Blame view
block/blk-throttle.c
67.9 KB
b24413180 License cleanup: ... |
1 |
// SPDX-License-Identifier: GPL-2.0 |
e43473b7f blkio: Core imple... |
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 blkcg: move block... |
13 |
#include <linux/blk-cgroup.h> |
bc9fcbf9c block: move blk_t... |
14 |
#include "blk.h" |
e43473b7f blkio: Core imple... |
15 16 17 18 19 20 |
/* Max dispatch from a group in 1 round */ static int throtl_grp_quantum = 8; /* Total max dispatch from all groups in one round */ static int throtl_quantum = 32; |
d61fcfa4b blk-throttle: cho... |
21 22 23 |
/* 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 blk-throttle: mak... |
24 |
#define MAX_THROTL_SLICE (HZ) |
9e234eeaf blk-throttle: add... |
25 |
#define MAX_IDLE_TIME (5L * 1000 * 1000) /* 5 s */ |
9bb67aeb9 blk-throttle: res... |
26 27 |
#define MIN_THROTL_BPS (320 * 1024) #define MIN_THROTL_IOPS (10) |
b4f428ef2 blk-throttle: for... |
28 29 |
#define DFL_LATENCY_TARGET (-1L) #define DFL_IDLE_THRESHOLD (0) |
6679a90c4 blk-throttle: set... |
30 31 32 33 34 35 36 |
#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 blkio: Core imple... |
37 |
|
3c798398e blkcg: mass renam... |
38 |
static struct blkcg_policy blkcg_policy_throtl; |
0381411e4 blkcg: let blkcg ... |
39 |
|
450adcbe5 blk-throttle: Do ... |
40 41 |
/* A workqueue to queue throttle related work */ static struct workqueue_struct *kthrotld_workqueue; |
450adcbe5 blk-throttle: Do ... |
42 |
|
c5cc2070b blk-throttle: add... |
43 44 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 |
/* * 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 blk-throttle: ren... |
71 |
struct throtl_service_queue { |
77216b048 blk-throttle: add... |
72 |
struct throtl_service_queue *parent_sq; /* the parent service_queue */ |
73f0d49a9 blk-throttle: mov... |
73 74 75 76 |
/* * Bios queued directly to this service_queue or dispatched from * children throtl_grp's. */ |
c5cc2070b blk-throttle: add... |
77 |
struct list_head queued[2]; /* throtl_qnode [READ/WRITE] */ |
73f0d49a9 blk-throttle: mov... |
78 79 80 81 82 83 |
unsigned int nr_queued[2]; /* number of queued bios */ /* * RB tree of active children throtl_grp's, which are sorted by * their ->disptime. */ |
9ff01255a Blk-throttle: upd... |
84 |
struct rb_root_cached pending_tree; /* RB tree of active tgs */ |
c9e0332e8 blk-throttle: ren... |
85 86 |
unsigned int nr_pending; /* # queued in the tree */ unsigned long first_pending_disptime; /* disptime of the first tg */ |
69df0ab03 blk-throttle: sep... |
87 |
struct timer_list pending_timer; /* fires on first_pending_disptime */ |
e43473b7f blkio: Core imple... |
88 |
}; |
5b2c16aae blk-throttle: sim... |
89 90 |
enum tg_state_flags { THROTL_TG_PENDING = 1 << 0, /* on parent's pending tree */ |
0e9f4164b blk-throttle: gen... |
91 |
THROTL_TG_WAS_EMPTY = 1 << 1, /* bio_lists[] became non-empty */ |
5b2c16aae blk-throttle: sim... |
92 |
}; |
e43473b7f blkio: Core imple... |
93 |
#define rb_entry_tg(node) rb_entry((node), struct throtl_grp, rb_node) |
9f626e372 blk-throttle: pre... |
94 |
enum { |
cd5ab1b0f blk-throttle: add... |
95 |
LIMIT_LOW, |
9f626e372 blk-throttle: pre... |
96 97 98 |
LIMIT_MAX, LIMIT_CNT, }; |
e43473b7f blkio: Core imple... |
99 |
struct throtl_grp { |
f95a04afa blkcg: embed stru... |
100 101 |
/* must be the first member */ struct blkg_policy_data pd; |
c9e0332e8 blk-throttle: ren... |
102 |
/* active throtl group service_queue member */ |
e43473b7f blkio: Core imple... |
103 |
struct rb_node rb_node; |
0f3457f60 blk-throttle: add... |
104 105 |
/* throtl_data this group belongs to */ struct throtl_data *td; |
49a2f1e3f blk-throttle: add... |
106 107 |
/* this group's service queue */ struct throtl_service_queue service_queue; |
e43473b7f blkio: Core imple... |
108 |
/* |
c5cc2070b blk-throttle: add... |
109 110 111 112 113 114 115 116 117 118 119 |
* 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 blkio: Core imple... |
120 121 122 123 124 |
* 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 blkio: Core imple... |
125 |
unsigned int flags; |
693e751e7 blk-throttle: imp... |
126 127 |
/* are there any throtl rules between this group and td? */ bool has_rules[2]; |
cd5ab1b0f blk-throttle: add... |
128 |
/* internally used bytes per second rate limits */ |
9f626e372 blk-throttle: pre... |
129 |
uint64_t bps[2][LIMIT_CNT]; |
cd5ab1b0f blk-throttle: add... |
130 131 |
/* user configured bps limits */ uint64_t bps_conf[2][LIMIT_CNT]; |
e43473b7f blkio: Core imple... |
132 |
|
cd5ab1b0f blk-throttle: add... |
133 |
/* internally used IOPS limits */ |
9f626e372 blk-throttle: pre... |
134 |
unsigned int iops[2][LIMIT_CNT]; |
cd5ab1b0f blk-throttle: add... |
135 136 |
/* user configured IOPS limits */ unsigned int iops_conf[2][LIMIT_CNT]; |
8e89d13f4 blkio: Implementa... |
137 |
|
e43473b7f blkio: Core imple... |
138 139 |
/* Number of bytes disptached in current slice */ uint64_t bytes_disp[2]; |
8e89d13f4 blkio: Implementa... |
140 141 |
/* Number of bio's dispatched in current slice */ unsigned int io_disp[2]; |
e43473b7f blkio: Core imple... |
142 |
|
3f0abd806 blk-throttle: add... |
143 144 145 146 147 148 |
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 blk-throttle: add... |
149 |
unsigned long latency_target; /* us */ |
5b81fc3cc blk-throttle: add... |
150 |
unsigned long latency_target_conf; /* us */ |
e43473b7f blkio: Core imple... |
151 152 153 |
/* When did we start a new slice */ unsigned long slice_start[2]; unsigned long slice_end[2]; |
9e234eeaf blk-throttle: add... |
154 155 156 157 158 |
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 blk-throttle: add... |
159 |
unsigned long idletime_threshold_conf; /* us */ |
53696b8d2 blk-throttle: add... |
160 161 162 163 |
unsigned int bio_cnt; /* total bios */ unsigned int bad_bio_cnt; /* bios exceeding latency threshold */ unsigned long bio_cnt_reset_time; |
e43473b7f blkio: Core imple... |
164 |
}; |
b9147dd1b blk-throttle: add... |
165 166 167 168 169 170 171 172 173 174 175 176 |
/* 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 blkio: Core imple... |
177 178 |
struct throtl_data { |
e43473b7f blkio: Core imple... |
179 |
/* service tree for active throtl groups */ |
c9e0332e8 blk-throttle: ren... |
180 |
struct throtl_service_queue service_queue; |
e43473b7f blkio: Core imple... |
181 |
|
e43473b7f blkio: Core imple... |
182 183 184 185 |
struct request_queue *queue; /* Total Number of queued bios on READ and WRITE lists */ unsigned int nr_queued[2]; |
297e3d854 blk-throttle: mak... |
186 |
unsigned int throtl_slice; |
e43473b7f blkio: Core imple... |
187 |
/* Work for dispatching throttled bios */ |
69df0ab03 blk-throttle: sep... |
188 |
struct work_struct dispatch_work; |
9f626e372 blk-throttle: pre... |
189 190 |
unsigned int limit_index; bool limit_valid[LIMIT_CNT]; |
3f0abd806 blk-throttle: add... |
191 192 193 |
unsigned long low_upgrade_time; unsigned long low_downgrade_time; |
7394e31fa blk-throttle: mak... |
194 195 |
unsigned int scale; |
b9147dd1b blk-throttle: add... |
196 |
|
b889bf66d blk-throttle: tra... |
197 198 199 |
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 blk-throttle: add... |
200 |
unsigned long last_calculate_time; |
6679a90c4 blk-throttle: set... |
201 |
unsigned long filtered_latency; |
b9147dd1b blk-throttle: add... |
202 203 |
bool track_bio_latency; |
e43473b7f blkio: Core imple... |
204 |
}; |
e99e88a9d treewide: setup_t... |
205 |
static void throtl_pending_timer_fn(struct timer_list *t); |
69df0ab03 blk-throttle: sep... |
206 |
|
f95a04afa blkcg: embed stru... |
207 208 209 210 |
static inline struct throtl_grp *pd_to_tg(struct blkg_policy_data *pd) { return pd ? container_of(pd, struct throtl_grp, pd) : NULL; } |
3c798398e blkcg: mass renam... |
211 |
static inline struct throtl_grp *blkg_to_tg(struct blkcg_gq *blkg) |
0381411e4 blkcg: let blkcg ... |
212 |
{ |
f95a04afa blkcg: embed stru... |
213 |
return pd_to_tg(blkg_to_pd(blkg, &blkcg_policy_throtl)); |
0381411e4 blkcg: let blkcg ... |
214 |
} |
3c798398e blkcg: mass renam... |
215 |
static inline struct blkcg_gq *tg_to_blkg(struct throtl_grp *tg) |
0381411e4 blkcg: let blkcg ... |
216 |
{ |
f95a04afa blkcg: embed stru... |
217 |
return pd_to_blkg(&tg->pd); |
0381411e4 blkcg: let blkcg ... |
218 |
} |
fda6f272c blk-throttle: imp... |
219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 |
/** * 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 scripts/spelling.... |
238 |
* A service_queue can be embedded in either a throtl_grp or throtl_data. |
fda6f272c blk-throttle: imp... |
239 240 241 242 243 244 245 246 247 248 249 |
* 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 blk-throttle: mak... |
250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 |
/* * 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 blk-throttle: pre... |
267 268 |
static uint64_t tg_bps_limit(struct throtl_grp *tg, int rw) { |
b22c417c8 blk-throttle: con... |
269 |
struct blkcg_gq *blkg = tg_to_blkg(tg); |
7394e31fa blk-throttle: mak... |
270 |
struct throtl_data *td; |
b22c417c8 blk-throttle: con... |
271 272 273 274 |
uint64_t ret; if (cgroup_subsys_on_dfl(io_cgrp_subsys) && !blkg->parent) return U64_MAX; |
7394e31fa blk-throttle: mak... |
275 276 277 |
td = tg->td; ret = tg->bps[rw][td->limit_index]; |
9bb67aeb9 blk-throttle: res... |
278 279 280 281 282 283 284 285 |
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 blk-throttle: mak... |
286 287 288 289 290 291 292 293 |
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 blk-throttle: con... |
294 |
return ret; |
9f626e372 blk-throttle: pre... |
295 296 297 298 |
} static unsigned int tg_iops_limit(struct throtl_grp *tg, int rw) { |
b22c417c8 blk-throttle: con... |
299 |
struct blkcg_gq *blkg = tg_to_blkg(tg); |
7394e31fa blk-throttle: mak... |
300 |
struct throtl_data *td; |
b22c417c8 blk-throttle: con... |
301 302 303 304 |
unsigned int ret; if (cgroup_subsys_on_dfl(io_cgrp_subsys) && !blkg->parent) return UINT_MAX; |
9bb67aeb9 blk-throttle: res... |
305 |
|
7394e31fa blk-throttle: mak... |
306 307 |
td = tg->td; ret = tg->iops[rw][td->limit_index]; |
9bb67aeb9 blk-throttle: res... |
308 309 310 311 312 313 314 315 |
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 blk-throttle: mak... |
316 317 318 319 320 321 322 323 324 325 |
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 blk-throttle: con... |
326 |
return ret; |
9f626e372 blk-throttle: pre... |
327 |
} |
b9147dd1b blk-throttle: add... |
328 329 |
#define request_bucket_index(sectors) \ clamp_t(int, order_base_2(sectors) - 3, 0, LATENCY_BUCKET_SIZE - 1) |
