latency, 0, sizeof(avg_latency)); |
b889bf66d
|
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 |
for (rw = READ; rw <= WRITE; rw++) { for (i = 0; i < LATENCY_BUCKET_SIZE; i++) { struct latency_bucket *tmp = &td->tmp_buckets[rw][i]; for_each_possible_cpu(cpu) { struct latency_bucket *bucket; /* this isn't race free, but ok in practice */ bucket = per_cpu_ptr(td->latency_buckets[rw], cpu); tmp->total_latency += bucket[i].total_latency; tmp->samples += bucket[i].samples; bucket[i].total_latency = 0; bucket[i].samples = 0; } |
b9147dd1b
|
2015 |
|
b889bf66d
|
2016 2017 |
if (tmp->samples >= 32) { int samples = tmp->samples; |
b9147dd1b
|
2018 |
|
b889bf66d
|
2019 |
latency[rw] = tmp->total_latency; |
b9147dd1b
|
2020 |
|
b889bf66d
|
2021 2022 2023 2024 2025 2026 2027 |
tmp->total_latency = 0; tmp->samples = 0; latency[rw] /= samples; if (latency[rw] == 0) continue; avg_latency[rw][i].latency = latency[rw]; } |
b9147dd1b
|
2028 2029 |
} } |
b889bf66d
|
2030 2031 2032 2033 2034 2035 2036 2037 |
for (rw = READ; rw <= WRITE; rw++) { for (i = 0; i < LATENCY_BUCKET_SIZE; i++) { if (!avg_latency[rw][i].latency) { if (td->avg_buckets[rw][i].latency < last_latency[rw]) td->avg_buckets[rw][i].latency = last_latency[rw]; continue; } |
b9147dd1b
|
2038 |
|
b889bf66d
|
2039 2040 2041 2042 2043 |
if (!td->avg_buckets[rw][i].valid) latency[rw] = avg_latency[rw][i].latency; else latency[rw] = (td->avg_buckets[rw][i].latency * 7 + avg_latency[rw][i].latency) >> 3; |
b9147dd1b
|
2044 |
|
b889bf66d
|
2045 2046 2047 2048 2049 |
td->avg_buckets[rw][i].latency = max(latency[rw], last_latency[rw]); td->avg_buckets[rw][i].valid = true; last_latency[rw] = td->avg_buckets[rw][i].latency; } |
b9147dd1b
|
2050 |
} |
4cff729f6
|
2051 2052 2053 |
for (i = 0; i < LATENCY_BUCKET_SIZE; i++) throtl_log(&td->service_queue, |
b889bf66d
|
2054 2055 2056 2057 2058 2059 |
"Latency bucket %d: read latency=%ld, read valid=%d, " "write latency=%ld, write valid=%d", i, td->avg_buckets[READ][i].latency, td->avg_buckets[READ][i].valid, td->avg_buckets[WRITE][i].latency, td->avg_buckets[WRITE][i].valid); |
b9147dd1b
|
2060 2061 2062 2063 2064 2065 |
} #else static inline void throtl_update_latency_buckets(struct throtl_data *td) { } #endif |
4f1e9630a
|
2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 |
void blk_throtl_charge_bio_split(struct bio *bio) { struct blkcg_gq *blkg = bio->bi_blkg; struct throtl_grp *parent = blkg_to_tg(blkg); struct throtl_service_queue *parent_sq; bool rw = bio_data_dir(bio); do { if (!parent->has_rules[rw]) break; atomic_inc(&parent->io_split_cnt[rw]); atomic_inc(&parent->last_io_split_cnt[rw]); parent_sq = parent->service_queue.parent_sq; parent = sq_to_tg(parent_sq); } while (parent); } |
db18a53e5
|
2084 |
bool blk_throtl_bio(struct bio *bio) |
e43473b7f
|
2085 |
{ |
309dca309
|
2086 |
struct request_queue *q = bio->bi_bdev->bd_disk->queue; |
db18a53e5
|
2087 |
