Blame view
block/blk-throttle.c
42.7 KB
e43473b7f
|
1 2 3 4 5 6 7 8 9 10 11 |
/* * 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
|
12 |
#include <linux/blk-cgroup.h> |
bc9fcbf9c
|
13 |
#include "blk.h" |
e43473b7f
|
14 15 16 17 18 19 20 21 22 |
/* 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; /* Throttling is performed over 100ms slice and after that slice is renewed */ static unsigned long throtl_slice = HZ/10; /* 100 ms */ |
3c798398e
|
23 |
static struct blkcg_policy blkcg_policy_throtl; |
0381411e4
|
24 |
|
450adcbe5
|
25 26 |
/* A workqueue to queue throttle related work */ static struct workqueue_struct *kthrotld_workqueue; |
450adcbe5
|
27 |
|
c5cc2070b
|
28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 |
/* * 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
|
56 |
struct throtl_service_queue { |
77216b048
|
57 |
struct throtl_service_queue *parent_sq; /* the parent service_queue */ |
73f0d49a9
|
58 59 60 61 |
/* * Bios queued directly to this service_queue or dispatched from * children throtl_grp's. */ |
c5cc2070b
|
62 |
struct list_head queued[2]; /* throtl_qnode [READ/WRITE] */ |
73f0d49a9
|
63 64 65 66 67 68 |
unsigned int nr_queued[2]; /* number of queued bios */ /* * RB tree of active children throtl_grp's, which are sorted by * their ->disptime. */ |
c9e0332e8
|
69 70 71 72 |
struct rb_root pending_tree; /* RB tree of active tgs */ struct rb_node *first_pending; /* first node in the tree */ unsigned int nr_pending; /* # queued in the tree */ unsigned long first_pending_disptime; /* disptime of the first tg */ |
69df0ab03
|
73 |
struct timer_list pending_timer; /* fires on first_pending_disptime */ |
e43473b7f
|
74 |
}; |
5b2c16aae
|
75 76 |
enum tg_state_flags { THROTL_TG_PENDING = 1 << 0, /* on parent's pending tree */ |
0e9f4164b
|
77 |
THROTL_TG_WAS_EMPTY = 1 << 1, /* bio_lists[] became non-empty */ |
5b2c16aae
|
78 |
}; |
e43473b7f
|
79 80 81 |
#define rb_entry_tg(node) rb_entry((node), struct throtl_grp, rb_node) struct throtl_grp { |
f95a04afa
|
82 83 |
/* must be the first member */ struct blkg_policy_data pd; |
c9e0332e8
|
84 |
/* active throtl group service_queue member */ |
e43473b7f
|
85 |
struct rb_node rb_node; |
0f3457f60
|
86 87 |
/* throtl_data this group belongs to */ struct throtl_data *td; |
49a2f1e3f
|
88 89 |
/* this group's service queue */ struct throtl_service_queue service_queue; |
e43473b7f
|
90 |
/* |
c5cc2070b
|
91 92 93 94 95 96 97 98 99 100 101 |
* 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
|
102 103 104 105 106 |
* 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
|
107 |
unsigned int flags; |
693e751e7
|
108 109 |
/* are there any throtl rules between this group and td? */ bool has_rules[2]; |
e43473b7f
|
110 111 |
/* bytes per second rate limits */ uint64_t bps[2]; |
8e89d13f4
|
112 113 |
/* IOPS limits */ unsigned int iops[2]; |
e43473b7f
|
114 115 |
/* Number of bytes disptached in current slice */ uint64_t bytes_disp[2]; |
8e89d13f4
|
116 117 |
/* Number of bio's dispatched in current slice */ unsigned int io_disp[2]; |
e43473b7f
|
118 119 120 121 122 123 124 125 |
/* When did we start a new slice */ unsigned long slice_start[2]; unsigned long slice_end[2]; }; struct throtl_data { |
e43473b7f
|
126 |
/* service tree for active throtl groups */ |
c9e0332e8
|
127 |
struct throtl_service_queue service_queue; |
e43473b7f
|
128 |
|
e43473b7f
|
129 130 131 132 |
struct request_queue *queue; /* Total Number of queued bios on READ and WRITE lists */ unsigned int nr_queued[2]; |
e43473b7f
|
133 |
/* Work for dispatching throttled bios */ |
69df0ab03
|
134 |
struct work_struct dispatch_work; |
e43473b7f
|
135 |
}; |
69df0ab03
|
136 |
static void throtl_pending_timer_fn(unsigned long arg); |
f95a04afa
|
137 138 139 140 |
static inline struct throtl_grp *pd_to_tg(struct blkg_policy_data *pd) { return pd ? container_of(pd, struct throtl_grp, pd) : NULL; } |
3c798398e
|
141 |
static inline struct throtl_grp *blkg_to_tg(struct blkcg_gq *blkg) |
0381411e4
|
142 |
{ |
f95a04afa
|
143 |
return pd_to_tg(blkg_to_pd(blkg, &blkcg_policy_throtl)); |
0381411e4
|
144 |
} |
3c798398e
|
145 |
static inline struct blkcg_gq *tg_to_blkg(struct throtl_grp *tg) |
0381411e4
|
146 |
{ |
f95a04afa
|
147 |
return pd_to_blkg(&tg->pd); |
0381411e4
|
148 |
} |
fda6f272c
|
149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 |
/** * 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 * * A service_queue can be embeded in either a throtl_grp or throtl_data. * 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); } /** * 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
|
189 190 191 192 193 194 |
*/ #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
|
195 196 |
if (likely(!blk_trace_note_message_enabled(__td->queue))) \ break; \ |
fda6f272c
|
197 198 |
if ((__tg)) { \ char __pbuf[128]; \ |
54e7ed12b
|
199 |
\ |
fda6f272c
|
200 201 202 203 204 |
blkg_path(tg_to_blkg(__tg), __pbuf, sizeof(__pbuf)); \ blk_add_trace_msg(__td->queue, "throtl %s " fmt, __pbuf, ##args); \ } else { \ blk_add_trace_msg(__td->queue, "throtl " fmt, ##args); \ } \ |