fda6f272c blk-throttle: imp... |
330 331 332 333 334 335 336 337 |
/** * 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 blk-throttle: imp... |
338 339 340 341 342 343 |
*/ #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 blk-throttle: don... |
344 345 |
if (likely(!blk_trace_note_message_enabled(__td->queue))) \ break; \ |
fda6f272c blk-throttle: imp... |
346 |
if ((__tg)) { \ |
35fe6d763 block: use standa... |
347 348 |
blk_add_cgroup_trace_msg(__td->queue, \ tg_to_blkg(__tg)->blkcg, "throtl " fmt, ##args);\ |
fda6f272c blk-throttle: imp... |
349 350 351 |
} else { \ blk_add_trace_msg(__td->queue, "throtl " fmt, ##args); \ } \ |
54e7ed12b blkcg: remove blk... |
352 |
} while (0) |
e43473b7f blkio: Core imple... |
353 |
|
ea0ea2bc6 blk-throttle: cap... |
354 355 356 357 358 359 360 |
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 blk-throttle: add... |
361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 |
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) { struct throtl_qnode *qn = list_first_entry(queued, struct throtl_qnode, node); struct bio *bio; if (list_empty(queued)) return NULL; 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) { struct throtl_qnode *qn = list_first_entry(queued, struct throtl_qnode, node); struct bio *bio; if (list_empty(queued)) return NULL; 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 blk-throttle: add... |
443 |
/* init a service_queue, assumes the caller zeroed it */ |
b2ce2643c blk-throttle: cle... |
444 |
static void throtl_service_queue_init(struct throtl_service_queue *sq) |
49a2f1e3f blk-throttle: add... |
445 |
{ |
c5cc2070b blk-throttle: add... |
446 447 |
INIT_LIST_HEAD(&sq->queued[0]); INIT_LIST_HEAD(&sq->queued[1]); |
9ff01255a Blk-throttle: upd... |
448 |
sq->pending_tree = RB_ROOT_CACHED; |
e99e88a9d treewide: setup_t... |
449 |
timer_setup(&sq->pending_timer, throtl_pending_timer_fn, 0); |
69df0ab03 blk-throttle: sep... |
450 |
} |
cf09a8ee1 blkcg: pass @q an... |
451 452 453 |
static struct blkg_policy_data *throtl_pd_alloc(gfp_t gfp, struct request_queue *q, struct blkcg *blkcg) |
001bea73e blkcg: replace bl... |
454 |
{ |
4fb72036f blk-throttle: rem... |
455 |
struct throtl_grp *tg; |
24bdb8ef0 blkcg: make blkcg... |
456 |
int rw; |
4fb72036f blk-throttle: rem... |
457 |
|
cf09a8ee1 blkcg: pass @q an... |
458 |
tg = kzalloc_node(sizeof(*tg), gfp, q->node); |
4fb72036f blk-throttle: rem... |
459 |
if (!tg) |
77ea73388 blkcg: move io_se... |
460 |
return NULL; |
4fb72036f blk-throttle: rem... |
461 |
|
b2ce2643c blk-throttle: cle... |
462 463 464 465 466 467 468 469 |
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 blk-throttle: pre... |
470 471 472 473 |
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 blk-throttle: add... |
474 475 476 477 478 |
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 blk-throttle: cle... |
479 |
|
ec80991d6 blk-throttle: add... |
480 |
tg->latency_target = DFL_LATENCY_TARGET; |
5b81fc3cc blk-throttle: add... |
481 |
tg->latency_target_conf = DFL_LATENCY_TARGET; |
b4f428ef2 blk-throttle: for... |
482 483 |
tg->idletime_threshold = DFL_IDLE_THRESHOLD; tg->idletime_threshold_conf = DFL_IDLE_THRESHOLD; |
ec80991d6 blk-throttle: add... |
484 |
|
4fb72036f blk-throttle: rem... |
485 |
return &tg->pd; |
001bea73e blkcg: replace bl... |
486 |
} |
a9520cd6f blkcg: make blkcg... |
487 |
static void throtl_pd_init(struct blkg_policy_data *pd) |
a29a171e7 blk-throttle: Do ... |
488 |
{ |
a9520cd6f blkcg: make blkcg... |
489 490 |
struct throtl_grp *tg = pd_to_tg(pd); struct blkcg_gq *blkg = tg_to_blkg(tg); |
77216b048 blk-throttle: add... |
491 |
struct throtl_data *td = blkg->q->td; |
b2ce2643c blk-throttle: cle... |
492 |
struct throtl_service_queue *sq = &tg->service_queue; |
cd1604fab blkcg: factor out... |
493 |
|
9138125be blk-throttle: imp... |
494 |
/* |
aa6ec29be cgroup: remove sa... |
495 |
* If on the default hierarchy, we switch to properly hierarchical |
9138125be blk-throttle: imp... |
496 497 498 499 500 |
* 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 cgroup: remove sa... |
501 |
* If not on the default hierarchy, the broken flat hierarchy |
9138125be blk-throttle: imp... |
502 503 504 505 506 |
* 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 blk-throttle: cle... |
507 |
sq->parent_sq = &td->service_queue; |
9e10a130d cgroup: replace c... |
508 |
if (cgroup_subsys_on_dfl(io_cgrp_subsys) && blkg->parent) |
b2ce2643c blk-throttle: cle... |
509 |
sq->parent_sq = &blkg_to_tg(blkg->parent)->service_queue; |
77216b048 blk-throttle: add... |
510 |
tg->td = td; |
8a3d26151 blkcg: move blkio... |
511 |
} |
693e751e7 blk-throttle: imp... |
512 513 514 515 516 517 518 519 |
/* * 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 blk-throttle: pre... |
520 |
struct throtl_data *td = tg->td; |
693e751e7 blk-throttle: imp... |
521 522 523 524 |
int rw; for (rw = READ; rw <= WRITE; rw++) tg->has_rules[rw] = (parent_tg && parent_tg->has_rules[rw]) || |
9f626e372 blk-throttle: pre... |
525 526 527 |
(td->limit_valid[td->limit_index] && (tg_bps_limit(tg, rw) != U64_MAX || tg_iops_limit(tg, rw) != UINT_MAX)); |
693e751e7 blk-throttle: imp... |
528 |
} |
a9520cd6f blkcg: make blkcg... |
529 |
static void throtl_pd_online(struct blkg_policy_data *pd) |
693e751e7 blk-throttle: imp... |
530 |
{ |
aec242468 blk-throttle: det... |
531 |
struct throtl_grp *tg = pd_to_tg(pd); |
693e751e7 blk-throttle: imp... |
532 533 534 535 |
/* * We don't want new groups to escape the limits of its ancestors. * Update has_rules[] after a new group is brought online. */ |
aec242468 blk-throttle: det... |
536 |
tg_update_has_rules(tg); |
693e751e7 blk-throttle: imp... |
537 |
} |
cd5ab1b0f blk-throttle: add... |
538 539 540 541 542 543 544 545 546 547 548 |
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 Block: blk-thrott... |
549 |
tg->iops[READ][LIMIT_LOW] || tg->iops[WRITE][LIMIT_LOW]) { |
cd5ab1b0f blk-throttle: add... |
550 |
low_valid = true; |
43ada7878 Block: blk-thrott... |
551 552 |
break; } |
cd5ab1b0f blk-throttle: add... |
553 554 555 556 557 |
} rcu_read_unlock(); td->limit_valid[LIMIT_LOW] = low_valid; } |
c79892c55 blk-throttle: add... |
558 |
static void throtl_upgrade_state(struct throtl_data *td); |
cd5ab1b0f blk-throttle: add... |
559 560 561 562 563 564 565 566 567 568 |
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 blk-throttle: add... |
569 570 |
if (!tg->td->limit_valid[tg->td->limit_index]) throtl_upgrade_state(tg->td); |
cd5ab1b0f blk-throttle: add... |
571 |
} |
001bea73e blkcg: replace bl... |
572 573 |
static void throtl_pd_free(struct blkg_policy_data *pd) { |
4fb72036f blk-throttle: rem... |
574 |
struct throtl_grp *tg = pd_to_tg(pd); |
b2ce2643c blk-throttle: cle... |
575 |
del_timer_sync(&tg->service_queue.pending_timer); |
4fb72036f blk-throttle: rem... |
576 |
kfree(tg); |
001bea73e blkcg: replace bl... |
577 |
} |
0049af73b blk-throttle: reo... |
578 579 |
static struct throtl_grp * throtl_rb_first(struct throtl_service_queue *parent_sq) |
e43473b7f blkio: Core imple... |
580 |
{ |
9ff01255a Blk-throttle: upd... |
581 |
struct rb_node *n; |
e43473b7f blkio: Core imple... |
582 |
/* Service tree is empty */ |
0049af73b blk-throttle: reo... |
583 |
if (!parent_sq->nr_pending) |
e43473b7f blkio: Core imple... |
584 |
return NULL; |
9ff01255a Blk-throttle: upd... |
585 586 587 588 589 |
n = rb_first_cached(&parent_sq->pending_tree); WARN_ON_ONCE(!n); if (!n) return NULL; return rb_entry_tg(n); |
e43473b7f blkio: Core imple... |
590 |
} |
0049af73b blk-throttle: reo... |
591 592 |
static void throtl_rb_erase(struct rb_node *n, struct throtl_service_queue *parent_sq) |
e43473b7f blkio: Core imple... |
593 |
{ |
9ff01255a Blk-throttle: upd... |
594 595 |
rb_erase_cached(n, &parent_sq->pending_tree); RB_CLEAR_NODE(n); |
0049af73b blk-throttle: reo... |
596 |
--parent_sq->nr_pending; |
e43473b7f blkio: Core imple... |
597 |
} |
0049af73b blk-throttle: reo... |
598 |
static void update_min_dispatch_time(struct throtl_service_queue *parent_sq) |
e43473b7f blkio: Core imple... |
599 600 |
{ struct throtl_grp *tg; |
0049af73b blk-throttle: reo... |
601 |
tg = throtl_rb_first(parent_sq); |
e43473b7f blkio: Core imple... |
602 603 |
if (!tg) return; |
0049af73b blk-throttle: reo... |
604 |
parent_sq->first_pending_disptime = tg->disptime; |
e43473b7f blkio: Core imple... |
605 |
} |
77216b048 blk-throttle: add... |
606 |
static void tg_service_queue_add(struct throtl_grp *tg) |
e43473b7f blkio: Core imple... |
607 |
{ |
77216b048 blk-throttle: add... |
608 |
struct throtl_service_queue *parent_sq = tg->service_queue.parent_sq; |
9ff01255a Blk-throttle: upd... |
609 |
struct rb_node **node = &parent_sq->pending_tree.rb_root.rb_node; |
e43473b7f blkio: Core imple... |
610 611 612 |
struct rb_node *parent = NULL; struct throtl_grp *__tg; unsigned long key = tg->disptime; |
9ff01255a Blk-throttle: upd... |
613 |
bool leftmost = true; |
e43473b7f blkio: Core imple... |
614 615 616 617 618 619 620 621 622 |
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 Blk-throttle: upd... |
623 |
leftmost = false; |
e43473b7f blkio: Core imple... |
624 625 |
} } |
e43473b7f blkio: Core imple... |
626 |
rb_link_node(&tg->rb_node, parent, node); |
9ff01255a Blk-throttle: upd... |
627 628 |
rb_insert_color_cached(&tg->rb_node, &parent_sq->pending_tree, leftmost); |
e43473b7f blkio: Core imple... |
629 |
} |
77216b048 blk-throttle: add... |
630 |
static void __throtl_enqueue_tg(struct throtl_grp *tg) |
e43473b7f blkio: Core imple... |
631 |
{ |
77216b048 blk-throttle: add... |
632 |
tg_service_queue_add(tg); |
5b2c16aae blk-throttle: sim... |
633 |
tg->flags |= THROTL_TG_PENDING; |
77216b048 blk-throttle: add... |
634 |
tg->service_queue.parent_sq->nr_pending++; |
e43473b7f blkio: Core imple... |
635 |
} |
77216b048 blk-throttle: add... |
636 |
static void throtl_enqueue_tg(struct throtl_grp *tg) |
e43473b7f blkio: Core imple... |
637 |
{ |
5b2c16aae blk-throttle: sim... |
638 |
if (!(tg->flags & THROTL_TG_PENDING)) |
77216b048 blk-throttle: add... |
639 |
__throtl_enqueue_tg(tg); |
e43473b7f blkio: Core imple... |
640 |
} |
77216b048 blk-throttle: add... |
641 |
static void __throtl_dequeue_tg(struct throtl_grp *tg) |
e43473b7f blkio: Core imple... |
642 |
{ |
77216b048 blk-throttle: add... |
643 |
throtl_rb_erase(&tg->rb_node, tg->service_queue.parent_sq); |
5b2c16aae blk-throttle: sim... |
644 |
tg->flags &= ~THROTL_TG_PENDING; |
e43473b7f blkio: Core imple... |
645 |
} |
77216b048 blk-throttle: add... |