struct blkcg_gq *blkg = bio->bi_blkg; |
c5cc2070b
|
2088 |
struct throtl_qnode *qn = NULL; |
a2e83ef9c
|
2089 |
struct throtl_grp *tg = blkg_to_tg(blkg); |
73f0d49a9
|
2090 |
struct throtl_service_queue *sq; |
0e9f4164b
|
2091 |
bool rw = bio_data_dir(bio); |
bc16a4f93
|
2092 |
bool throttled = false; |
b9147dd1b
|
2093 |
struct throtl_data *td = tg->td; |
e43473b7f
|
2094 |
|
93b806380
|
2095 |
rcu_read_lock(); |
ae1188963
|
2096 |
|
2a0f61e6e
|
2097 |
/* see throtl_charge_bio() */ |
7ca464383
|
2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 |
if (bio_flagged(bio, BIO_THROTTLED)) goto out; if (!cgroup_subsys_on_dfl(io_cgrp_subsys)) { blkg_rwstat_add(&tg->stat_bytes, bio->bi_opf, bio->bi_iter.bi_size); blkg_rwstat_add(&tg->stat_ios, bio->bi_opf, 1); } if (!tg->has_rules[rw]) |
bc16a4f93
|
2108 |
goto out; |
e43473b7f
|
2109 |
|
0d945c1f9
|
2110 |
spin_lock_irq(&q->queue_lock); |
c9589f03e
|
2111 |
|
b9147dd1b
|
2112 |
throtl_update_latency_buckets(td); |
9e234eeaf
|
2113 |
blk_throtl_update_idletime(tg); |
73f0d49a9
|
2114 |
sq = &tg->service_queue; |
c79892c55
|
2115 |
again: |
9e660acff
|
2116 |
while (true) { |
3f0abd806
|
2117 2118 2119 |
if (tg->last_low_overflow_time[rw] == 0) tg->last_low_overflow_time[rw] = jiffies; throtl_downgrade_check(tg); |
fa6fb5aab
|
2120 |
throtl_upgrade_check(tg); |
9e660acff
|
2121 2122 2123 |
/* throtl is FIFO - if bios are already queued, should queue */ if (sq->nr_queued[rw]) break; |
de701c74a
|
2124 |
|
9e660acff
|
2125 |
/* if above limits, break to queue */ |
c79892c55
|
2126 |
if (!tg_may_dispatch(tg, bio, NULL)) { |
3f0abd806
|
2127 |
tg->last_low_overflow_time[rw] = jiffies; |
b9147dd1b
|
2128 2129 |
if (throtl_can_upgrade(td, tg)) { throtl_upgrade_state(td); |
c79892c55
|
2130 2131 |
goto again; } |
9e660acff
|
2132 |
break; |
c79892c55
|
2133 |
} |
9e660acff
|
2134 2135 |
/* within limits, let's charge and dispatch directly */ |
e43473b7f
|
2136 |
throtl_charge_bio(tg, bio); |
04521db04
|
2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 |
/* * We need to trim slice even when bios are not being queued * otherwise it might happen that a bio is not queued for * a long time and slice keeps on extending and trim is not * called for a long time. Now if limits are reduced suddenly * we take into account all the IO dispatched so far at new * low rate and * newly queued IO gets a really long dispatch * time. * * So keep on trimming slice even if bio is not queued. */ |
0f3457f60
|
2149 |
throtl_trim_slice(tg, rw); |
9e660acff
|
2150 2151 2152 |
/* * @bio passed through this layer without being throttled. |
b53b072c4
|
2153 |
* Climb up the ladder. If we're already at the top, it |
9e660acff
|
2154 2155 |
* can be executed directly. */ |
c5cc2070b
|
2156 |
qn = &tg->qnode_on_parent[rw]; |
9e660acff
|
2157 2158 2159 2160 |
sq = sq->parent_sq; tg = sq_to_tg(sq); if (!tg) goto out_unlock; |
e43473b7f
|
2161 |
} |
9e660acff
|