54e7ed12b
|
205 |
} while (0) |
e43473b7f
|
206 |
|
c5cc2070b
|
207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 |
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
|
289 |
/* init a service_queue, assumes the caller zeroed it */ |
b2ce2643c
|
290 |
static void throtl_service_queue_init(struct throtl_service_queue *sq) |
49a2f1e3f
|
291 |
{ |
c5cc2070b
|
292 293 |
INIT_LIST_HEAD(&sq->queued[0]); INIT_LIST_HEAD(&sq->queued[1]); |
49a2f1e3f
|
294 |
sq->pending_tree = RB_ROOT; |
69df0ab03
|
295 296 297 |
setup_timer(&sq->pending_timer, throtl_pending_timer_fn, (unsigned long)sq); } |
001bea73e
|
298 299 |
static struct blkg_policy_data *throtl_pd_alloc(gfp_t gfp, int node) { |
4fb72036f
|
300 |
struct throtl_grp *tg; |
24bdb8ef0
|
301 |
int rw; |
4fb72036f
|
302 303 304 |
tg = kzalloc_node(sizeof(*tg), gfp, node); if (!tg) |
77ea73388
|
305 |
return NULL; |
4fb72036f
|
306 |
|
b2ce2643c
|
307 308 309 310 311 312 313 314 315 316 317 318 |
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); tg->bps[READ] = -1; tg->bps[WRITE] = -1; tg->iops[READ] = -1; tg->iops[WRITE] = -1; |
4fb72036f
|
319 |
return &tg->pd; |
001bea73e
|
320 |
} |
a9520cd6f
|
321 |
static void throtl_pd_init(struct blkg_policy_data *pd) |
a29a171e7
|
322 |
{ |
a9520cd6f
|
323 324 |
struct throtl_grp *tg = pd_to_tg(pd); struct blkcg_gq *blkg = tg_to_blkg(tg); |
77216b048
|
325 |
struct throtl_data *td = blkg->q->td; |
b2ce2643c
|
326 |
struct throtl_service_queue *sq = &tg->service_queue; |
cd1604fab
|
327 |
|
9138125be
|
328 |
/* |
aa6ec29be
|
329 |
* If on the default hierarchy, we switch to properly hierarchical |
9138125be
|
330 331 332 333 334 |
* 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
|
335 |
* If not on the default hierarchy, the broken flat hierarchy |
9138125be
|
336 337 338 339 340 |
* 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
|
341 |
sq->parent_sq = &td->service_queue; |
9e10a130d
|
342 |
if (cgroup_subsys_on_dfl(io_cgrp_subsys) && blkg->parent) |
b2ce2643c
|
343 |
sq->parent_sq = &blkg_to_tg(blkg->parent)->service_queue; |
77216b048
|
344 |
tg->td = td; |
8a3d26151
|
345 |
} |
693e751e7
|
346 347 348 349 350 351 352 353 354 355 356 357 358 359 |
/* * 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); int rw; for (rw = READ; rw <= WRITE; rw++) tg->has_rules[rw] = (parent_tg && parent_tg->has_rules[rw]) || (tg->bps[rw] != -1 || tg->iops[rw] != -1); } |
a9520cd6f
|
360 |
static void throtl_pd_online(struct blkg_policy_data *pd) |
693e751e7
|
361 362 363 364 365 |
{ /* * We don't want new groups to escape the limits of its ancestors. * Update has_rules[] after a new group is brought online. */ |
a9520cd6f
|
366 |
tg_update_has_rules(pd_to_tg(pd)); |
693e751e7
|
367 |
} |
001bea73e
|
368 369 |
static void throtl_pd_free(struct blkg_policy_data *pd) { |
4fb72036f
|
370 |
struct throtl_grp *tg = pd_to_tg(pd); |
b2ce2643c
|
371 |
del_timer_sync(&tg->service_queue.pending_timer); |
4fb72036f
|
372 |
kfree(tg); |
001bea73e
|
373 |
} |
0049af73b
|
374 375 |
static struct throtl_grp * throtl_rb_first(struct throtl_service_queue *parent_sq) |
e43473b7f
|
376 377 |
{ /* Service tree is empty */ |
0049af73b
|
378 |
if (!parent_sq->nr_pending) |
e43473b7f
|
379 |
return NULL; |
0049af73b
|
380 381 |
if (!parent_sq->first_pending) parent_sq->first_pending = rb_first(&parent_sq->pending_tree); |
e43473b7f
|
382 |
|
0049af73b
|
383 384 |
if (parent_sq->first_pending) return rb_entry_tg(parent_sq->first_pending); |
e43473b7f
|
385 386 387 388 389 390 391 392 393 |
return NULL; } static void rb_erase_init(struct rb_node *n, struct rb_root *root) { rb_erase(n, root); RB_CLEAR_NODE(n); } |
0049af73b
|
394 395 |
static void throtl_rb_erase(struct rb_node *n, struct throtl_service_queue *parent_sq) |
e43473b7f
|
396 |
{ |
0049af73b
|
397 398 399 400 |
if (parent_sq->first_pending == n) parent_sq->first_pending = NULL; rb_erase_init(n, &parent_sq->pending_tree); --parent_sq->nr_pending; |
e43473b7f
|
401 |
} |
0049af73b
|
402 |
static void update_min_dispatch_time(struct throtl_service_queue *parent_sq) |
e43473b7f
|
403 404 |
{ struct throtl_grp *tg; |
0049af73b
|
405 |
tg = throtl_rb_first(parent_sq); |
e43473b7f
|
406 407 |
if (!tg) return; |
0049af73b
|
408 |
parent_sq->first_pending_disptime = tg->disptime; |
e43473b7f
|
409 |
} |
77216b048
|
410 |
static void tg_service_queue_add(struct throtl_grp *tg) |
e43473b7f
|
411 |
{ |
77216b048
|
412 |
struct throtl_service_queue *parent_sq = tg->service_queue.parent_sq; |
0049af73b
|
413 |
struct rb_node **node = &parent_sq->pending_tree.rb_node; |
e43473b7f
|
414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 |
struct rb_node *parent = NULL; struct throtl_grp *__tg; unsigned long key = tg->disptime; int left = 1; 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; left = 0; } } if (left) |
0049af73b
|
432 |
parent_sq->first_pending = &tg->rb_node; |
e43473b7f
|
433 434 |