646 |
static void throtl_dequeue_tg(struct throtl_grp *tg) |
e43473b7f blkio: Core imple... |
647 |
{ |
5b2c16aae blk-throttle: sim... |
648 |
if (tg->flags & THROTL_TG_PENDING) |
77216b048 blk-throttle: add... |
649 |
__throtl_dequeue_tg(tg); |
e43473b7f blkio: Core imple... |
650 |
} |
a9131a27e blk-throttle: rel... |
651 |
/* Call with queue lock held */ |
69df0ab03 blk-throttle: sep... |
652 653 |
static void throtl_schedule_pending_timer(struct throtl_service_queue *sq, unsigned long expires) |
a9131a27e blk-throttle: rel... |
654 |
{ |
a41b816c1 blk-throttle: fix... |
655 |
unsigned long max_expire = jiffies + 8 * sq_to_td(sq)->throtl_slice; |
06cceedcc blk-throttle: mak... |
656 657 658 659 660 661 662 663 664 665 |
/* * 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 blk-throttle: sep... |
666 667 668 |
mod_timer(&sq->pending_timer, expires); throtl_log(sq, "schedule timer. delay=%lu jiffies=%lu", expires - jiffies, jiffies); |
a9131a27e blk-throttle: rel... |
669 |
} |
7f52f98c2 blk-throttle: imp... |
670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 |
/** * 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 blkio: Core imple... |
690 |
{ |
6a525600f blk-throttle: rem... |
691 |
/* any pending children left? */ |
c9e0332e8 blk-throttle: ren... |
692 |
if (!sq->nr_pending) |
7f52f98c2 blk-throttle: imp... |
693 |
return true; |
e43473b7f blkio: Core imple... |
694 |
|
c9e0332e8 blk-throttle: ren... |
695 |
update_min_dispatch_time(sq); |
e43473b7f blkio: Core imple... |
696 |
|
69df0ab03 blk-throttle: sep... |
697 |
/* is the next dispatch time in the future? */ |
7f52f98c2 blk-throttle: imp... |
698 |
if (force || time_after(sq->first_pending_disptime, jiffies)) { |
69df0ab03 blk-throttle: sep... |
699 |
throtl_schedule_pending_timer(sq, sq->first_pending_disptime); |
7f52f98c2 blk-throttle: imp... |
700 |
return true; |
69df0ab03 blk-throttle: sep... |
701 |
} |
7f52f98c2 blk-throttle: imp... |
702 703 |
/* tell the caller to continue dispatching */ return false; |
e43473b7f blkio: Core imple... |
704 |
} |
32ee5bc47 blk-throttle: Acc... |
705 706 707 708 709 710 711 712 713 714 715 716 717 718 |
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; /* * 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 blk-throttle: mak... |
719 |
tg->slice_end[rw] = jiffies + tg->td->throtl_slice; |
32ee5bc47 blk-throttle: Acc... |
720 721 722 723 724 |
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 blk-throttle: add... |
725 |
static inline void throtl_start_new_slice(struct throtl_grp *tg, bool rw) |
e43473b7f blkio: Core imple... |
726 727 |
{ tg->bytes_disp[rw] = 0; |
8e89d13f4 blkio: Implementa... |
728 |
tg->io_disp[rw] = 0; |
e43473b7f blkio: Core imple... |
729 |
tg->slice_start[rw] = jiffies; |
297e3d854 blk-throttle: mak... |
730 |
tg->slice_end[rw] = jiffies + tg->td->throtl_slice; |
fda6f272c blk-throttle: imp... |
731 732 733 734 |
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 blkio: Core imple... |
735 |
} |
0f3457f60 blk-throttle: add... |
736 737 |
static inline void throtl_set_slice_end(struct throtl_grp *tg, bool rw, unsigned long jiffy_end) |
d1ae8ffdf blk-throttle: Tri... |
738 |
{ |
297e3d854 blk-throttle: mak... |
739 |
tg->slice_end[rw] = roundup(jiffy_end, tg->td->throtl_slice); |
d1ae8ffdf blk-throttle: Tri... |
740 |
} |
0f3457f60 blk-throttle: add... |
741 742 |
static inline void throtl_extend_slice(struct throtl_grp *tg, bool rw, unsigned long jiffy_end) |
e43473b7f blkio: Core imple... |
743 |
{ |
297e3d854 blk-throttle: mak... |
744 |
tg->slice_end[rw] = roundup(jiffy_end, tg->td->throtl_slice); |
fda6f272c blk-throttle: imp... |
745 746 747 748 |
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 blkio: Core imple... |
749 750 751 |
} /* Determine if previously allocated or extended slice is complete or not */ |
0f3457f60 blk-throttle: add... |
752 |
static bool throtl_slice_used(struct throtl_grp *tg, bool rw) |
e43473b7f blkio: Core imple... |
753 754 |
{ if (time_in_range(jiffies, tg->slice_start[rw], tg->slice_end[rw])) |
5cf8c2277 block/blk-throttl... |
755 |
return false; |
e43473b7f blkio: Core imple... |
756 |
|
0b6bad7d6 blk-throttle: ret... |
757 |
return true; |
e43473b7f blkio: Core imple... |
758 759 760 |
} /* Trim the used slices and adjust slice start accordingly */ |
0f3457f60 blk-throttle: add... |
761 |
static inline void throtl_trim_slice(struct throtl_grp *tg, bool rw) |
e43473b7f blkio: Core imple... |
762 |
{ |
3aad5d3ee blkio-throttle: F... |
763 764 |
unsigned long nr_slices, time_elapsed, io_trim; u64 bytes_trim, tmp; |
e43473b7f blkio: Core imple... |
765 766 767 768 769 770 771 772 |
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 blk-throttle: add... |
773 |
if (throtl_slice_used(tg, rw)) |
e43473b7f blkio: Core imple... |
774 |
return; |
d1ae8ffdf blk-throttle: Tri... |
775 776 777 778 779 780 781 |
/* * A bio has been dispatched. Also adjust slice_end. It might happen * that initially cgroup limit was very low resulting in high * slice_end, but later limit was bumped up and bio was dispached * 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 blk-throttle: mak... |
782 |
throtl_set_slice_end(tg, rw, jiffies + tg->td->throtl_slice); |
d1ae8ffdf blk-throttle: Tri... |
783 |
|
e43473b7f blkio: Core imple... |
784 |
time_elapsed = jiffies - tg->slice_start[rw]; |
297e3d854 blk-throttle: mak... |
785 |
nr_slices = time_elapsed / tg->td->throtl_slice; |
e43473b7f blkio: Core imple... |
786 787 788 |
if (!nr_slices) return; |
297e3d854 blk-throttle: mak... |
789 |
tmp = tg_bps_limit(tg, rw) * tg->td->throtl_slice * nr_slices; |
3aad5d3ee blkio-throttle: F... |
790 791 |
do_div(tmp, HZ); bytes_trim = tmp; |
e43473b7f blkio: Core imple... |
792 |
|
297e3d854 blk-throttle: mak... |
793 794 |
io_trim = (tg_iops_limit(tg, rw) * tg->td->throtl_slice * nr_slices) / HZ; |
e43473b7f blkio: Core imple... |
795 |
|
8e89d13f4 blkio: Implementa... |
796 |
if (!bytes_trim && !io_trim) |
e43473b7f blkio: Core imple... |
797 798 799 800 801 802 |
return; if (tg->bytes_disp[rw] >= bytes_trim) tg->bytes_disp[rw] -= bytes_trim; else tg->bytes_disp[rw] = 0; |
8e89d13f4 blkio: Implementa... |
803 804 805 806 |
if (tg->io_disp[rw] >= io_trim) tg->io_disp[rw] -= io_trim; else tg->io_disp[rw] = 0; |
297e3d854 blk-throttle: mak... |
807 |
tg->slice_start[rw] += nr_slices * tg->td->throtl_slice; |
e43473b7f blkio: Core imple... |
808 |
|
fda6f272c blk-throttle: imp... |
809 810 811 812 |
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 blkio: Core imple... |
813 |
} |
0f3457f60 blk-throttle: add... |
814 815 |
static bool tg_with_in_iops_limit(struct throtl_grp *tg, struct bio *bio, unsigned long *wait) |
e43473b7f blkio: Core imple... |
816 817 |
{ bool rw = bio_data_dir(bio); |
8e89d13f4 blkio: Implementa... |
818 |
unsigned int io_allowed; |
e43473b7f blkio: Core imple... |
819 |
unsigned long jiffy_elapsed, jiffy_wait, jiffy_elapsed_rnd; |
c49c06e49 blkio-throttle: F... |
820 |
u64 tmp; |
e43473b7f blkio: Core imple... |
821 |
|
3a10f999f blk-throttle: fix... |
822 |
jiffy_elapsed = jiffies - tg->slice_start[rw]; |
e43473b7f blkio: Core imple... |
823 |
|
3a10f999f blk-throttle: fix... |
824 825 |
/* Round up to the next throttle slice, wait time must be nonzero */ jiffy_elapsed_rnd = roundup(jiffy_elapsed + 1, tg->td->throtl_slice); |
8e89d13f4 blkio: Implementa... |
826 |
|
c49c06e49 blkio-throttle: F... |
827 828 829 830 831 832 |
/* * 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. */ |
9f626e372 blk-throttle: pre... |
833 |
tmp = (u64)tg_iops_limit(tg, rw) * jiffy_elapsed_rnd; |
c49c06e49 blkio-throttle: F... |
834 835 836 837 838 839 |
do_div(tmp, HZ); if (tmp > UINT_MAX) io_allowed = UINT_MAX; else io_allowed = tmp; |
8e89d13f4 blkio: Implementa... |
840 841 |
if (tg->io_disp[rw] + 1 <= io_allowed) { |
e43473b7f blkio: Core imple... |
842 843 |
if (wait) *wait = 0; |
5cf8c2277 block/blk-throttl... |
844 |
return true; |
e43473b7f blkio: Core imple... |
845 |
} |
8e89d13f4 blkio: Implementa... |
846 |
/* Calc approx time to dispatch */ |
991f61fe7 Blk-throttle: red... |
847 |
jiffy_wait = jiffy_elapsed_rnd - jiffy_elapsed; |
8e89d13f4 blkio: Implementa... |
848 849 850 |
if (wait) *wait = jiffy_wait; |
0b6bad7d6 blk-throttle: ret... |
851 |
return false; |
8e89d13f4 blkio: Implementa... |
852 |
} |
0f3457f60 blk-throttle: add... |
853 854 |
static bool tg_with_in_bps_limit(struct throtl_grp *tg, struct bio *bio, unsigned long *wait) |
8e89d13f4 blkio: Implementa... |
855 856 |
{ bool rw = bio_data_dir(bio); |
3aad5d3ee blkio-throttle: F... |
857 |
u64 bytes_allowed, extra_bytes, tmp; |
8e89d13f4 blkio: Implementa... |
858 |
unsigned long jiffy_elapsed, jiffy_wait, jiffy_elapsed_rnd; |
ea0ea2bc6 blk-throttle: cap... |
859 |
unsigned int bio_size = throtl_bio_data_size(bio); |
e43473b7f blkio: Core imple... |
860 861 862 863 864 |
jiffy_elapsed = jiffy_elapsed_rnd = jiffies - tg->slice_start[rw]; /* Slice has just started. Consider one slice interval */ if (!jiffy_elapsed) |
297e3d854 blk-throttle: mak... |
865 |
jiffy_elapsed_rnd = tg->td->throtl_slice; |
e43473b7f blkio: Core imple... |
866 |
|
297e3d854 blk-throttle: mak... |
867 |
jiffy_elapsed_rnd = roundup(jiffy_elapsed_rnd, tg->td->throtl_slice); |
e43473b7f blkio: Core imple... |
868 |
|
9f626e372 blk-throttle: pre... |
869 |
tmp = tg_bps_limit(tg, rw) * jiffy_elapsed_rnd; |
5e901a2b9 blkio-throttle: T... |
870 |
do_div(tmp, HZ); |
3aad5d3ee blkio-throttle: F... |
871 |
bytes_allowed = tmp; |
e43473b7f blkio: Core imple... |
872 |
|
ea0ea2bc6 blk-throttle: cap... |
873 |
if (tg->bytes_disp[rw] + bio_size <= bytes_allowed) { |
e43473b7f blkio: Core imple... |
874 875 |
if (wait) *wait = 0; |
5cf8c2277 block/blk-throttl... |
876 |
return true; |
e43473b7f blkio: Core imple... |
877 878 879 |
} /* Calc approx time to dispatch */ |
ea0ea2bc6 blk-throttle: cap... |
880 |
extra_bytes = tg->bytes_disp[rw] + bio_size - bytes_allowed; |
9f626e372 blk-throttle: pre... |
881 |
jiffy_wait = div64_u64(extra_bytes * HZ, tg_bps_limit(tg, rw)); |
e43473b7f blkio: Core imple... |
882 883 884 885 886 887 888 889 890 |
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 blkio: Core imple... |
891 892 |
if (wait) *wait = jiffy_wait; |
0b6bad7d6 blk-throttle: ret... |
893 |
return false; |
8e89d13f4 blkio: Implementa... |
894 895 896 897 898 899 |
} /* * 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 blk-throttle: add... |
900 901 |
static bool tg_may_dispatch(struct throtl_grp *tg, struct bio *bio, unsigned long *wait) |