2162 |
/* out-of-limit, queue to @tg */ |
fda6f272c
|
2163 2164 |
throtl_log(sq, "[%c] bio. bdisp=%llu sz=%u bps=%llu iodisp=%u iops=%u queued=%d/%d", rw == READ ? 'R' : 'W', |
9f626e372
|
2165 2166 2167 |
tg->bytes_disp[rw], bio->bi_iter.bi_size, tg_bps_limit(tg, rw), tg->io_disp[rw], tg_iops_limit(tg, rw), |
fda6f272c
|
2168 |
sq->nr_queued[READ], sq->nr_queued[WRITE]); |
e43473b7f
|
2169 |
|
3f0abd806
|
2170 |
tg->last_low_overflow_time[rw] = jiffies; |
b9147dd1b
|
2171 |
td->nr_queued[rw]++; |
c5cc2070b
|
2172 |
throtl_add_bio_tg(bio, qn, tg); |
bc16a4f93
|
2173 |
throttled = true; |
e43473b7f
|
2174 |
|
7f52f98c2
|
2175 2176 2177 2178 2179 2180 |
/* * Update @tg's dispatch time and force schedule dispatch if @tg * was empty before @bio. The forced scheduling isn't likely to * cause undue delay as @bio is likely to be dispatched directly if * its @tg's disptime is not in the future. */ |
0e9f4164b
|
2181 |
if (tg->flags & THROTL_TG_WAS_EMPTY) { |
77216b048
|
2182 |
tg_update_disptime(tg); |
7f52f98c2
|
2183 |
throtl_schedule_next_dispatch(tg->service_queue.parent_sq, true); |
e43473b7f
|
2184 |
} |
bc16a4f93
|
2185 |
out_unlock: |
0d945c1f9
|
2186 |
spin_unlock_irq(&q->queue_lock); |
bc16a4f93
|
2187 |
out: |
111be8839
|
2188 |
bio_set_flag(bio, BIO_THROTTLED); |
b9147dd1b
|
2189 2190 2191 |
#ifdef CONFIG_BLK_DEV_THROTTLING_LOW if (throttled || !td->track_bio_latency) |
5238dcf41
|
2192 |
bio->bi_issue.value |= BIO_ISSUE_THROTL_SKIP_LATENCY; |
b9147dd1b
|
2193 |
#endif |
93b806380
|
2194 |
rcu_read_unlock(); |
bc16a4f93
|
2195 |
return throttled; |
e43473b7f
|
2196 |
} |
9e234eeaf
|
2197 |
#ifdef CONFIG_BLK_DEV_THROTTLING_LOW |
b9147dd1b
|
2198 2199 2200 2201 2202 |
static void throtl_track_latency(struct throtl_data *td, sector_t size, int op, unsigned long time) { struct latency_bucket *latency; int index; |
b889bf66d
|
2203 2204 |
if (!td || td->limit_index != LIMIT_LOW || !(op == REQ_OP_READ || op == REQ_OP_WRITE) || |
b9147dd1b
|
2205 2206 2207 2208 |
!blk_queue_nonrot(td->queue)) return; index = request_bucket_index(size); |
b889bf66d
|
2209 |
latency = get_cpu_ptr(td->latency_buckets[op]); |
b9147dd1b
|
2210 2211 |
latency[index].total_latency += time; latency[index].samples++; |
b889bf66d
|
2212 |
put_cpu_ptr(td->latency_buckets[op]); |
b9147dd1b
|
2213 2214 2215 2216 2217 2218 |
} void blk_throtl_stat_add(struct request *rq, u64 time_ns) { struct request_queue *q = rq->q; struct throtl_data *td = q->td; |
3d2443069
|
2219 2220 |
throtl_track_latency(td, blk_rq_stats_sectors(rq), req_op(rq), time_ns >> 10); |
b9147dd1b
|
2221 |
} |
9e234eeaf
|
2222 2223 |
void blk_throtl_bio_endio(struct bio *bio) { |
08e18eab0
|
2224 |
struct blkcg_gq *blkg; |
9e234eeaf
|
2225 |
struct throtl_grp *tg; |
b9147dd1b
|
2226 2227 2228 2229 |
u64 finish_time_ns; unsigned long finish_time; unsigned long start_time; unsigned long lat; |
b889bf66d
|
2230 |
int rw = bio_data_dir(bio); |