rb_link_node(&tg->rb_node, parent, node); |
0049af73b
|
435 |
rb_insert_color(&tg->rb_node, &parent_sq->pending_tree); |
e43473b7f
|
436 |
} |
77216b048
|
437 |
static void __throtl_enqueue_tg(struct throtl_grp *tg) |
e43473b7f
|
438 |
{ |
77216b048
|
439 |
tg_service_queue_add(tg); |
5b2c16aae
|
440 |
tg->flags |= THROTL_TG_PENDING; |
77216b048
|
441 |
tg->service_queue.parent_sq->nr_pending++; |
e43473b7f
|
442 |
} |
77216b048
|
443 |
static void throtl_enqueue_tg(struct throtl_grp *tg) |
e43473b7f
|
444 |
{ |
5b2c16aae
|
445 |
if (!(tg->flags & THROTL_TG_PENDING)) |
77216b048
|
446 |
__throtl_enqueue_tg(tg); |
e43473b7f
|
447 |
} |
77216b048
|
448 |
static void __throtl_dequeue_tg(struct throtl_grp *tg) |
e43473b7f
|
449 |
{ |
77216b048
|
450 |
throtl_rb_erase(&tg->rb_node, tg->service_queue.parent_sq); |
5b2c16aae
|
451 |
tg->flags &= ~THROTL_TG_PENDING; |
e43473b7f
|
452 |
} |
77216b048
|
453 |
static void throtl_dequeue_tg(struct throtl_grp *tg) |
e43473b7f
|
454 |
{ |
5b2c16aae
|
455 |
if (tg->flags & THROTL_TG_PENDING) |
77216b048
|
456 |
__throtl_dequeue_tg(tg); |
e43473b7f
|
457 |
} |
a9131a27e
|
458 |
/* Call with queue lock held */ |
69df0ab03
|
459 460 |
static void throtl_schedule_pending_timer(struct throtl_service_queue *sq, unsigned long expires) |
a9131a27e
|
461 |
{ |
69df0ab03
|
462 463 464 |
mod_timer(&sq->pending_timer, expires); throtl_log(sq, "schedule timer. delay=%lu jiffies=%lu", expires - jiffies, jiffies); |
a9131a27e
|
465 |
} |
7f52f98c2
|
466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 |
/** * 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
|
486 |
{ |
6a525600f
|
487 |
/* any pending children left? */ |
c9e0332e8
|
488 |
if (!sq->nr_pending) |
7f52f98c2
|
489 |
return true; |
e43473b7f
|
490 |
|
c9e0332e8
|
491 |
update_min_dispatch_time(sq); |
e43473b7f
|
492 |
|
69df0ab03
|
493 |
/* is the next dispatch time in the future? */ |
7f52f98c2
|
494 |
if (force || time_after(sq->first_pending_disptime, jiffies)) { |
69df0ab03
|
495 |
throtl_schedule_pending_timer(sq, sq->first_pending_disptime); |
7f52f98c2
|
496 |
return true; |
69df0ab03
|
497 |
} |
7f52f98c2
|
498 499 |
/* tell the caller to continue dispatching */ return false; |
e43473b7f
|
500 |
} |
32ee5bc47
|
501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 |
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; tg->slice_end[rw] = jiffies + throtl_slice; 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
|
522 |
static inline void throtl_start_new_slice(struct throtl_grp *tg, bool rw) |
e43473b7f
|
523 524 |
{ tg->bytes_disp[rw] = 0; |
8e89d13f4
|
525 |
tg->io_disp[rw] = 0; |
e43473b7f
|
526 527 |
tg->slice_start[rw] = jiffies; tg->slice_end[rw] = jiffies + throtl_slice; |
fda6f272c
|
528 529 530 531 |
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
|
532 |
} |
0f3457f60
|
533 534 |
static inline void throtl_set_slice_end(struct throtl_grp *tg, bool rw, unsigned long jiffy_end) |
d1ae8ffdf
|
535 536 537 |
{ tg->slice_end[rw] = roundup(jiffy_end, throtl_slice); } |
0f3457f60
|
538 539 |
static inline void throtl_extend_slice(struct throtl_grp *tg, bool rw, unsigned long jiffy_end) |
e43473b7f
|
540 541 |
{ tg->slice_end[rw] = roundup(jiffy_end, throtl_slice); |
fda6f272c
|
542 543 544 545 |
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
|
546 547 548 |
} /* Determine if previously allocated or extended slice is complete or not */ |
0f3457f60
|
549 |
static bool throtl_slice_used(struct throtl_grp *tg, bool rw) |
e43473b7f
|
550 551 |
{ if (time_in_range(jiffies, tg->slice_start[rw], tg->slice_end[rw])) |
5cf8c2277
|
552 |
return false; |
e43473b7f
|
553 554 555 556 557 |
return 1; } /* Trim the used slices and adjust slice start accordingly */ |
0f3457f60
|
558 |
static inline void throtl_trim_slice(struct throtl_grp *tg, bool rw) |
e43473b7f
|
559 |
{ |
3aad5d3ee
|
560 561 |
unsigned long nr_slices, time_elapsed, io_trim; u64 bytes_trim, tmp; |
e43473b7f
|
562 563 564 565 566 567 568 569 |
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
|
570 |
if (throtl_slice_used(tg, rw)) |
e43473b7f
|
571 |
return; |
d1ae8ffdf
|
572 573 574 575 576 577 578 |
/* * 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. */ |
0f3457f60
|
579 |
throtl_set_slice_end(tg, rw, jiffies + throtl_slice); |
d1ae8ffdf
|
580 |
|
e43473b7f
|
581 582 583 584 585 586 |
time_elapsed = jiffies - tg->slice_start[rw]; nr_slices = time_elapsed / throtl_slice; if (!nr_slices) return; |
3aad5d3ee
|
587 588 589 |
tmp = tg->bps[rw] * throtl_slice * nr_slices; do_div(tmp, HZ); bytes_trim = tmp; |
e43473b7f
|
590 |
|
8e89d13f4
|
591 |
io_trim = (tg->iops[rw] * throtl_slice * nr_slices)/HZ; |
e43473b7f
|
592 |
|
8e89d13f4
|
593 |
if (!bytes_trim && !io_trim) |
e43473b7f
|
594 595 596 597 598 599 |
return; if (tg->bytes_disp[rw] >= bytes_trim) tg->bytes_disp[rw] -= bytes_trim; else tg->bytes_disp[rw] = 0; |
8e89d13f4
|
600 601 602 603 |
if (tg->io_disp[rw] >= io_trim) tg->io_disp[rw] -= io_trim; else tg->io_disp[rw] = 0; |
e43473b7f
|
604 |