8e89d13f4 blkio: Implementa... |
902 903 904 905 906 907 908 909 910 911 |
{ bool rw = bio_data_dir(bio); unsigned long bps_wait = 0, iops_wait = 0, max_wait = 0; /* * 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 blk-throttle: mov... |
912 |
BUG_ON(tg->service_queue.nr_queued[rw] && |
c5cc2070b blk-throttle: add... |
913 |
bio != throtl_peek_queued(&tg->service_queue.queued[rw])); |
e43473b7f blkio: Core imple... |
914 |
|
8e89d13f4 blkio: Implementa... |
915 |
/* If tg->bps = -1, then BW is unlimited */ |
9f626e372 blk-throttle: pre... |
916 917 |
if (tg_bps_limit(tg, rw) == U64_MAX && tg_iops_limit(tg, rw) == UINT_MAX) { |
8e89d13f4 blkio: Implementa... |
918 919 |
if (wait) *wait = 0; |
5cf8c2277 block/blk-throttl... |
920 |
return true; |
8e89d13f4 blkio: Implementa... |
921 922 923 924 925 |
} /* * If previous slice expired, start a new one otherwise renew/extend * existing slice to make sure it is at least throtl_slice interval |
164c80ed8 blk-throttle: Ext... |
926 927 928 |
* 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 blkio: Implementa... |
929 |
*/ |
164c80ed8 blk-throttle: Ext... |
930 |
if (throtl_slice_used(tg, rw) && !(tg->service_queue.nr_queued[rw])) |
0f3457f60 blk-throttle: add... |
931 |
throtl_start_new_slice(tg, rw); |
8e89d13f4 blkio: Implementa... |
932 |
else { |
297e3d854 blk-throttle: mak... |
933 934 935 936 |
if (time_before(tg->slice_end[rw], jiffies + tg->td->throtl_slice)) throtl_extend_slice(tg, rw, jiffies + tg->td->throtl_slice); |
8e89d13f4 blkio: Implementa... |
937 |
} |
0f3457f60 blk-throttle: add... |
938 939 |
if (tg_with_in_bps_limit(tg, bio, &bps_wait) && tg_with_in_iops_limit(tg, bio, &iops_wait)) { |
8e89d13f4 blkio: Implementa... |
940 941 |
if (wait) *wait = 0; |
0b6bad7d6 blk-throttle: ret... |
942 |
return true; |
8e89d13f4 blkio: Implementa... |
943 944 945 946 947 948 949 950 |
} max_wait = max(bps_wait, iops_wait); if (wait) *wait = max_wait; if (time_before(tg->slice_end[rw], jiffies + max_wait)) |
0f3457f60 blk-throttle: add... |
951 |
throtl_extend_slice(tg, rw, jiffies + max_wait); |
e43473b7f blkio: Core imple... |
952 |
|
0b6bad7d6 blk-throttle: ret... |
953 |
return false; |
e43473b7f blkio: Core imple... |
954 955 956 957 958 |
} static void throtl_charge_bio(struct throtl_grp *tg, struct bio *bio) { bool rw = bio_data_dir(bio); |
ea0ea2bc6 blk-throttle: cap... |
959 |
unsigned int bio_size = throtl_bio_data_size(bio); |
e43473b7f blkio: Core imple... |
960 961 |
/* Charge the bio to the group */ |
ea0ea2bc6 blk-throttle: cap... |
962 |
tg->bytes_disp[rw] += bio_size; |
8e89d13f4 blkio: Implementa... |
963 |
tg->io_disp[rw]++; |
ea0ea2bc6 blk-throttle: cap... |
964 |
tg->last_bytes_disp[rw] += bio_size; |
3f0abd806 blk-throttle: add... |
965 |
tg->last_io_disp[rw]++; |
e43473b7f blkio: Core imple... |
966 |
|
2a0f61e6e blk-throttle: set... |
967 |
/* |
8d2bbd4c8 block: replace RE... |
968 |
* BIO_THROTTLED is used to prevent the same bio to be throttled |
2a0f61e6e blk-throttle: set... |
969 970 971 |
* 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 blk-throttle: set... |
972 |
*/ |
8d2bbd4c8 block: replace RE... |
973 974 |
if (!bio_flagged(bio, BIO_THROTTLED)) bio_set_flag(bio, BIO_THROTTLED); |
e43473b7f blkio: Core imple... |
975 |
} |
c5cc2070b blk-throttle: add... |
976 977 978 979 980 981 982 983 984 985 986 |
/** * 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 blkio: Core imple... |
987 |
{ |
73f0d49a9 blk-throttle: mov... |
988 |
struct throtl_service_queue *sq = &tg->service_queue; |
e43473b7f blkio: Core imple... |
989 |
bool rw = bio_data_dir(bio); |
c5cc2070b blk-throttle: add... |
990 991 |
if (!qn) qn = &tg->qnode_on_self[rw]; |
0e9f4164b blk-throttle: gen... |
992 993 994 995 996 997 998 999 |
/* * 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 * cleaered on the next tg_update_disptime(). */ if (!sq->nr_queued[rw]) tg->flags |= THROTL_TG_WAS_EMPTY; |
c5cc2070b blk-throttle: add... |
1000 |
throtl_qnode_add_bio(bio, qn, &sq->queued[rw]); |
73f0d49a9 blk-throttle: mov... |
1001 |
sq->nr_queued[rw]++; |
77216b048 blk-throttle: add... |
1002 |
throtl_enqueue_tg(tg); |
e43473b7f blkio: Core imple... |
1003 |
} |
77216b048 blk-throttle: add... |
1004 |
static void tg_update_disptime(struct throtl_grp *tg) |
e43473b7f blkio: Core imple... |
1005 |
{ |
73f0d49a9 blk-throttle: mov... |
1006 |
struct throtl_service_queue *sq = &tg->service_queue; |
e43473b7f blkio: Core imple... |
1007 1008 |
unsigned long read_wait = -1, write_wait = -1, min_wait = -1, disptime; struct bio *bio; |
d609af3a1 blk-throttle: Adj... |
1009 1010 |
bio = throtl_peek_queued(&sq->queued[READ]); if (bio) |
0f3457f60 blk-throttle: add... |
1011 |
tg_may_dispatch(tg, bio, &read_wait); |
e43473b7f blkio: Core imple... |
1012 |
|
d609af3a1 blk-throttle: Adj... |
1013 1014 |
bio = throtl_peek_queued(&sq->queued[WRITE]); if (bio) |
0f3457f60 blk-throttle: add... |
1015 |
tg_may_dispatch(tg, bio, &write_wait); |
e43473b7f blkio: Core imple... |
1016 1017 1018 |
min_wait = min(read_wait, write_wait); disptime = jiffies + min_wait; |
e43473b7f blkio: Core imple... |
1019 |
/* Update dispatch time */ |
77216b048 blk-throttle: add... |
1020 |
throtl_dequeue_tg(tg); |
e43473b7f blkio: Core imple... |
1021 |
tg->disptime = disptime; |
77216b048 blk-throttle: add... |
1022 |
throtl_enqueue_tg(tg); |
0e9f4164b blk-throttle: gen... |
1023 1024 1025 |
/* see throtl_add_bio_tg() */ tg->flags &= ~THROTL_TG_WAS_EMPTY; |
e43473b7f blkio: Core imple... |
1026 |
} |
32ee5bc47 blk-throttle: Acc... |
1027 1028 1029 1030 1031 1032 1033 1034 1035 |
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 blk-throttle: add... |
1036 |
static void tg_dispatch_one_bio(struct throtl_grp *tg, bool rw) |
e43473b7f blkio: Core imple... |
1037 |
{ |
73f0d49a9 blk-throttle: mov... |
1038 |
struct throtl_service_queue *sq = &tg->service_queue; |
6bc9c2b46 blk-throttle: mak... |
1039 1040 |
struct throtl_service_queue *parent_sq = sq->parent_sq; struct throtl_grp *parent_tg = sq_to_tg(parent_sq); |
c5cc2070b blk-throttle: add... |
1041 |
struct throtl_grp *tg_to_put = NULL; |
e43473b7f blkio: Core imple... |
1042 |
struct bio *bio; |
c5cc2070b blk-throttle: add... |
1043 1044 1045 1046 1047 1048 1049 |
/* * @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 blk-throttle: mov... |
1050 |
sq->nr_queued[rw]--; |
e43473b7f blkio: Core imple... |
1051 1052 |
throtl_charge_bio(tg, bio); |
6bc9c2b46 blk-throttle: mak... |
1053 1054 1055 1056 1057 1058 1059 1060 1061 |
/* * 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 blk-throttle: add... |
1062 |
throtl_add_bio_tg(bio, &tg->qnode_on_parent[rw], parent_tg); |
32ee5bc47 blk-throttle: Acc... |
1063 |
start_parent_slice_with_credit(tg, parent_tg, rw); |
6bc9c2b46 blk-throttle: mak... |
1064 |
} else { |
c5cc2070b blk-throttle: add... |
1065 1066 |
throtl_qnode_add_bio(bio, &tg->qnode_on_parent[rw], &parent_sq->queued[rw]); |
6bc9c2b46 blk-throttle: mak... |
1067 1068 1069 |
BUG_ON(tg->td->nr_queued[rw] <= 0); tg->td->nr_queued[rw]--; } |
e43473b7f blkio: Core imple... |
1070 |
|
0f3457f60 blk-throttle: add... |
1071 |
throtl_trim_slice(tg, rw); |
6bc9c2b46 blk-throttle: mak... |
1072 |
|
c5cc2070b blk-throttle: add... |
1073 1074 |
if (tg_to_put) blkg_put(tg_to_blkg(tg_to_put)); |
e43473b7f blkio: Core imple... |
1075 |
} |
77216b048 blk-throttle: add... |
1076 |
static int throtl_dispatch_tg(struct throtl_grp *tg) |
e43473b7f blkio: Core imple... |
1077 |
{ |
73f0d49a9 blk-throttle: mov... |
1078 |
struct throtl_service_queue *sq = &tg->service_queue; |
e43473b7f blkio: Core imple... |
1079 1080 |
unsigned int nr_reads = 0, nr_writes = 0; unsigned int max_nr_reads = throtl_grp_quantum*3/4; |
c2f6805d4 blk-throttle: Fix... |
1081 |
unsigned int max_nr_writes = throtl_grp_quantum - max_nr_reads; |
e43473b7f blkio: Core imple... |
1082 1083 1084 |
struct bio *bio; /* Try to dispatch 75% READS and 25% WRITES */ |
c5cc2070b blk-throttle: add... |
1085 |
while ((bio = throtl_peek_queued(&sq->queued[READ])) && |
0f3457f60 blk-throttle: add... |
1086 |
tg_may_dispatch(tg, bio, NULL)) { |
e43473b7f blkio: Core imple... |
1087 |
|
77216b048 blk-throttle: add... |
1088 |
tg_dispatch_one_bio(tg, bio_data_dir(bio)); |
e43473b7f blkio: Core imple... |
1089 1090 1091 1092 1093 |
nr_reads++; if (nr_reads >= max_nr_reads) break; } |
c5cc2070b blk-throttle: add... |
1094 |
while ((bio = throtl_peek_queued(&sq->queued[WRITE])) && |
0f3457f60 blk-throttle: add... |
1095 |
tg_may_dispatch(tg, bio, NULL)) { |
e43473b7f blkio: Core imple... |
1096 |
|
77216b048 blk-throttle: add... |
1097 |
tg_dispatch_one_bio(tg, bio_data_dir(bio)); |
e43473b7f blkio: Core imple... |
1098 1099 1100 1101 1102 1103 1104 1105 |
nr_writes++; if (nr_writes >= max_nr_writes) break; } return nr_reads + nr_writes; } |
651930bc1 blk-throttle: dis... |
1106 |
static int throtl_select_dispatch(struct throtl_service_queue *parent_sq) |
e43473b7f blkio: Core imple... |
1107 1108 |
{ unsigned int nr_disp = 0; |
e43473b7f blkio: Core imple... |
1109 1110 |
while (1) { |
73f0d49a9 blk-throttle: mov... |
1111 |
struct throtl_grp *tg = throtl_rb_first(parent_sq); |
2ab74cd29 blk-throttle: fix... |
1112 |
struct throtl_service_queue *sq; |
e43473b7f blkio: Core imple... |
1113 1114 1115 1116 1117 1118 |
if (!tg) break; if (time_before(jiffies, tg->disptime)) break; |
77216b048 blk-throttle: add... |
1119 |
throtl_dequeue_tg(tg); |
e43473b7f blkio: Core imple... |
1120 |
|
77216b048 blk-throttle: add... |
1121 |
nr_disp += throtl_dispatch_tg(tg); |
e43473b7f blkio: Core imple... |
1122 |
|
2ab74cd29 blk-throttle: fix... |
1123 |
sq = &tg->service_queue; |
73f0d49a9 blk-throttle: mov... |
1124 |
if (sq->nr_queued[0] || sq->nr_queued[1]) |
77216b048 blk-throttle: add... |
1125 |
tg_update_disptime(tg); |
e43473b7f blkio: Core imple... |
1126 1127 1128 1129 1130 1131 1132 |
if (nr_disp >= throtl_quantum) break; } return nr_disp; } |
c79892c55 blk-throttle: add... |
1133 1134 |
static bool throtl_can_upgrade(struct throtl_data *td, struct throtl_grp *this_tg); |
6e1a5704c blk-throttle: dis... |
1135 1136 |
/** * throtl_pending_timer_fn - timer function for service_queue->pending_timer |
216382dcc block: Fix throtl... |
1137 |
* @t: the pending_timer member of the throtl_service_queue being serviced |
6e1a5704c blk-throttle: dis... |
1138 1139 1140 1141 |
* * 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 blk-throttle: mak... |
1142 1143 1144 1145 1146 1147 1148 |
* 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 blk-throttle: dis... |