9e234eeaf
|
2231 |
|
08e18eab0
|
2232 2233 |
blkg = bio->bi_blkg; if (!blkg) |
9e234eeaf
|
2234 |
return; |
08e18eab0
|
2235 |
tg = blkg_to_tg(blkg); |
b185efa78
|
2236 2237 |
if (!tg->td->limit_valid[LIMIT_LOW]) return; |
9e234eeaf
|
2238 |
|
b9147dd1b
|
2239 2240 |
finish_time_ns = ktime_get_ns(); tg->last_finish_time = finish_time_ns >> 10; |
5238dcf41
|
2241 2242 |
start_time = bio_issue_time(&bio->bi_issue) >> 10; finish_time = __bio_issue_time(finish_time_ns) >> 10; |
08e18eab0
|
2243 |
if (!start_time || finish_time <= start_time) |
53696b8d2
|
2244 2245 2246 |
return; lat = finish_time - start_time; |
b9147dd1b
|
2247 |
/* this is only for bio based driver */ |
5238dcf41
|
2248 2249 2250 |
if (!(bio->bi_issue.value & BIO_ISSUE_THROTL_SKIP_LATENCY)) throtl_track_latency(tg->td, bio_issue_size(&bio->bi_issue), bio_op(bio), lat); |
53696b8d2
|
2251 |
|
6679a90c4
|
2252 |
if (tg->latency_target && lat >= tg->td->filtered_latency) { |
53696b8d2
|
2253 2254 |
int bucket; unsigned int threshold; |
5238dcf41
|
2255 |
bucket = request_bucket_index(bio_issue_size(&bio->bi_issue)); |
b889bf66d
|
2256 |
threshold = tg->td->avg_buckets[rw][bucket].latency + |
53696b8d2
|
2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 |
tg->latency_target; if (lat > threshold) tg->bad_bio_cnt++; /* * Not race free, could get wrong count, which means cgroups * will be throttled */ tg->bio_cnt++; } if (time_after(jiffies, tg->bio_cnt_reset_time) || tg->bio_cnt > 1024) { tg->bio_cnt_reset_time = tg->td->throtl_slice + jiffies; tg->bio_cnt /= 2; tg->bad_bio_cnt /= 2; |
b9147dd1b
|
2271 |
} |
9e234eeaf
|
2272 2273 |
} #endif |
e43473b7f
|
2274 2275 2276 |
int blk_throtl_init(struct request_queue *q) { struct throtl_data *td; |
a2b1693ba
|
2277 |
int ret; |
e43473b7f
|
2278 2279 2280 2281 |
td = kzalloc_node(sizeof(*td), GFP_KERNEL, q->node); if (!td) return -ENOMEM; |
b889bf66d
|
2282 |
td->latency_buckets[READ] = __alloc_percpu(sizeof(struct latency_bucket) * |
b9147dd1b
|
2283 |
LATENCY_BUCKET_SIZE, __alignof__(u64)); |
b889bf66d
|
2284 2285 2286 2287 2288 |
if (!td->latency_buckets[READ]) { kfree(td); return -ENOMEM; } td->latency_buckets[WRITE] = __alloc_percpu(sizeof(struct latency_bucket) * |
b9147dd1b
|
2289 |
LATENCY_BUCKET_SIZE, __alignof__(u64)); |
b889bf66d
|
2290 2291 |
if (!td->latency_buckets[WRITE]) { free_percpu(td->latency_buckets[READ]); |
b9147dd1b
|
2292 2293 2294 |
kfree(td); return -ENOMEM; } |
e43473b7f
|
2295 |
|
69df0ab03
|
2296 |
INIT_WORK(&td->dispatch_work, blk_throtl_dispatch_work_fn); |
b2ce2643c
|
2297 |
throtl_service_queue_init(&td->service_queue); |
e43473b7f
|
2298 |
|
cd1604fab
|
2299 |
q->td = td; |
29b125892
|
2300 |
td->queue = q; |
02977e4af
|
2301 |
|
9f626e372
|
2302 |
td->limit_valid[LIMIT_MAX] = true; |
cd5ab1b0f
|
2303 |
td->limit_index = LIMIT_MAX; |
3f0abd806
|
2304 2305 |
td->low_upgrade_time = jiffies; td->low_downgrade_time = jiffies; |