tg->slice_start[rw] += nr_slices * throtl_slice; |
fda6f272c
|
605 606 607 608 |
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
|
609 |
} |
0f3457f60
|
610 611 |
static bool tg_with_in_iops_limit(struct throtl_grp *tg, struct bio *bio, unsigned long *wait) |
e43473b7f
|
612 613 |
{ bool rw = bio_data_dir(bio); |
8e89d13f4
|
614 |
unsigned int io_allowed; |
e43473b7f
|
615 |
unsigned long jiffy_elapsed, jiffy_wait, jiffy_elapsed_rnd; |
c49c06e49
|
616 |
u64 tmp; |
e43473b7f
|
617 |
|
8e89d13f4
|
618 |
jiffy_elapsed = jiffy_elapsed_rnd = jiffies - tg->slice_start[rw]; |
e43473b7f
|
619 |
|
8e89d13f4
|
620 621 622 623 624 |
/* Slice has just started. Consider one slice interval */ if (!jiffy_elapsed) jiffy_elapsed_rnd = throtl_slice; jiffy_elapsed_rnd = roundup(jiffy_elapsed_rnd, throtl_slice); |
c49c06e49
|
625 626 627 628 629 630 631 632 633 634 635 636 637 638 |
/* * 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. */ tmp = (u64)tg->iops[rw] * jiffy_elapsed_rnd; do_div(tmp, HZ); if (tmp > UINT_MAX) io_allowed = UINT_MAX; else io_allowed = tmp; |
8e89d13f4
|
639 640 |
if (tg->io_disp[rw] + 1 <= io_allowed) { |
e43473b7f
|
641 642 |
if (wait) *wait = 0; |
5cf8c2277
|
643 |
return true; |
e43473b7f
|
644 |
} |
8e89d13f4
|
645 646 647 648 649 650 651 652 653 654 655 656 |
/* Calc approx time to dispatch */ jiffy_wait = ((tg->io_disp[rw] + 1) * HZ)/tg->iops[rw] + 1; if (jiffy_wait > jiffy_elapsed) jiffy_wait = jiffy_wait - jiffy_elapsed; else jiffy_wait = 1; if (wait) *wait = jiffy_wait; return 0; } |
0f3457f60
|
657 658 |
static bool tg_with_in_bps_limit(struct throtl_grp *tg, struct bio *bio, unsigned long *wait) |
8e89d13f4
|
659 660 |
{ bool rw = bio_data_dir(bio); |
3aad5d3ee
|
661 |
u64 bytes_allowed, extra_bytes, tmp; |
8e89d13f4
|
662 |
unsigned long jiffy_elapsed, jiffy_wait, jiffy_elapsed_rnd; |
e43473b7f
|
663 664 665 666 667 668 669 670 |
jiffy_elapsed = jiffy_elapsed_rnd = jiffies - tg->slice_start[rw]; /* Slice has just started. Consider one slice interval */ if (!jiffy_elapsed) jiffy_elapsed_rnd = throtl_slice; jiffy_elapsed_rnd = roundup(jiffy_elapsed_rnd, throtl_slice); |
5e901a2b9
|
671 672 |
tmp = tg->bps[rw] * jiffy_elapsed_rnd; do_div(tmp, HZ); |
3aad5d3ee
|
673 |
bytes_allowed = tmp; |
e43473b7f
|
674 |
|
4f024f379
|
675 |
if (tg->bytes_disp[rw] + bio->bi_iter.bi_size <= bytes_allowed) { |
e43473b7f
|
676 677 |
if (wait) *wait = 0; |
5cf8c2277
|
678 |
return true; |
e43473b7f
|
679 680 681 |
} /* Calc approx time to dispatch */ |
4f024f379
|
682 |
extra_bytes = tg->bytes_disp[rw] + bio->bi_iter.bi_size - bytes_allowed; |
e43473b7f
|
683 684 685 686 687 688 689 690 691 692 |
jiffy_wait = div64_u64(extra_bytes * HZ, tg->bps[rw]); 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
|
693 694 |
if (wait) *wait = jiffy_wait; |
8e89d13f4
|
695 696 697 698 699 700 701 |
return 0; } /* * 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
|
702 703 |
static bool tg_may_dispatch(struct throtl_grp *tg, struct bio *bio, unsigned long *wait) |
8e89d13f4
|
704 705 706 707 708 709 710 711 712 713 |
{ 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
|
714 |
BUG_ON(tg->service_queue.nr_queued[rw] && |
c5cc2070b
|
715 |
bio != throtl_peek_queued(&tg->service_queue.queued[rw])); |
e43473b7f
|
716 |
|
8e89d13f4
|
717 718 719 720 |
/* If tg->bps = -1, then BW is unlimited */ if (tg->bps[rw] == -1 && tg->iops[rw] == -1) { if (wait) *wait = 0; |
5cf8c2277
|
721 |
return true; |
8e89d13f4
|
722 723 724 725 726 |
} /* * If previous slice expired, start a new one otherwise renew/extend * existing slice to make sure it is at least throtl_slice interval |
164c80ed8
|
727 728 729 |
* 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
|
730 |
*/ |
164c80ed8
|
731 |
if (throtl_slice_used(tg, rw) && !(tg->service_queue.nr_queued[rw])) |
0f3457f60
|
732 |
throtl_start_new_slice(tg, rw); |
8e89d13f4
|
733 734 |
else { if (time_before(tg->slice_end[rw], jiffies + throtl_slice)) |
0f3457f60
|
735 |
throtl_extend_slice(tg, rw, jiffies + throtl_slice); |
8e89d13f4
|
736 |
} |
0f3457f60
|
737 738 |
if (tg_with_in_bps_limit(tg, bio, &bps_wait) && tg_with_in_iops_limit(tg, bio, &iops_wait)) { |
8e89d13f4
|
739 740 741 742 743 744 745 746 747 748 749 |
if (wait) *wait = 0; return 1; } max_wait = max(bps_wait, iops_wait); if (wait) *wait = max_wait; if (time_before(tg->slice_end[rw], jiffies + max_wait)) |
0f3457f60
|
750 |
throtl_extend_slice(tg, rw, jiffies + max_wait); |
e43473b7f
|
751 752 753 754 755 756 757 |
return 0; } static void throtl_charge_bio(struct throtl_grp *tg, struct bio *bio) { bool rw = bio_data_dir(bio); |
e43473b7f
|
758 759 |
/* Charge the bio to the group */ |
4f024f379
|
760 |
tg->bytes_disp[rw] += bio->bi_iter.bi_size; |
8e89d13f4
|
761 |
tg->io_disp[rw]++; |
e43473b7f
|
762 |
|
2a0f61e6e
|
763 764 765 766 767 |
/* * REQ_THROTTLED is used to prevent the same bio to be throttled * 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
|
768 |
*/ |
1eff9d322
|
769 770 |
if (!(bio->bi_opf & REQ_THROTTLED)) bio->bi_opf |= REQ_THROTTLED; |