1149 |
*/ |
e99e88a9d treewide: setup_t... |
1150 |
static void throtl_pending_timer_fn(struct timer_list *t) |
69df0ab03 blk-throttle: sep... |
1151 |
{ |
e99e88a9d treewide: setup_t... |
1152 |
struct throtl_service_queue *sq = from_timer(sq, t, pending_timer); |
2e48a530a blk-throttle: mak... |
1153 |
struct throtl_grp *tg = sq_to_tg(sq); |
69df0ab03 blk-throttle: sep... |
1154 |
struct throtl_data *td = sq_to_td(sq); |
cb76199c3 blk-throttle: col... |
1155 |
struct request_queue *q = td->queue; |
2e48a530a blk-throttle: mak... |
1156 1157 |
struct throtl_service_queue *parent_sq; bool dispatched; |
6e1a5704c blk-throttle: dis... |
1158 |
int ret; |
e43473b7f blkio: Core imple... |
1159 |
|
0d945c1f9 block: remove the... |
1160 |
spin_lock_irq(&q->queue_lock); |
c79892c55 blk-throttle: add... |
1161 1162 |
if (throtl_can_upgrade(td, NULL)) throtl_upgrade_state(td); |
2e48a530a blk-throttle: mak... |
1163 1164 1165 |
again: parent_sq = sq->parent_sq; dispatched = false; |
e43473b7f blkio: Core imple... |
1166 |
|
7f52f98c2 blk-throttle: imp... |
1167 1168 |
while (true) { throtl_log(sq, "dispatch nr_queued=%u read=%u write=%u", |
2e48a530a blk-throttle: mak... |
1169 1170 |
sq->nr_queued[READ] + sq->nr_queued[WRITE], sq->nr_queued[READ], sq->nr_queued[WRITE]); |
7f52f98c2 blk-throttle: imp... |
1171 1172 1173 |
ret = throtl_select_dispatch(sq); if (ret) { |
7f52f98c2 blk-throttle: imp... |
1174 1175 1176 |
throtl_log(sq, "bios disp=%u", ret); dispatched = true; } |
e43473b7f blkio: Core imple... |
1177 |
|
7f52f98c2 blk-throttle: imp... |
1178 1179 |
if (throtl_schedule_next_dispatch(sq, false)) break; |
e43473b7f blkio: Core imple... |
1180 |
|
7f52f98c2 blk-throttle: imp... |
1181 |
/* this dispatch windows is still open, relax and repeat */ |
0d945c1f9 block: remove the... |
1182 |
spin_unlock_irq(&q->queue_lock); |
7f52f98c2 blk-throttle: imp... |
1183 |
cpu_relax(); |
0d945c1f9 block: remove the... |
1184 |
spin_lock_irq(&q->queue_lock); |
651930bc1 blk-throttle: dis... |
1185 |
} |
e43473b7f blkio: Core imple... |
1186 |
|
2e48a530a blk-throttle: mak... |
1187 1188 |
if (!dispatched) goto out_unlock; |
6e1a5704c blk-throttle: dis... |
1189 |
|
2e48a530a blk-throttle: mak... |
1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 |
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 { /* reached the top-level, queue issueing */ queue_work(kthrotld_workqueue, &td->dispatch_work); } out_unlock: |
0d945c1f9 block: remove the... |
1206 |
spin_unlock_irq(&q->queue_lock); |
6e1a5704c blk-throttle: dis... |
1207 |
} |
e43473b7f blkio: Core imple... |
1208 |
|
6e1a5704c blk-throttle: dis... |
1209 1210 1211 1212 1213 1214 1215 1216 |
/** * blk_throtl_dispatch_work_fn - work function for throtl_data->dispatch_work * @work: work item being executed * * This function is queued for execution when bio's reach the bio_lists[] * of throtl_data->service_queue. Those bio's are ready and issued by this * function. */ |
8876e140e block/blk-throttl... |
1217 |
static void blk_throtl_dispatch_work_fn(struct work_struct *work) |
6e1a5704c blk-throttle: dis... |
1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 |
{ 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 block: remove the... |
1229 |
spin_lock_irq(&q->queue_lock); |
c5cc2070b blk-throttle: add... |
1230 1231 1232 |
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 block: remove the... |
1233 |
spin_unlock_irq(&q->queue_lock); |
6e1a5704c blk-throttle: dis... |
1234 1235 |
if (!bio_list_empty(&bio_list_on_stack)) { |
69d60eb96 blk-throttle: Use... |
1236 |
blk_start_plug(&plug); |
e43473b7f blkio: Core imple... |
1237 1238 |
while((bio = bio_list_pop(&bio_list_on_stack))) generic_make_request(bio); |
69d60eb96 blk-throttle: Use... |
1239 |
blk_finish_plug(&plug); |
e43473b7f blkio: Core imple... |
1240 |
} |
e43473b7f blkio: Core imple... |
1241 |
} |
f95a04afa blkcg: embed stru... |
1242 1243 |
static u64 tg_prfill_conf_u64(struct seq_file *sf, struct blkg_policy_data *pd, int off) |
60c2bc2d5 blkcg: move conf/... |
1244 |
{ |
f95a04afa blkcg: embed stru... |
1245 1246 |
struct throtl_grp *tg = pd_to_tg(pd); u64 v = *(u64 *)((void *)tg + off); |
60c2bc2d5 blkcg: move conf/... |
1247 |
|
2ab5492de blk-throttle: use... |
1248 |
if (v == U64_MAX) |
60c2bc2d5 blkcg: move conf/... |
1249 |
return 0; |
f95a04afa blkcg: embed stru... |
1250 |
return __blkg_prfill_u64(sf, pd, v); |
60c2bc2d5 blkcg: move conf/... |
1251 |
} |
f95a04afa blkcg: embed stru... |
1252 1253 |
static u64 tg_prfill_conf_uint(struct seq_file *sf, struct blkg_policy_data *pd, int off) |
e43473b7f blkio: Core imple... |
1254 |
{ |
f95a04afa blkcg: embed stru... |
1255 1256 |
struct throtl_grp *tg = pd_to_tg(pd); unsigned int v = *(unsigned int *)((void *)tg + off); |
fe0714377 blkio: Recalculat... |
1257 |
|
2ab5492de blk-throttle: use... |
1258 |
if (v == UINT_MAX) |
af133ceb2 blkcg: move blkio... |
1259 |
return 0; |
f95a04afa blkcg: embed stru... |
1260 |
return __blkg_prfill_u64(sf, pd, v); |
e43473b7f blkio: Core imple... |
1261 |
} |
2da8ca822 cgroup: replace c... |
1262 |
static int tg_print_conf_u64(struct seq_file *sf, void *v) |
8e89d13f4 blkio: Implementa... |
1263 |
{ |
2da8ca822 cgroup: replace c... |
1264 1265 |
blkcg_print_blkgs(sf, css_to_blkcg(seq_css(sf)), tg_prfill_conf_u64, &blkcg_policy_throtl, seq_cft(sf)->private, false); |
af133ceb2 blkcg: move blkio... |
1266 |
return 0; |
8e89d13f4 blkio: Implementa... |
1267 |
} |
2da8ca822 cgroup: replace c... |
1268 |
static int tg_print_conf_uint(struct seq_file *sf, void *v) |
8e89d13f4 blkio: Implementa... |
1269 |
{ |
2da8ca822 cgroup: replace c... |
1270 1271 |
blkcg_print_blkgs(sf, css_to_blkcg(seq_css(sf)), tg_prfill_conf_uint, &blkcg_policy_throtl, seq_cft(sf)->private, false); |
af133ceb2 blkcg: move blkio... |
1272 |
return 0; |
60c2bc2d5 blkcg: move conf/... |
1273 |
} |
9bb67aeb9 blk-throttle: res... |
1274 |
static void tg_conf_updated(struct throtl_grp *tg, bool global) |
60c2bc2d5 blkcg: move conf/... |
1275 |
{ |
69948b070 blkcg: separate o... |
1276 |
struct throtl_service_queue *sq = &tg->service_queue; |
492eb21b9 cgroup: make hier... |
1277 |
struct cgroup_subsys_state *pos_css; |
69948b070 blkcg: separate o... |
1278 |
struct blkcg_gq *blkg; |
af133ceb2 blkcg: move blkio... |
1279 |
|
fda6f272c blk-throttle: imp... |
1280 1281 |
throtl_log(&tg->service_queue, "limit change rbps=%llu wbps=%llu riops=%u wiops=%u", |
9f626e372 blk-throttle: pre... |
1282 1283 |
tg_bps_limit(tg, READ), tg_bps_limit(tg, WRITE), tg_iops_limit(tg, READ), tg_iops_limit(tg, WRITE)); |
632b44935 blk-throttle: rem... |
1284 1285 |
/* |
693e751e7 blk-throttle: imp... |
1286 1287 1288 1289 1290 1291 |
* 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 blk-throttle: res... |
1292 1293 |
blkg_for_each_descendant_pre(blkg, pos_css, global ? tg->td->queue->root_blkg : tg_to_blkg(tg)) { |
5b81fc3cc blk-throttle: add... |
1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 |
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 blk-throttle: imp... |
1312 1313 |
/* |
632b44935 blk-throttle: rem... |
1314 1315 1316 1317 1318 1319 1320 |
* 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. */ |
0f3457f60 blk-throttle: add... |
1321 1322 |
throtl_start_new_slice(tg, 0); throtl_start_new_slice(tg, 1); |
632b44935 blk-throttle: rem... |
1323 |
|
5b2c16aae blk-throttle: sim... |
1324 |
if (tg->flags & THROTL_TG_PENDING) { |
77216b048 blk-throttle: add... |
1325 |
tg_update_disptime(tg); |
7f52f98c2 blk-throttle: imp... |
1326 |
throtl_schedule_next_dispatch(sq->parent_sq, true); |
632b44935 blk-throttle: rem... |
1327 |
} |
69948b070 blkcg: separate o... |
1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 |
} 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 blk-throttle: use... |
1347 |
v = U64_MAX; |
69948b070 blkcg: separate o... |
1348 1349 1350 1351 1352 1353 1354 |
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 blkcg: move conf/... |
1355 |
|
9bb67aeb9 blk-throttle: res... |
1356 |
tg_conf_updated(tg, false); |
36aa9e5f5 blkcg: move body ... |
1357 1358 |
ret = 0; out_finish: |
60c2bc2d5 blkcg: move conf/... |
1359 |
blkg_conf_finish(&ctx); |
36aa9e5f5 blkcg: move body ... |
1360 |
return ret ?: nbytes; |
8e89d13f4 blkio: Implementa... |
1361 |
} |
451af504d cgroup: replace c... |
1362 1363 |
static ssize_t tg_set_conf_u64(struct kernfs_open_file *of, char *buf, size_t nbytes, loff_t off) |
60c2bc2d5 blkcg: move conf/... |
1364 |
{ |
451af504d cgroup: replace c... |
1365 |
return tg_set_conf(of, buf, nbytes, off, true); |
60c2bc2d5 blkcg: move conf/... |
1366 |
} |
451af504d cgroup: replace c... |
1367 1368 |
static ssize_t tg_set_conf_uint(struct kernfs_open_file *of, char *buf, size_t nbytes, loff_t off) |
60c2bc2d5 blkcg: move conf/... |
1369 |
{ |
451af504d cgroup: replace c... |
1370 |
return tg_set_conf(of, buf, nbytes, off, false); |
60c2bc2d5 blkcg: move conf/... |
1371 |
} |
880f50e22 blkcg: mark exist... |
1372 |
static struct cftype throtl_legacy_files[] = { |
60c2bc2d5 blkcg: move conf/... |
1373 1374 |
{ .name = "throttle.read_bps_device", |
9f626e372 blk-throttle: pre... |
1375 |
.private = offsetof(struct throtl_grp, bps[READ][LIMIT_MAX]), |
2da8ca822 cgroup: replace c... |
1376 |
.seq_show = tg_print_conf_u64, |
451af504d cgroup: replace c... |
1377 |
.write = tg_set_conf_u64, |
60c2bc2d5 blkcg: move conf/... |
1378 1379 1380 |
}, { .name = "throttle.write_bps_device", |
9f626e372 blk-throttle: pre... |
1381 |
.private = offsetof(struct throtl_grp, bps[WRITE][LIMIT_MAX]), |
2da8ca822 cgroup: replace c... |
1382 |
.seq_show = tg_print_conf_u64, |
451af504d cgroup: replace c... |
1383 |
.write = tg_set_conf_u64, |
60c2bc2d5 blkcg: move conf/... |
1384 1385 1386 |
}, { .name = "throttle.read_iops_device", |
9f626e372 blk-throttle: pre... |
1387 |
.private = offsetof(struct throtl_grp, iops[READ][LIMIT_MAX]), |
2da8ca822 cgroup: replace c... |
1388 |
.seq_show = tg_print_conf_uint, |
451af504d cgroup: replace c... |
1389 |
.write = tg_set_conf_uint, |
60c2bc2d5 blkcg: move conf/... |
1390 1391 1392 |
}, { .name = "throttle.write_iops_device", |
9f626e372 blk-throttle: pre... |
1393 |
.private = offsetof(struct throtl_grp, iops[WRITE][LIMIT_MAX]), |
2da8ca822 cgroup: replace c... |
1394 |
.seq_show = tg_print_conf_uint, |
451af504d cgroup: replace c... |
1395 |
.write = tg_set_conf_uint, |
60c2bc2d5 blkcg: move conf/... |
1396 1397 1398 |
}, { .name = "throttle.io_service_bytes", |
77ea73388 blkcg: move io_se... |
1399 1400 |