9e234eeaf
|
2306 |
|
a2b1693ba
|
2307 |
/* activate policy */ |
3c798398e
|
2308 |
ret = blkcg_activate_policy(q, &blkcg_policy_throtl); |
b9147dd1b
|
2309 |
if (ret) { |
b889bf66d
|
2310 2311 |
free_percpu(td->latency_buckets[READ]); free_percpu(td->latency_buckets[WRITE]); |
f51b802c1
|
2312 |
kfree(td); |
b9147dd1b
|
2313 |
} |
a2b1693ba
|
2314 |
return ret; |
e43473b7f
|
2315 2316 2317 2318 |
} void blk_throtl_exit(struct request_queue *q) { |
c875f4d02
|
2319 |
BUG_ON(!q->td); |
884f0e84f
|
2320 |
del_timer_sync(&q->td->service_queue.pending_timer); |
da5277700
|
2321 |
throtl_shutdown_wq(q); |
3c798398e
|
2322 |
blkcg_deactivate_policy(q, &blkcg_policy_throtl); |
b889bf66d
|
2323 2324 |
free_percpu(q->td->latency_buckets[READ]); free_percpu(q->td->latency_buckets[WRITE]); |
c9a929dde
|
2325 |
kfree(q->td); |
e43473b7f
|
2326 |
} |
d61fcfa4b
|
2327 2328 2329 |
void blk_throtl_register_queue(struct request_queue *q) { struct throtl_data *td; |
6679a90c4
|
2330 |
int i; |
d61fcfa4b
|
2331 2332 2333 |
td = q->td; BUG_ON(!td); |
6679a90c4
|
2334 |
if (blk_queue_nonrot(q)) { |
d61fcfa4b
|
2335 |
td->throtl_slice = DFL_THROTL_SLICE_SSD; |
6679a90c4
|
2336 2337 |
td->filtered_latency = LATENCY_FILTERED_SSD; } else { |
d61fcfa4b
|
2338 |
td->throtl_slice = DFL_THROTL_SLICE_HD; |
6679a90c4
|
2339 |
td->filtered_latency = LATENCY_FILTERED_HD; |
b889bf66d
|
2340 2341 2342 2343 |
for (i = 0; i < LATENCY_BUCKET_SIZE; i++) { td->avg_buckets[READ][i].latency = DFL_HD_BASELINE_LATENCY; td->avg_buckets[WRITE][i].latency = DFL_HD_BASELINE_LATENCY; } |
6679a90c4
|
2344 |
} |
d61fcfa4b
|
2345 2346 2347 2348 |
#ifndef CONFIG_BLK_DEV_THROTTLING_LOW /* if no low limit, use previous default */ td->throtl_slice = DFL_THROTL_SLICE_HD; #endif |
9e234eeaf
|
2349 |
|
344e9ffcb
|
2350 |
td->track_bio_latency = !queue_is_mq(q); |
b9147dd1b
|
2351 2352 |
if (!td->track_bio_latency) blk_stat_enable_accounting(q); |
d61fcfa4b
|
2353 |
} |
297e3d854
|
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 |
#ifdef CONFIG_BLK_DEV_THROTTLING_LOW ssize_t blk_throtl_sample_time_show(struct request_queue *q, char *page) { if (!q->td) return -EINVAL; return sprintf(page, "%u ", jiffies_to_msecs(q->td->throtl_slice)); } ssize_t blk_throtl_sample_time_store(struct request_queue *q, const char *page, size_t count) { unsigned long v; unsigned long t; if (!q->td) return -EINVAL; if (kstrtoul(page, 10, &v)) return -EINVAL; t = msecs_to_jiffies(v); if (t == 0 || t > MAX_THROTL_SLICE) return -EINVAL; q->td->throtl_slice = t; return count; } #endif |
e43473b7f
|
2380 2381 |
static int __init throtl_init(void) { |
450adcbe5
|
2382 2383 2384 2385 |
kthrotld_workqueue = alloc_workqueue("kthrotld", WQ_MEM_RECLAIM, 0); if (!kthrotld_workqueue) panic("Failed to create kthrotld "); |
3c798398e
|
2386 |
return blkcg_policy_register(&blkcg_policy_throtl); |
e43473b7f
|
2387 2388 2389 |
} module_init(throtl_init); |