e43473b7f
|
771 |
} |
c5cc2070b
|
772 773 774 775 776 777 778 779 780 781 782 |
/** * 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
|
783 |
{ |
73f0d49a9
|
784 |
struct throtl_service_queue *sq = &tg->service_queue; |
e43473b7f
|
785 |
bool rw = bio_data_dir(bio); |
c5cc2070b
|
786 787 |
if (!qn) qn = &tg->qnode_on_self[rw]; |
0e9f4164b
|
788 789 790 791 792 793 794 795 |
/* * 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
|
796 |
throtl_qnode_add_bio(bio, qn, &sq->queued[rw]); |
73f0d49a9
|
797 |
sq->nr_queued[rw]++; |
77216b048
|
798 |
throtl_enqueue_tg(tg); |
e43473b7f
|
799 |
} |
77216b048
|
800 |
static void tg_update_disptime(struct throtl_grp *tg) |
e43473b7f
|
801 |
{ |
73f0d49a9
|
802 |
struct throtl_service_queue *sq = &tg->service_queue; |
e43473b7f
|
803 804 |
unsigned long read_wait = -1, write_wait = -1, min_wait = -1, disptime; struct bio *bio; |
c5cc2070b
|
805 |
if ((bio = throtl_peek_queued(&sq->queued[READ]))) |
0f3457f60
|
806 |
tg_may_dispatch(tg, bio, &read_wait); |
e43473b7f
|
807 |
|
c5cc2070b
|
808 |
if ((bio = throtl_peek_queued(&sq->queued[WRITE]))) |
0f3457f60
|
809 |
tg_may_dispatch(tg, bio, &write_wait); |
e43473b7f
|
810 811 812 |
min_wait = min(read_wait, write_wait); disptime = jiffies + min_wait; |
e43473b7f
|
813 |
/* Update dispatch time */ |
77216b048
|
814 |
throtl_dequeue_tg(tg); |
e43473b7f
|
815 |
tg->disptime = disptime; |
77216b048
|
816 |
throtl_enqueue_tg(tg); |
0e9f4164b
|
817 818 819 |
/* see throtl_add_bio_tg() */ tg->flags &= ~THROTL_TG_WAS_EMPTY; |
e43473b7f
|
820 |
} |
32ee5bc47
|
821 822 823 824 825 826 827 828 829 |
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
|
830 |
static void tg_dispatch_one_bio(struct throtl_grp *tg, bool rw) |
e43473b7f
|
831 |
{ |
73f0d49a9
|
832 |
struct throtl_service_queue *sq = &tg->service_queue; |
6bc9c2b46
|
833 834 |
struct throtl_service_queue *parent_sq = sq->parent_sq; struct throtl_grp *parent_tg = sq_to_tg(parent_sq); |
c5cc2070b
|
835 |
struct throtl_grp *tg_to_put = NULL; |
e43473b7f
|
836 |
struct bio *bio; |
c5cc2070b
|
837 838 839 840 841 842 843 |
/* * @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
|
844 |
sq->nr_queued[rw]--; |
e43473b7f
|
845 846 |
throtl_charge_bio(tg, bio); |
6bc9c2b46
|
847 848 849 850 851 852 853 854 855 |
/* * 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
|
856 |
throtl_add_bio_tg(bio, &tg->qnode_on_parent[rw], parent_tg); |
32ee5bc47
|
857 |
start_parent_slice_with_credit(tg, parent_tg, rw); |
6bc9c2b46
|
858 |
} else { |
c5cc2070b
|
859 860 |
throtl_qnode_add_bio(bio, &tg->qnode_on_parent[rw], &parent_sq->queued[rw]); |
6bc9c2b46
|
861 862 863 |
BUG_ON(tg->td->nr_queued[rw] <= 0); tg->td->nr_queued[rw]--; } |
e43473b7f
|
864 |
|
0f3457f60
|
865 |
throtl_trim_slice(tg, rw); |
6bc9c2b46
|
866 |
|
c5cc2070b
|
867 868 |
if (tg_to_put) blkg_put(tg_to_blkg(tg_to_put)); |
e43473b7f
|
869 |
} |
77216b048
|
870 |
static int throtl_dispatch_tg(struct throtl_grp *tg) |
e43473b7f
|
871 |
{ |
73f0d49a9
|
872 |
struct throtl_service_queue *sq = &tg->service_queue; |
e43473b7f
|
873 874 |
unsigned int nr_reads = 0, nr_writes = 0; unsigned int max_nr_reads = throtl_grp_quantum*3/4; |
c2f6805d4
|
875 |
unsigned int max_nr_writes = throtl_grp_quantum - max_nr_reads; |
e43473b7f
|
876 877 878 |
struct bio *bio; /* Try to dispatch 75% READS and 25% WRITES */ |
c5cc2070b
|
879 |
while ((bio = throtl_peek_queued(&sq->queued[READ])) && |
0f3457f60
|
880 |
tg_may_dispatch(tg, bio, NULL)) { |
e43473b7f
|
881 |
|
77216b048
|
882 |
tg_dispatch_one_bio(tg, bio_data_dir(bio)); |
e43473b7f
|
883 884 885 886 887 |
nr_reads++; if (nr_reads >= max_nr_reads) break; } |
c5cc2070b
|
888 |
while ((bio = throtl_peek_queued(&sq->queued[WRITE])) && |
0f3457f60
|
889 |
tg_may_dispatch(tg, bio, NULL)) { |
e43473b7f
|
890 |
|
77216b048
|
891 |
tg_dispatch_one_bio(tg, bio_data_dir(bio)); |
e43473b7f
|
892 893 894 895 896 897 898 899 |
nr_writes++; if (nr_writes >= max_nr_writes) break; } return nr_reads + nr_writes; } |
651930bc1
|
900 |
static int throtl_select_dispatch(struct throtl_service_queue *parent_sq) |
e43473b7f
|
901 902 |
{ unsigned int nr_disp = 0; |
e43473b7f
|
903 904 |
while (1) { |
73f0d49a9
|
905 906 |
struct throtl_grp *tg = throtl_rb_first(parent_sq); struct throtl_service_queue *sq = &tg->service_queue; |
e43473b7f
|
907 908 909 910 911 912 |
if (!tg) break; if (time_before(jiffies, tg->disptime)) break; |
77216b048
|
913 |
throtl_dequeue_tg(tg); |
e43473b7f
|
914 |
|
77216b048
|
915 |
nr_disp += throtl_dispatch_tg(tg); |
e43473b7f
|
916 |
|
73f0d49a9
|
917 |
if (sq->nr_queued[0] || sq->nr_queued[1]) |
77216b048
|
918 |
tg_update_disptime(tg); |
e43473b7f
|
919 920 921 922 923 924 925 |
if (nr_disp >= throtl_quantum) break; } return nr_disp; } |
6e1a5704c
|
926 927 928 929 930 931 932 |
/** * throtl_pending_timer_fn - timer function for service_queue->pending_timer * @arg: the throtl_service_queue being serviced * * 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