.private = (unsigned long)&blkcg_policy_throtl, .seq_show = blkg_print_stat_bytes, |
60c2bc2d5 blkcg: move conf/... |
1401 1402 |
}, { |
17534c6f2 blk-throttle: exp... |
1403 1404 1405 1406 1407 |
.name = "throttle.io_service_bytes_recursive", .private = (unsigned long)&blkcg_policy_throtl, .seq_show = blkg_print_stat_bytes_recursive, }, { |
60c2bc2d5 blkcg: move conf/... |
1408 |
.name = "throttle.io_serviced", |
77ea73388 blkcg: move io_se... |
1409 1410 |
.private = (unsigned long)&blkcg_policy_throtl, .seq_show = blkg_print_stat_ios, |
60c2bc2d5 blkcg: move conf/... |
1411 |
}, |
17534c6f2 blk-throttle: exp... |
1412 1413 1414 1415 1416 |
{ .name = "throttle.io_serviced_recursive", .private = (unsigned long)&blkcg_policy_throtl, .seq_show = blkg_print_stat_ios_recursive, }, |
60c2bc2d5 blkcg: move conf/... |
1417 1418 |
{ } /* terminate */ }; |
cd5ab1b0f blk-throttle: add... |
1419 |
static u64 tg_prfill_limit(struct seq_file *sf, struct blkg_policy_data *pd, |
2ee867dcf blkcg: implement ... |
1420 1421 1422 1423 1424 |
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 blk-throttle: add... |
1425 1426 |
u64 bps_dft; unsigned int iops_dft; |
ada75b6e5 blk-throttle: add... |
1427 |
char idle_time[26] = ""; |
ec80991d6 blk-throttle: add... |
1428 |
char latency_time[26] = ""; |
2ee867dcf blkcg: implement ... |
1429 1430 1431 |
if (!dname) return 0; |
9f626e372 blk-throttle: pre... |
1432 |
|
cd5ab1b0f blk-throttle: add... |
1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 |
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 blk-throttle: add... |
1444 |
tg->iops_conf[WRITE][off] == iops_dft && |
ec80991d6 blk-throttle: add... |
1445 |
(off != LIMIT_LOW || |
b4f428ef2 blk-throttle: for... |
1446 |
(tg->idletime_threshold_conf == DFL_IDLE_THRESHOLD && |
5b81fc3cc blk-throttle: add... |
1447 |
tg->latency_target_conf == DFL_LATENCY_TARGET))) |
2ee867dcf blkcg: implement ... |
1448 |
return 0; |
9bb67aeb9 blk-throttle: res... |
1449 |
if (tg->bps_conf[READ][off] != U64_MAX) |
9f626e372 blk-throttle: pre... |
1450 |
snprintf(bufs[0], sizeof(bufs[0]), "%llu", |
cd5ab1b0f blk-throttle: add... |
1451 |
tg->bps_conf[READ][off]); |
9bb67aeb9 blk-throttle: res... |
1452 |
if (tg->bps_conf[WRITE][off] != U64_MAX) |
9f626e372 blk-throttle: pre... |
1453 |
snprintf(bufs[1], sizeof(bufs[1]), "%llu", |
cd5ab1b0f blk-throttle: add... |
1454 |
tg->bps_conf[WRITE][off]); |
9bb67aeb9 blk-throttle: res... |
1455 |
if (tg->iops_conf[READ][off] != UINT_MAX) |
9f626e372 blk-throttle: pre... |
1456 |
snprintf(bufs[2], sizeof(bufs[2]), "%u", |
cd5ab1b0f blk-throttle: add... |
1457 |
tg->iops_conf[READ][off]); |
9bb67aeb9 blk-throttle: res... |
1458 |
if (tg->iops_conf[WRITE][off] != UINT_MAX) |
9f626e372 blk-throttle: pre... |
1459 |
snprintf(bufs[3], sizeof(bufs[3]), "%u", |
cd5ab1b0f blk-throttle: add... |
1460 |
tg->iops_conf[WRITE][off]); |
ada75b6e5 blk-throttle: add... |
1461 |
if (off == LIMIT_LOW) { |
5b81fc3cc blk-throttle: add... |
1462 |
if (tg->idletime_threshold_conf == ULONG_MAX) |
ada75b6e5 blk-throttle: add... |
1463 1464 1465 |
strcpy(idle_time, " idle=max"); else snprintf(idle_time, sizeof(idle_time), " idle=%lu", |
5b81fc3cc blk-throttle: add... |
1466 |
tg->idletime_threshold_conf); |
ec80991d6 blk-throttle: add... |
1467 |
|
5b81fc3cc blk-throttle: add... |
1468 |
if (tg->latency_target_conf == ULONG_MAX) |
ec80991d6 blk-throttle: add... |
1469 1470 1471 |
strcpy(latency_time, " latency=max"); else snprintf(latency_time, sizeof(latency_time), |
5b81fc3cc blk-throttle: add... |
1472 |
" latency=%lu", tg->latency_target_conf); |
ada75b6e5 blk-throttle: add... |
1473 |
} |
2ee867dcf blkcg: implement ... |
1474 |
|
ec80991d6 blk-throttle: add... |
1475 1476 1477 1478 |
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 blkcg: implement ... |
1479 1480 |
return 0; } |
cd5ab1b0f blk-throttle: add... |
1481 |
static int tg_print_limit(struct seq_file *sf, void *v) |
2ee867dcf blkcg: implement ... |
1482 |
{ |
cd5ab1b0f blk-throttle: add... |
1483 |
blkcg_print_blkgs(sf, css_to_blkcg(seq_css(sf)), tg_prfill_limit, |
2ee867dcf blkcg: implement ... |
1484 1485 1486 |
&blkcg_policy_throtl, seq_cft(sf)->private, false); return 0; } |
cd5ab1b0f blk-throttle: add... |
1487 |
static ssize_t tg_set_limit(struct kernfs_open_file *of, |
2ee867dcf blkcg: implement ... |
1488 1489 1490 1491 1492 1493 |
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 blk-throttle: add... |
1494 |
unsigned long idle_time; |
ec80991d6 blk-throttle: add... |
1495 |
unsigned long latency_time; |
2ee867dcf blkcg: implement ... |
1496 |
int ret; |
cd5ab1b0f blk-throttle: add... |
1497 |
int index = of_cft(of)->private; |
2ee867dcf blkcg: implement ... |
1498 1499 1500 1501 1502 1503 |
ret = blkg_conf_prep(blkcg, &blkcg_policy_throtl, buf, &ctx); if (ret) return ret; tg = blkg_to_tg(ctx.blkg); |
cd5ab1b0f blk-throttle: add... |
1504 1505 1506 1507 |
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 blkcg: implement ... |
1508 |
|
5b81fc3cc blk-throttle: add... |
1509 1510 |
idle_time = tg->idletime_threshold_conf; latency_time = tg->latency_target_conf; |
2ee867dcf blkcg: implement ... |
1511 1512 1513 |
while (true) { char tok[27]; /* wiops=18446744073709551616 */ char *p; |
2ab5492de blk-throttle: use... |
1514 |
u64 val = U64_MAX; |
2ee867dcf blkcg: implement ... |
1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 |
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; if (!strcmp(tok, "rbps")) v[0] = val; else if (!strcmp(tok, "wbps")) v[1] = val; else if (!strcmp(tok, "riops")) v[2] = min_t(u64, val, UINT_MAX); else if (!strcmp(tok, "wiops")) v[3] = min_t(u64, val, UINT_MAX); |
ada75b6e5 blk-throttle: add... |
1542 1543 |
else if (off == LIMIT_LOW && !strcmp(tok, "idle")) idle_time = val; |
ec80991d6 blk-throttle: add... |
1544 1545 |
else if (off == LIMIT_LOW && !strcmp(tok, "latency")) latency_time = val; |
2ee867dcf blkcg: implement ... |
1546 1547 1548 |
else goto out_finish; } |
cd5ab1b0f blk-throttle: add... |
1549 1550 1551 1552 |
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 blkcg: implement ... |
1553 |
|
cd5ab1b0f blk-throttle: add... |
1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 |
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 blk-throttle: for... |
1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 |
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 blk-throttle: add... |
1583 |
tg->idletime_threshold = tg->idletime_threshold_conf; |
5b81fc3cc blk-throttle: add... |
1584 |
tg->latency_target = tg->latency_target_conf; |
cd5ab1b0f blk-throttle: add... |
1585 |
} |
b4f428ef2 blk-throttle: for... |
1586 1587 1588 1589 1590 1591 1592 |
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 blk-throttle: res... |
1593 1594 |
tg_conf_updated(tg, index == LIMIT_LOW && tg->td->limit_valid[LIMIT_LOW]); |
2ee867dcf blkcg: implement ... |
1595 1596 1597 1598 1599 1600 1601 |
ret = 0; out_finish: blkg_conf_finish(&ctx); return ret ?: nbytes; } static struct cftype throtl_files[] = { |
cd5ab1b0f blk-throttle: add... |
1602 1603 1604 1605 1606 1607 1608 1609 1610 |
#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 blkcg: implement ... |
1611 1612 1613 |
{ .name = "max", .flags = CFTYPE_NOT_ON_ROOT, |
cd5ab1b0f blk-throttle: add... |
1614 1615 1616 |
.seq_show = tg_print_limit, .write = tg_set_limit, .private = LIMIT_MAX, |
2ee867dcf blkcg: implement ... |
1617 1618 1619 |
}, { } /* terminate */ }; |
da5277700 block: Move blk_t... |
1620 |
static void throtl_shutdown_wq(struct request_queue *q) |
e43473b7f blkio: Core imple... |
1621 1622 |
{ struct throtl_data *td = q->td; |
69df0ab03 blk-throttle: sep... |
1623 |
cancel_work_sync(&td->dispatch_work); |
e43473b7f blkio: Core imple... |
1624 |
} |
3c798398e blkcg: mass renam... |
1625 |
static struct blkcg_policy blkcg_policy_throtl = { |
2ee867dcf blkcg: implement ... |
1626 |
.dfl_cftypes = throtl_files, |
880f50e22 blkcg: mark exist... |
1627 |
.legacy_cftypes = throtl_legacy_files, |
f9fcc2d39 blkcg: collapse b... |
1628 |
|
001bea73e blkcg: replace bl... |
1629 |
.pd_alloc_fn = throtl_pd_alloc, |
f9fcc2d39 blkcg: collapse b... |
1630 |
.pd_init_fn = throtl_pd_init, |
693e751e7 blk-throttle: imp... |
1631 |
.pd_online_fn = throtl_pd_online, |
cd5ab1b0f blk-throttle: add... |
1632 |
.pd_offline_fn = throtl_pd_offline, |
001bea73e blkcg: replace bl... |
1633 |
.pd_free_fn = throtl_pd_free, |
e43473b7f blkio: Core imple... |
1634 |
}; |
3f0abd806 blk-throttle: add... |
1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 |
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 blk-throttle: add... |
1673 1674 1675 1676 1677 |
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 blk-throttle: for... |
1678 |
* configure a too big threshold) or 4 times of idletime threshold |
9e234eeaf blk-throttle: add... |
1679 |
* - average think time is more than threshold |
53696b8d2 blk-throttle: add... |
1680 |
* - IO latency is largely below threshold |
9e234eeaf blk-throttle: add... |
1681 |
*/ |
b4f428ef2 blk-throttle: for... |
1682 |
unsigned long time; |
4cff729f6 blk-throttle: out... |
1683 |
bool ret; |
9e234eeaf blk-throttle: add... |
1684 |
|
b4f428ef2 blk-throttle: for... |
1685 1686 1687 1688 1689 1690 |
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 blk-throttle: add... |
1691 |
tg->bad_bio_cnt * 5 < tg->bio_cnt); |
4cff729f6 blk-throttle: out... |
1692 1693 1694 1695 1696 |
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 blk-throttle: add... |
1697 |
} |
c79892c55 blk-throttle: add... |
1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 |
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 blk-throttle: det... |
1717 1718 |
if (time_after_eq(jiffies, |
fa6fb5aab blk-throttle: ign... |
1719 1720 |
tg_last_low_overflow_time(tg) + tg->td->throtl_slice) && throtl_tg_is_idle(tg)) |
aec242468 blk-throttle: det... |
1721 |
return true; |
c79892c55 blk-throttle: add... |
1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 |
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 blk-throttle: mak... |
1745 |
if (time_before(jiffies, td->low_downgrade_time + td->throtl_slice)) |
3f0abd806 blk-throttle: add... |
1746 |
return false; |
c79892c55 blk-throttle: add... |
1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 |
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 blk-throttle: ign... |
1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 |
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 blk-throttle: add... |
1782 1783 1784 1785 |