|
933 934 935 936 937 938 939 |
* 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
|
940 |
*/ |
69df0ab03
|
941 942 943 |
static void throtl_pending_timer_fn(unsigned long arg) { struct throtl_service_queue *sq = (void *)arg; |
2e48a530a
|
944 |
struct throtl_grp *tg = sq_to_tg(sq); |
69df0ab03
|
945 |
struct throtl_data *td = sq_to_td(sq); |
cb76199c3
|
946 |
struct request_queue *q = td->queue; |
2e48a530a
|
947 948 |
struct throtl_service_queue *parent_sq; bool dispatched; |
6e1a5704c
|
949 |
int ret; |
e43473b7f
|
950 951 |
spin_lock_irq(q->queue_lock); |
2e48a530a
|
952 953 954 |
again: parent_sq = sq->parent_sq; dispatched = false; |
e43473b7f
|
955 |
|
7f52f98c2
|
956 957 |
while (true) { throtl_log(sq, "dispatch nr_queued=%u read=%u write=%u", |
2e48a530a
|
958 959 |
sq->nr_queued[READ] + sq->nr_queued[WRITE], sq->nr_queued[READ], sq->nr_queued[WRITE]); |
7f52f98c2
|
960 961 962 |
ret = throtl_select_dispatch(sq); if (ret) { |
7f52f98c2
|
963 964 965 |
throtl_log(sq, "bios disp=%u", ret); dispatched = true; } |
e43473b7f
|
966 |
|
7f52f98c2
|
967 968 |
if (throtl_schedule_next_dispatch(sq, false)) break; |
e43473b7f
|
969 |
|
7f52f98c2
|
970 971 972 973 |
/* this dispatch windows is still open, relax and repeat */ spin_unlock_irq(q->queue_lock); cpu_relax(); spin_lock_irq(q->queue_lock); |
651930bc1
|
974 |
} |
e43473b7f
|
975 |
|
2e48a530a
|
976 977 |
if (!dispatched) goto out_unlock; |
6e1a5704c
|
978 |
|
2e48a530a
|
979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 |
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: |
e43473b7f
|
995 |
spin_unlock_irq(q->queue_lock); |
6e1a5704c
|
996 |
} |
e43473b7f
|
997 |
|
6e1a5704c
|
998 999 1000 1001 1002 1003 1004 1005 |
/** * 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
|
1006 |
static void blk_throtl_dispatch_work_fn(struct work_struct *work) |
6e1a5704c
|
1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 |
{ 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); spin_lock_irq(q->queue_lock); |
c5cc2070b
|
1020 1021 1022 |
for (rw = READ; rw <= WRITE; rw++) while ((bio = throtl_pop_queued(&td_sq->queued[rw], NULL))) bio_list_add(&bio_list_on_stack, bio); |
6e1a5704c
|
1023 1024 1025 |
spin_unlock_irq(q->queue_lock); if (!bio_list_empty(&bio_list_on_stack)) { |
69d60eb96
|
1026 |
blk_start_plug(&plug); |
e43473b7f
|
1027 1028 |
while((bio = bio_list_pop(&bio_list_on_stack))) generic_make_request(bio); |
69d60eb96
|
1029 |
blk_finish_plug(&plug); |
e43473b7f
|
1030 |
} |
e43473b7f
|
1031 |
} |
f95a04afa
|
1032 1033 |
static u64 tg_prfill_conf_u64(struct seq_file *sf, struct blkg_policy_data *pd, int off) |
60c2bc2d5
|
1034 |
{ |
f95a04afa
|
1035 1036 |
struct throtl_grp *tg = pd_to_tg(pd); u64 v = *(u64 *)((void *)tg + off); |
60c2bc2d5
|
1037 |
|
af133ceb2
|
1038 |
if (v == -1) |
60c2bc2d5
|
1039 |
return 0; |
f95a04afa
|
1040 |
return __blkg_prfill_u64(sf, pd, v); |
60c2bc2d5
|
1041 |
} |
f95a04afa
|
1042 1043 |
static u64 tg_prfill_conf_uint(struct seq_file *sf, struct blkg_policy_data *pd, int off) |
e43473b7f
|
1044 |
{ |
f95a04afa
|
1045 1046 |
struct throtl_grp *tg = pd_to_tg(pd); unsigned int v = *(unsigned int *)((void *)tg + off); |
fe0714377
|
1047 |
|
af133ceb2
|
1048 1049 |
if (v == -1) return 0; |
f95a04afa
|
1050 |
return __blkg_prfill_u64(sf, pd, v); |
e43473b7f
|
1051 |
} |
2da8ca822
|
1052 |
static int tg_print_conf_u64(struct seq_file *sf, void *v) |
8e89d13f4
|
1053 |
{ |
2da8ca822
|
1054 1055 |
blkcg_print_blkgs(sf, css_to_blkcg(seq_css(sf)), tg_prfill_conf_u64, &blkcg_policy_throtl, seq_cft(sf)->private, false); |
af133ceb2
|
1056 |
return 0; |
8e89d13f4
|
1057 |
} |
2da8ca822
|
1058 |
static int tg_print_conf_uint(struct seq_file *sf, void *v) |
8e89d13f4
|
1059 |
{ |
2da8ca822
|
1060 1061 |
blkcg_print_blkgs(sf, css_to_blkcg(seq_css(sf)), tg_prfill_conf_uint, &blkcg_policy_throtl, seq_cft(sf)->private, false); |
af133ceb2
|
1062 |
return 0; |
60c2bc2d5
|
1063 |
} |
69948b070
|
1064 |
static void tg_conf_updated(struct throtl_grp *tg) |
60c2bc2d5
|
1065 |
{ |
69948b070
|
1066 |
struct throtl_service_queue *sq = &tg->service_queue; |
492eb21b9
|
1067 |
struct cgroup_subsys_state *pos_css; |
69948b070
|
1068 |
struct blkcg_gq *blkg; |
af133ceb2
|
1069 |
|
fda6f272c
|
1070 1071 1072 1073 |
throtl_log(&tg->service_queue, "limit change rbps=%llu wbps=%llu riops=%u wiops=%u", tg->bps[READ], tg->bps[WRITE], tg->iops[READ], tg->iops[WRITE]); |
632b44935
|
1074 1075 |
/* |
693e751e7
|
1076 1077 1078 1079 1080 1081 |
* 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. */ |
69948b070
|
1082 |
blkg_for_each_descendant_pre(blkg, pos_css, tg_to_blkg(tg)) |
693e751e7
|
1083 1084 1085 |
tg_update_has_rules(blkg_to_tg(blkg)); /* |
632b44935
|
1086 1087 1088 1089 1090 1091 1092 |
* 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
|
1093 1094 |
throtl_start_new_slice(tg, 0); throtl_start_new_slice(tg, 1); |
632b44935
|
1095 |
|
5b2c16aae
|
1096 |
if (tg->flags & THROTL_TG_PENDING) { |
77216b048
|
1097 |
tg_update_disptime(tg); |
7f52f98c2
|
1098 |
throtl_schedule_next_dispatch(sq->parent_sq, true); |
632b44935
|
1099 |
} |
69948b070
|
1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 |
} 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) v = -1; 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
|
1127 |
|
69948b070
|
1128 |
tg_conf_updated(tg); |
36aa9e5f5
|
1129 1130 |
ret = 0; out_finish: |
60c2bc2d5
|
1131 |
blkg_conf_finish(&ctx); |
36aa9e5f5
|
1132 |
return ret ?: nbytes; |
8e89d13f4
|
1133 |
} |
451af504d
|
1134 1135 |
static ssize_t tg_set_conf_u64(struct kernfs_open_file *of, char *buf, size_t nbytes, loff_t off) |
60c2bc2d5
|
1136 |
{ |
451af504d
|
1137 |
return tg_set_conf(of, buf, nbytes, off, true); |
60c2bc2d5
|
1138 |
} |
451af504d
|
1139 1140 |
static ssize_t tg_set_conf_uint(struct kernfs_open_file *of, char *buf, size_t nbytes, loff_t off) |
60c2bc2d5
|
1141 |
{ |
451af504d
|
1142 |
return tg_set_conf(of, buf, nbytes, off, false); |
60c2bc2d5
|
1143 |
} |
880f50e22
|
1144 |
static struct cftype throtl_legacy_files[] = { |
60c2bc2d5
|
1145 1146 |
{ .name = "throttle.read_bps_device", |
af133ceb2
|
1147 |
.private = offsetof(struct throtl_grp, bps[READ]), |
2da8ca822
|
1148 |
.seq_show = tg_print_conf_u64, |
451af504d
|
1149 |
.write = tg_set_conf_u64, |
60c2bc2d5
|
1150 1151 1152 |
}, { .name = "throttle.write_bps_device", |
af133ceb2
|
1153 |
.private = offsetof(struct throtl_grp, bps[WRITE]), |
2da8ca822
|
1154 |
.seq_show = tg_print_conf_u64, |
451af504d
|
1155 |
.write = tg_set_conf_u64, |
60c2bc2d5
|
1156 1157 1158 |
}, { .name = "throttle.read_iops_device", |
af133ceb2
|
1159 |
.private = offsetof(struct throtl_grp, iops[READ]), |
2da8ca822
|
1160 |
.seq_show = tg_print_conf_uint, |
451af504d
|
1161 |
.write = tg_set_conf_uint, |
60c2bc2d5
|
1162 1163 1164 |
}, { .name = "throttle.write_iops_device", |
af133ceb2
|
1165 |
.private = offsetof(struct throtl_grp, iops[WRITE]), |
2da8ca822
|
1166 |
.seq_show = tg_print_conf_uint, |
451af504d
|
1167 |
.write = tg_set_conf_uint, |
60c2bc2d5
|
1168 1169 1170 |
}, { .name = "throttle.io_service_bytes", |
77ea73388
|
1171 1172 |
.private = (unsigned long)&blkcg_policy_throtl, .seq_show = blkg_print_stat_bytes, |
60c2bc2d5
|
1173 1174 1175 |
}, { .name = "throttle.io_serviced", |
77ea73388
|
1176 1177 |
.private = (unsigned long)&blkcg_policy_throtl, .seq_show = blkg_print_stat_ios, |
60c2bc2d5
|
1178 1179 1180 |
}, { } /* terminate */ }; |
2ee867dcf
|
1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 |
static u64 tg_prfill_max(struct seq_file *sf, struct blkg_policy_data *pd, 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" }; if (!dname) return 0; if (tg->bps[READ] == -1 && tg->bps[WRITE] == -1 && tg->iops[READ] == -1 && tg->iops[WRITE] == -1) return 0; if (tg->bps[READ] != -1) snprintf(bufs[0], sizeof(bufs[0]), "%llu", tg->bps[READ]); if (tg->bps[WRITE] != -1) snprintf(bufs[1], sizeof(bufs[1]), "%llu", tg->bps[WRITE]); if (tg->iops[READ] != -1) snprintf(bufs[2], sizeof(bufs[2]), "%u", tg->iops[READ]); if (tg->iops[WRITE] != -1) snprintf(bufs[3], sizeof(bufs[3]), "%u", tg->iops[WRITE]); seq_printf(sf, "%s rbps=%s wbps=%s riops=%s wiops=%s ", dname, bufs[0], bufs[1], bufs[2], bufs[3]); return 0; } static int tg_print_max(struct seq_file *sf, void *v) { blkcg_print_blkgs(sf, css_to_blkcg(seq_css(sf)), tg_prfill_max, &blkcg_policy_throtl, seq_cft(sf)->private, false); return 0; } static ssize_t tg_set_max(struct kernfs_open_file *of, 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]; int ret; ret = blkg_conf_prep(blkcg, &blkcg_policy_throtl, buf, &ctx); if (ret) return ret; tg = blkg_to_tg(ctx.blkg); v[0] = tg->bps[READ]; v[1] = tg->bps[WRITE]; v[2] = tg->iops[READ]; v[3] = tg->iops[WRITE]; while (true) { char tok[27]; /* wiops=18446744073709551616 */ char *p; u64 val = -1; 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); else goto out_finish; } tg->bps[READ] = v[0]; tg->bps[WRITE] = v[1]; tg->iops[READ] = v[2]; tg->iops[WRITE] = v[3]; tg_conf_updated(tg); ret = 0; out_finish: blkg_conf_finish(&ctx); return ret ?: nbytes; } static struct cftype throtl_files[] = { { .name = "max", .flags = CFTYPE_NOT_ON_ROOT, .seq_show = tg_print_max, .write = tg_set_max, }, { } /* terminate */ }; |
da5277700
|
1292 |
static void throtl_shutdown_wq(struct request_queue *q) |
e43473b7f
|
1293 1294 |
{ struct throtl_data *td = q->td; |
69df0ab03
|
1295 |
cancel_work_sync(&td->dispatch_work); |
e43473b7f
|
1296 |
} |
3c798398e
|
1297 |
static struct blkcg_policy blkcg_policy_throtl = { |
2ee867dcf
|
1298 |
.dfl_cftypes = throtl_files, |
880f50e22
|
1299 |
.legacy_cftypes = throtl_legacy_files, |
f9fcc2d39
|
1300 |
|
001bea73e
|
1301 |
.pd_alloc_fn = throtl_pd_alloc, |
f9fcc2d39
|
1302 |
.pd_init_fn = throtl_pd_init, |
693e751e7
|
1303 |
.pd_online_fn = throtl_pd_online, |
001bea73e
|
1304 |
.pd_free_fn = throtl_pd_free, |
e43473b7f
|
1305 |
}; |
ae1188963
|
1306 1307 |
bool blk_throtl_bio(struct request_queue *q, struct blkcg_gq *blkg, struct bio *bio) |
e43473b7f