static void throtl_upgrade_state(struct throtl_data *td) { struct cgroup_subsys_state *pos_css; struct blkcg_gq *blkg; |
4cff729f6 blk-throttle: out... |
1786 |
throtl_log(&td->service_queue, "upgrade to max"); |
c79892c55 blk-throttle: add... |
1787 |
td->limit_index = LIMIT_MAX; |
3f0abd806 blk-throttle: add... |
1788 |
td->low_upgrade_time = jiffies; |
7394e31fa blk-throttle: mak... |
1789 |
td->scale = 0; |
c79892c55 blk-throttle: add... |
1790 1791 1792 1793 1794 1795 1796 |
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 blk-throttle: fix... |
1797 |
throtl_schedule_next_dispatch(sq, true); |
c79892c55 blk-throttle: add... |
1798 1799 1800 |
} rcu_read_unlock(); throtl_select_dispatch(&td->service_queue); |
4f02fb761 blk-throttle: fix... |
1801 |
throtl_schedule_next_dispatch(&td->service_queue, true); |
c79892c55 blk-throttle: add... |
1802 1803 |
queue_work(kthrotld_workqueue, &td->dispatch_work); } |
3f0abd806 blk-throttle: add... |
1804 1805 |
static void throtl_downgrade_state(struct throtl_data *td, int new) { |
7394e31fa blk-throttle: mak... |
1806 |
td->scale /= 2; |
4cff729f6 blk-throttle: out... |
1807 |
throtl_log(&td->service_queue, "downgrade, scale %d", td->scale); |
7394e31fa blk-throttle: mak... |
1808 1809 1810 1811 |
if (td->scale) { td->low_upgrade_time = jiffies - td->scale * td->throtl_slice; return; } |
3f0abd806 blk-throttle: add... |
1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 |
td->limit_index = new; 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 blk-throttle: mak... |
1825 1826 |
if (time_after_eq(now, td->low_upgrade_time + td->throtl_slice) && time_after_eq(now, tg_last_low_overflow_time(tg) + |
fa6fb5aab blk-throttle: ign... |
1827 1828 1829 |
td->throtl_slice) && (!throtl_tg_is_idle(tg) || !list_empty(&tg_to_blkg(tg)->blkcg->css.children))) |
3f0abd806 blk-throttle: add... |
1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 |
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 blk-throttle: mak... |
1858 |
if (time_after(tg->last_check_time + tg->td->throtl_slice, now)) |
3f0abd806 blk-throttle: add... |
1859 1860 1861 1862 |
return; elapsed_time = now - tg->last_check_time; tg->last_check_time = now; |
297e3d854 blk-throttle: mak... |
1863 1864 |
if (time_before(now, tg_last_low_overflow_time(tg) + tg->td->throtl_slice)) |
3f0abd806 blk-throttle: add... |
1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 |
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]) { 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]) { 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)) throtl_downgrade_state(tg->td, LIMIT_LOW); 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 blk-throttle: add... |
1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 |
static void blk_throtl_update_idletime(struct throtl_grp *tg) { unsigned long now = ktime_get_ns() >> 10; unsigned long last_finish_time = tg->last_finish_time; if (now <= last_finish_time || last_finish_time == 0 || 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 blk-throttle: add... |
1917 1918 1919 |
#ifdef CONFIG_BLK_DEV_THROTTLING_LOW static void throtl_update_latency_buckets(struct throtl_data *td) { |
b889bf66d blk-throttle: tra... |
1920 1921 1922 1923 |
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 blk-throttle: add... |
1924 1925 1926 1927 1928 1929 1930 1931 |
if (!blk_queue_nonrot(td->queue)) return; if (time_before(jiffies, td->last_calculate_time + HZ)) return; td->last_calculate_time = jiffies; memset(avg_latency, 0, sizeof(avg_latency)); |
b889bf66d blk-throttle: tra... |
1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 |
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 blk-throttle: add... |
1947 |
|
b889bf66d blk-throttle: tra... |
1948 1949 |
if (tmp->samples >= 32) { int samples = tmp->samples; |
b9147dd1b blk-throttle: add... |
1950 |
|
b889bf66d blk-throttle: tra... |
1951 |
latency[rw] = tmp->total_latency; |
b9147dd1b blk-throttle: add... |
1952 |
|
b889bf66d blk-throttle: tra... |
1953 1954 1955 1956 1957 1958 1959 |
tmp->total_latency = 0; tmp->samples = 0; latency[rw] /= samples; if (latency[rw] == 0) continue; avg_latency[rw][i].latency = latency[rw]; } |
b9147dd1b blk-throttle: add... |
1960 1961 |
} } |
b889bf66d blk-throttle: tra... |
1962 1963 1964 1965 1966 1967 1968 1969 |
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 blk-throttle: add... |
1970 |
|
b889bf66d blk-throttle: tra... |
1971 1972 1973 1974 1975 |
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 blk-throttle: add... |
1976 |
|
b889bf66d blk-throttle: tra... |
1977 1978 1979 1980 1981 |
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 blk-throttle: add... |
1982 |
} |
4cff729f6 blk-throttle: out... |
1983 1984 1985 |
for (i = 0; i < LATENCY_BUCKET_SIZE; i++) throtl_log(&td->service_queue, |
b889bf66d blk-throttle: tra... |
1986 1987 1988 1989 1990 1991 |
"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 blk-throttle: add... |
1992 1993 1994 1995 1996 1997 |
} #else static inline void throtl_update_latency_buckets(struct throtl_data *td) { } #endif |
ae1188963 blkcg: consolidat... |
1998 1999 |
bool blk_throtl_bio(struct request_queue *q, struct blkcg_gq *blkg, struct bio *bio) |
e43473b7f blkio: Core imple... |
2000 |
{ |
c5cc2070b blk-throttle: add... |
2001 |
struct throtl_qnode *qn = NULL; |
b5f2954d3 blkcg: revert blk... |
2002 |
struct throtl_grp *tg = blkg_to_tg(blkg ?: q->root_blkg); |
73f0d49a9 blk-throttle: mov... |
2003 |
struct throtl_service_queue *sq; |
0e9f4164b blk-throttle: gen... |
2004 |
bool rw = bio_data_dir(bio); |
bc16a4f93 block: reorganize... |
2005 |
bool throttled = false; |
b9147dd1b blk-throttle: add... |
2006 |
struct throtl_data *td = tg->td; |
e43473b7f blkio: Core imple... |
2007 |
|
ae1188963 blkcg: consolidat... |
2008 |
WARN_ON_ONCE(!rcu_read_lock_held()); |
2a0f61e6e blk-throttle: set... |
2009 |
/* see throtl_charge_bio() */ |
8d2bbd4c8 block: replace RE... |
2010 |
if (bio_flagged(bio, BIO_THROTTLED) || !tg->has_rules[rw]) |
bc16a4f93 block: reorganize... |
2011 |
goto out; |
e43473b7f blkio: Core imple... |
2012 |
|
0d945c1f9 block: remove the... |
2013 |
spin_lock_irq(&q->queue_lock); |
c9589f03e blk-throttle: imp... |
2014 |
|
b9147dd1b blk-throttle: add... |
2015 |
throtl_update_latency_buckets(td); |
9e234eeaf blk-throttle: add... |
2016 |
blk_throtl_update_idletime(tg); |
73f0d49a9 blk-throttle: mov... |
2017 |
sq = &tg->service_queue; |
c79892c55 blk-throttle: add... |
2018 |
again: |
9e660acff blk-throttle: mak... |
2019 |
while (true) { |
3f0abd806 blk-throttle: add... |
2020 2021 2022 |
if (tg->last_low_overflow_time[rw] == 0) tg->last_low_overflow_time[rw] = jiffies; throtl_downgrade_check(tg); |
fa6fb5aab blk-throttle: ign... |
2023 |
throtl_upgrade_check(tg); |
9e660acff blk-throttle: mak... |
2024 2025 2026 |
/* throtl is FIFO - if bios are already queued, should queue */ if (sq->nr_queued[rw]) break; |
de701c74a blk-throttle: Som... |
2027 |
|
9e660acff blk-throttle: mak... |
2028 |
/* if above limits, break to queue */ |
c79892c55 blk-throttle: add... |
2029 |
if (!tg_may_dispatch(tg, bio, NULL)) { |
3f0abd806 blk-throttle: add... |
2030 |
tg->last_low_overflow_time[rw] = jiffies; |
b9147dd1b blk-throttle: add... |
2031 2032 |
if (throtl_can_upgrade(td, tg)) { throtl_upgrade_state(td); |
c79892c55 blk-throttle: add... |
2033 2034 |
goto again; } |
9e660acff blk-throttle: mak... |
2035 |
break; |
c79892c55 blk-throttle: add... |
2036 |
} |
9e660acff blk-throttle: mak... |
2037 2038 |
/* within limits, let's charge and dispatch directly */ |
e43473b7f blkio: Core imple... |
2039 |
throtl_charge_bio(tg, bio); |
04521db04 blk-throttle: Res... |
2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 |
/* * 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 blk-throttle: add... |
2052 |
throtl_trim_slice(tg, rw); |
9e660acff blk-throttle: mak... |
2053 2054 2055 2056 2057 2058 |
/* * @bio passed through this layer without being throttled. * Climb up the ladder. If we''re already at the top, it * can be executed directly. */ |
c5cc2070b blk-throttle: add... |
2059 |
qn = &tg->qnode_on_parent[rw]; |
9e660acff blk-throttle: mak... |
2060 2061 2062 2063 |
sq = sq->parent_sq; tg = sq_to_tg(sq); if (!tg) goto out_unlock; |
e43473b7f blkio: Core imple... |
2064 |
} |
9e660acff blk-throttle: mak... |
2065 |
/* out-of-limit, queue to @tg */ |
fda6f272c blk-throttle: imp... |
2066 2067 |
throtl_log(sq, "[%c] bio. bdisp=%llu sz=%u bps=%llu iodisp=%u iops=%u queued=%d/%d", rw == READ ? 'R' : 'W', |
9f626e372 blk-throttle: pre... |
2068 2069 2070 |
tg->bytes_disp[rw], bio->bi_iter.bi_size, tg_bps_limit(tg, rw), tg->io_disp[rw], tg_iops_limit(tg, rw), |
fda6f272c blk-throttle: imp... |
2071 |
sq->nr_queued[READ], sq->nr_queued[WRITE]); |
e43473b7f blkio: Core imple... |
2072 |
|
3f0abd806 blk-throttle: add... |
2073 |
tg->last_low_overflow_time[rw] = jiffies; |
b9147dd1b blk-throttle: add... |
2074 |
td->nr_queued[rw]++; |
c5cc2070b blk-throttle: add... |
2075 |
throtl_add_bio_tg(bio, qn, tg); |
bc16a4f93 block: reorganize... |
2076 |
throttled = true; |
e43473b7f blkio: Core imple... |
2077 |
|
7f52f98c2 blk-throttle: imp... |
2078 2079 2080 2081 2082 2083 |
/* * 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 blk-throttle: gen... |
2084 |
if (tg->flags & THROTL_TG_WAS_EMPTY) { |
77216b048 blk-throttle: add... |
2085 |
tg_update_disptime(tg); |
7f52f98c2 blk-throttle: imp... |
2086 |
throtl_schedule_next_dispatch(tg->service_queue.parent_sq, true); |
e43473b7f blkio: Core imple... |
2087 |
} |
bc16a4f93 block: reorganize... |
2088 |
out_unlock: |
0d945c1f9 block: remove the... |
2089 |
spin_unlock_irq(&q->queue_lock); |
bc16a4f93 block: reorganize... |
2090 |
out: |
111be8839 block-throttle: a... |
2091 |
bio_set_flag(bio, BIO_THROTTLED); |
b9147dd1b blk-throttle: add... |
2092 2093 2094 |
#ifdef CONFIG_BLK_DEV_THROTTLING_LOW if (throttled || !td->track_bio_latency) |
5238dcf41 block: replace bi... |
2095 |
bio->bi_issue.value |= BIO_ISSUE_THROTL_SKIP_LATENCY; |
b9147dd1b blk-throttle: add... |
2096 |
#endif |
bc16a4f93 block: reorganize... |
2097 |
return throttled; |
e43473b7f blkio: Core imple... |
2098 |
} |
9e234eeaf blk-throttle: add... |
2099 |
#ifdef CONFIG_BLK_DEV_THROTTLING_LOW |
b9147dd1b blk-throttle: add... |
2100 2101 2102 2103 2104 |
static void throtl_track_latency(struct throtl_data *td, sector_t size, int op, unsigned long time) { struct latency_bucket *latency; int index; |
b889bf66d blk-throttle: tra... |
2105 2106 |
if (!td || td->limit_index != LIMIT_LOW || !(op == REQ_OP_READ || op == REQ_OP_WRITE) || |
b9147dd1b blk-throttle: add... |
2107 2108 2109 2110 |
!blk_queue_nonrot(td->queue)) return; index = request_bucket_index(size); |
b889bf66d blk-throttle: tra... |
2111 |
latency = get_cpu_ptr(td->latency_buckets[op]); |
b9147dd1b blk-throttle: add... |
2112 2113 |
latency[index].total_latency += time; latency[index].samples++; |
b889bf66d blk-throttle: tra... |
2114 |
put_cpu_ptr(td->latency_buckets[op]); |
b9147dd1b blk-throttle: add... |
2115 2116 2117 2118 2119 2120 |
} void blk_throtl_stat_add(struct request *rq, u64 time_ns) { struct request_queue *q = rq->q; struct throtl_data *td = q->td; |
3d2443069 block: make rq se... |
2121 2122 |
throtl_track_latency(td, blk_rq_stats_sectors(rq), req_op(rq), time_ns >> 10); |
b9147dd1b blk-throttle: add... |
2123 |
} |
9e234eeaf blk-throttle: add... |
2124 2125 |
void blk_throtl_bio_endio(struct bio *bio) { |
08e18eab0 block: add bi_blk... |
2126 |
struct blkcg_gq *blkg; |
9e234eeaf blk-throttle: add... |
2127 |
struct throtl_grp *tg; |
b9147dd1b blk-throttle: add... |
2128 2129 2130 2131 |
u64 finish_time_ns; unsigned long finish_time; unsigned long start_time; unsigned long lat; |
b889bf66d blk-throttle: tra... |
2132 |
int rw = bio_data_dir(bio); |
9e234eeaf blk-throttle: add... |
2133 |
|
08e18eab0 block: add bi_blk... |
2134 2135 |
blkg = bio->bi_blkg; if (!blkg) |
9e234eeaf blk-throttle: add... |
2136 |
return; |
08e18eab0 block: add bi_blk... |
2137 |
tg = blkg_to_tg(blkg); |
9e234eeaf blk-throttle: add... |
2138 |
|
b9147dd1b blk-throttle: add... |
2139 2140 |
finish_time_ns = ktime_get_ns(); tg->last_finish_time = finish_time_ns >> 10; |
5238dcf41 block: replace bi... |
2141 2142 |
start_time = bio_issue_time(&bio->bi_issue) >> 10; finish_time = __bio_issue_time(finish_time_ns) >> 10; |
08e18eab0 block: add bi_blk... |
2143 |
if (!start_time || finish_time <= start_time) |
53696b8d2 blk-throttle: add... |
2144 2145 2146 |
return; lat = finish_time - start_time; |
b9147dd1b blk-throttle: add... |
2147 |
/* this is only for bio based driver */ |
5238dcf41 block: replace bi... |
2148 2149 2150 |
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 blk-throttle: add... |
2151 |
|
6679a90c4 blk-throttle: set... |
2152 |
if (tg->latency_target && lat >= tg->td->filtered_latency) { |
53696b8d2 blk-throttle: add... |
2153 2154 |
int bucket; unsigned int threshold; |
5238dcf41 block: replace bi... |
2155 |
bucket = request_bucket_index(bio_issue_size(&bio->bi_issue)); |
b889bf66d blk-throttle: tra... |
2156 |
threshold = tg->td->avg_buckets[rw][bucket].latency + |
53696b8d2 blk-throttle: add... |
2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 |
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 blk-throttle: add... |
2171 |
} |
9e234eeaf blk-throttle: add... |
2172 2173 |
} #endif |
2a12f0dcd blk-throttle: mak... |
2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 |
/* * Dispatch all bios from all children tg's queued on @parent_sq. On * return, @parent_sq is guaranteed to not have any active children tg's * and all bios from previously active tg's are on @parent_sq->bio_lists[]. */ static void tg_drain_bios(struct throtl_service_queue *parent_sq) { struct throtl_grp *tg; while ((tg = throtl_rb_first(parent_sq))) { struct throtl_service_queue *sq = &tg->service_queue; struct bio *bio; throtl_dequeue_tg(tg); |
c5cc2070b blk-throttle: add... |
2188 |
while ((bio = throtl_peek_queued(&sq->queued[READ]))) |
2a12f0dcd blk-throttle: mak... |
2189 |
tg_dispatch_one_bio(tg, bio_data_dir(bio)); |
c5cc2070b blk-throttle: add... |
2190 |
while ((bio = throtl_peek_queued(&sq->queued[WRITE]))) |
2a12f0dcd blk-throttle: mak... |
2191 2192 2193 |
tg_dispatch_one_bio(tg, bio_data_dir(bio)); } } |
c9a929dde block: fix reques... |
2194 2195 2196 2197 2198 2199 2200 |
/** * blk_throtl_drain - drain throttled bios * @q: request_queue to drain throttled bios for * * Dispatch all currently throttled bios on @q through ->make_request_fn(). */ void blk_throtl_drain(struct request_queue *q) |
0d945c1f9 block: remove the... |
2201 |
__releases(&q->queue_lock) __acquires(&q->queue_lock) |
c9a929dde block: fix reques... |
2202 2203 |
{ struct throtl_data *td = q->td; |
2a12f0dcd blk-throttle: mak... |
2204 |
struct blkcg_gq *blkg; |
492eb21b9 cgroup: make hier... |
2205 |
struct cgroup_subsys_state *pos_css; |
c9a929dde block: fix reques... |
2206 |
struct bio *bio; |
651930bc1 blk-throttle: dis... |
2207 |
int rw; |
c9a929dde block: fix reques... |
2208 |
|
2a12f0dcd blk-throttle: mak... |
2209 |
rcu_read_lock(); |
c9a929dde block: fix reques... |
2210 |
|
2a12f0dcd blk-throttle: mak... |
2211 2212 2213 2214 2215 2216 |
/* * Drain each tg while doing post-order walk on the blkg tree, so * that all bios are propagated to td->service_queue. It'd be * better to walk service_queue tree directly but blkg walk is * easier. */ |
492eb21b9 cgroup: make hier... |
2217 |
blkg_for_each_descendant_post(blkg, pos_css, td->queue->root_blkg) |
2a12f0dcd blk-throttle: mak... |
2218 |
tg_drain_bios(&blkg_to_tg(blkg)->service_queue); |
73f0d49a9 blk-throttle: mov... |
2219 |
|
2a12f0dcd blk-throttle: mak... |
2220 2221 2222 2223 |
/* finally, transfer bios from top-level tg's into the td */ tg_drain_bios(&td->service_queue); rcu_read_unlock(); |
0d945c1f9 block: remove the... |
2224 |
spin_unlock_irq(&q->queue_lock); |
c9a929dde block: fix reques... |
2225 |
|
2a12f0dcd blk-throttle: mak... |
2226 |
/* all bios now should be in td->service_queue, issue them */ |
651930bc1 blk-throttle: dis... |
2227 |
for (rw = READ; rw <= WRITE; rw++) |
c5cc2070b blk-throttle: add... |
2228 2229 |
while ((bio = throtl_pop_queued(&td->service_queue.queued[rw], NULL))) |
651930bc1 blk-throttle: dis... |
2230 |
generic_make_request(bio); |
c9a929dde block: fix reques... |
2231 |
|
0d945c1f9 block: remove the... |
2232 |
spin_lock_irq(&q->queue_lock); |
c9a929dde block: fix reques... |
2233 |
} |
e43473b7f blkio: Core imple... |
2234 2235 2236 |
int blk_throtl_init(struct request_queue *q) { struct throtl_data *td; |
a2b1693ba blkcg: implement ... |
2237 |
int ret; |
e43473b7f blkio: Core imple... |
2238 2239 2240 2241 |
td = kzalloc_node(sizeof(*td), GFP_KERNEL, q->node); if (!td) return -ENOMEM; |
b889bf66d blk-throttle: tra... |
2242 |
td->latency_buckets[READ] = __alloc_percpu(sizeof(struct latency_bucket) * |
b9147dd1b blk-throttle: add... |
2243 |
LATENCY_BUCKET_SIZE, __alignof__(u64)); |
b889bf66d blk-throttle: tra... |
2244 2245 2246 2247 2248 |
if (!td->latency_buckets[READ]) { kfree(td); return -ENOMEM; } td->latency_buckets[WRITE] = __alloc_percpu(sizeof(struct latency_bucket) * |
b9147dd1b blk-throttle: add... |
2249 |
LATENCY_BUCKET_SIZE, __alignof__(u64)); |
b889bf66d blk-throttle: tra... |
2250 2251 |
if (!td->latency_buckets[WRITE]) { free_percpu(td->latency_buckets[READ]); |
b9147dd1b blk-throttle: add... |
2252 2253 2254 |
kfree(td); return -ENOMEM; } |
e43473b7f blkio: Core imple... |
2255 |
|
69df0ab03 blk-throttle: sep... |
2256 |
INIT_WORK(&td->dispatch_work, blk_throtl_dispatch_work_fn); |
b2ce2643c blk-throttle: cle... |
2257 |
throtl_service_queue_init(&td->service_queue); |
e43473b7f blkio: Core imple... |
2258 |
|
cd1604fab blkcg: factor out... |
2259 |
q->td = td; |
29b125892 blk-throttle: Dyn... |
2260 |
td->queue = q; |
02977e4af blkio: Add root g... |
2261 |
|
9f626e372 blk-throttle: pre... |
2262 |
td->limit_valid[LIMIT_MAX] = true; |
cd5ab1b0f blk-throttle: add... |
2263 |
td->limit_index = LIMIT_MAX; |
3f0abd806 blk-throttle: add... |
2264 2265 |
td->low_upgrade_time = jiffies; td->low_downgrade_time = jiffies; |
9e234eeaf blk-throttle: add... |
2266 |
|
a2b1693ba blkcg: implement ... |
2267 |
/* activate policy */ |
3c798398e blkcg: mass renam... |
2268 |
ret = blkcg_activate_policy(q, &blkcg_policy_throtl); |
b9147dd1b blk-throttle: add... |
2269 |
if (ret) { |
b889bf66d blk-throttle: tra... |
2270 2271 |
free_percpu(td->latency_buckets[READ]); free_percpu(td->latency_buckets[WRITE]); |
f51b802c1 blkcg: use the us... |
2272 |
kfree(td); |
b9147dd1b blk-throttle: add... |
2273 |
} |
a2b1693ba blkcg: implement ... |
2274 |
return ret; |
e43473b7f blkio: Core imple... |
2275 2276 2277 2278 |
} void blk_throtl_exit(struct request_queue *q) { |
c875f4d02 blkcg: drop unnec... |
2279 |
BUG_ON(!q->td); |
da5277700 block: Move blk_t... |
2280 |
throtl_shutdown_wq(q); |
3c798398e blkcg: mass renam... |
2281 |
blkcg_deactivate_policy(q, &blkcg_policy_throtl); |
b889bf66d blk-throttle: tra... |
2282 2283 |
free_percpu(q->td->latency_buckets[READ]); free_percpu(q->td->latency_buckets[WRITE]); |
c9a929dde block: fix reques... |
2284 |
kfree(q->td); |
e43473b7f blkio: Core imple... |
2285 |
} |
d61fcfa4b blk-throttle: cho... |
2286 2287 2288 |
void blk_throtl_register_queue(struct request_queue *q) { struct throtl_data *td; |
6679a90c4 blk-throttle: set... |
2289 |
int i; |
d61fcfa4b blk-throttle: cho... |
2290 2291 2292 |
td = q->td; BUG_ON(!td); |
6679a90c4 blk-throttle: set... |
2293 |
if (blk_queue_nonrot(q)) { |
d61fcfa4b blk-throttle: cho... |
2294 |
td->throtl_slice = DFL_THROTL_SLICE_SSD; |
6679a90c4 blk-throttle: set... |
2295 2296 |
td->filtered_latency = LATENCY_FILTERED_SSD; } else { |
d61fcfa4b blk-throttle: cho... |
2297 |
td->throtl_slice = DFL_THROTL_SLICE_HD; |
6679a90c4 blk-throttle: set... |
2298 |
td->filtered_latency = LATENCY_FILTERED_HD; |
b889bf66d blk-throttle: tra... |
2299 2300 2301 2302 |
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 blk-throttle: set... |
2303 |
} |
d61fcfa4b blk-throttle: cho... |
2304 2305 2306 2307 |
#ifndef CONFIG_BLK_DEV_THROTTLING_LOW /* if no low limit, use previous default */ td->throtl_slice = DFL_THROTL_SLICE_HD; #endif |
9e234eeaf blk-throttle: add... |
2308 |
|
344e9ffcb block: add queue_... |
2309 |
td->track_bio_latency = !queue_is_mq(q); |
b9147dd1b blk-throttle: add... |
2310 2311 |
if (!td->track_bio_latency) blk_stat_enable_accounting(q); |
d61fcfa4b blk-throttle: cho... |
2312 |
} |
297e3d854 blk-throttle: mak... |
2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 |
#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 blkio: Core imple... |
2339 2340 |
static int __init throtl_init(void) { |
450adcbe5 blk-throttle: Do ... |
2341 2342 2343 2344 |
kthrotld_workqueue = alloc_workqueue("kthrotld", WQ_MEM_RECLAIM, 0); if (!kthrotld_workqueue) panic("Failed to create kthrotld "); |
3c798398e blkcg: mass renam... |
2345 |
return blkcg_policy_register(&blkcg_policy_throtl); |
e43473b7f blkio: Core imple... |
2346 2347 2348 |
} module_init(throtl_init); |