|
1308 |
{ |
c5cc2070b
|
1309 |
struct throtl_qnode *qn = NULL; |
ae1188963
|
1310 |
struct throtl_grp *tg = blkg_to_tg(blkg ?: q->root_blkg); |
73f0d49a9
|
1311 |
struct throtl_service_queue *sq; |
0e9f4164b
|
1312 |
bool rw = bio_data_dir(bio); |
bc16a4f93
|
1313 |
bool throttled = false; |
e43473b7f
|
1314 |
|
ae1188963
|
1315 |
WARN_ON_ONCE(!rcu_read_lock_held()); |
2a0f61e6e
|
1316 |
/* see throtl_charge_bio() */ |
1eff9d322
|
1317 |
if ((bio->bi_opf & REQ_THROTTLED) || !tg->has_rules[rw]) |
bc16a4f93
|
1318 |
goto out; |
e43473b7f
|
1319 1320 |
spin_lock_irq(q->queue_lock); |
c9589f03e
|
1321 1322 |
if (unlikely(blk_queue_bypass(q))) |
bc16a4f93
|
1323 |
goto out_unlock; |
f469a7b4d
|
1324 |
|
73f0d49a9
|
1325 |
sq = &tg->service_queue; |
9e660acff
|
1326 1327 1328 1329 |
while (true) { /* throtl is FIFO - if bios are already queued, should queue */ if (sq->nr_queued[rw]) break; |
de701c74a
|
1330 |
|
9e660acff
|
1331 1332 1333 1334 1335 |
/* if above limits, break to queue */ if (!tg_may_dispatch(tg, bio, NULL)) break; /* within limits, let's charge and dispatch directly */ |
e43473b7f
|
1336 |
throtl_charge_bio(tg, bio); |
04521db04
|
1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 |
/* * 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
|
1349 |
throtl_trim_slice(tg, rw); |
9e660acff
|
1350 1351 1352 1353 1354 1355 |
/* * @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
|
1356 |
qn = &tg->qnode_on_parent[rw]; |
9e660acff
|
1357 1358 1359 1360 |
sq = sq->parent_sq; tg = sq_to_tg(sq); if (!tg) goto out_unlock; |
e43473b7f
|
1361 |
} |
9e660acff
|
1362 |
/* out-of-limit, queue to @tg */ |
fda6f272c
|
1363 1364 |
throtl_log(sq, "[%c] bio. bdisp=%llu sz=%u bps=%llu iodisp=%u iops=%u queued=%d/%d", rw == READ ? 'R' : 'W', |
4f024f379
|
1365 |
tg->bytes_disp[rw], bio->bi_iter.bi_size, tg->bps[rw], |
fda6f272c
|
1366 1367 |
tg->io_disp[rw], tg->iops[rw], sq->nr_queued[READ], sq->nr_queued[WRITE]); |
e43473b7f
|
1368 |
|
671058fb2
|
1369 |
bio_associate_current(bio); |
6bc9c2b46
|
1370 |
tg->td->nr_queued[rw]++; |
c5cc2070b
|
1371 |
throtl_add_bio_tg(bio, qn, tg); |
bc16a4f93
|
1372 |
throttled = true; |
e43473b7f
|
1373 |
|
7f52f98c2
|
1374 1375 1376 1377 1378 1379 |
/* * 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
|
1380 |
if (tg->flags & THROTL_TG_WAS_EMPTY) { |
77216b048
|
1381 |
tg_update_disptime(tg); |
7f52f98c2
|
1382 |
throtl_schedule_next_dispatch(tg->service_queue.parent_sq, true); |
e43473b7f
|
1383 |
} |
bc16a4f93
|
1384 |
out_unlock: |
e43473b7f
|
1385 |
spin_unlock_irq(q->queue_lock); |
bc16a4f93
|
1386 |
out: |
2a0f61e6e
|
1387 1388 1389 1390 1391 1392 |
/* * As multiple blk-throtls may stack in the same issue path, we * don't want bios to leave with the flag set. Clear the flag if * being issued. */ if (!throttled) |
1eff9d322
|
1393 |
bio->bi_opf &= ~REQ_THROTTLED; |
bc16a4f93
|
1394 |
return throttled; |
e43473b7f
|
1395 |
} |
2a12f0dcd
|
1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 |
/* * 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
|
1410 |
while ((bio = throtl_peek_queued(&sq->queued[READ]))) |
2a12f0dcd
|
1411 |
tg_dispatch_one_bio(tg, bio_data_dir(bio)); |
c5cc2070b
|
1412 |
while ((bio = throtl_peek_queued(&sq->queued[WRITE]))) |
2a12f0dcd
|
1413 1414 1415 |
tg_dispatch_one_bio(tg, bio_data_dir(bio)); } } |
c9a929dde
|
1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 |
/** * 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) __releases(q->queue_lock) __acquires(q->queue_lock) { struct throtl_data *td = q->td; |
2a12f0dcd
|
1426 |
struct blkcg_gq *blkg; |
492eb21b9
|
1427 |
struct cgroup_subsys_state *pos_css; |
c9a929dde
|
1428 |
struct bio *bio; |
651930bc1
|
1429 |
int rw; |
c9a929dde
|
1430 |
|
8bcb6c7d4
|
1431 |
queue_lockdep_assert_held(q); |
2a12f0dcd
|
1432 |
rcu_read_lock(); |
c9a929dde
|
1433 |
|
2a12f0dcd
|
1434 1435 1436 1437 1438 1439 |
/* * 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
|
1440 |
blkg_for_each_descendant_post(blkg, pos_css, td->queue->root_blkg) |
2a12f0dcd
|
1441 |
tg_drain_bios(&blkg_to_tg(blkg)->service_queue); |
73f0d49a9
|
1442 |
|
2a12f0dcd
|
1443 1444 1445 1446 |
/* finally, transfer bios from top-level tg's into the td */ tg_drain_bios(&td->service_queue); rcu_read_unlock(); |
c9a929dde
|
1447 |
spin_unlock_irq(q->queue_lock); |
2a12f0dcd
|
1448 |
/* all bios now should be in td->service_queue, issue them */ |
651930bc1
|
1449 |
for (rw = READ; rw <= WRITE; rw++) |
c5cc2070b
|
1450 1451 |
while ((bio = throtl_pop_queued(&td->service_queue.queued[rw], NULL))) |
651930bc1
|
1452 |
generic_make_request(bio); |
c9a929dde
|
1453 1454 1455 |
spin_lock_irq(q->queue_lock); } |
e43473b7f
|
1456 1457 1458 |
int blk_throtl_init(struct request_queue *q) { struct throtl_data *td; |
a2b1693ba
|
1459 |
int ret; |
e43473b7f
|
1460 1461 1462 1463 |
td = kzalloc_node(sizeof(*td), GFP_KERNEL, q->node); if (!td) return -ENOMEM; |
69df0ab03
|
1464 |
INIT_WORK(&td->dispatch_work, blk_throtl_dispatch_work_fn); |
b2ce2643c
|
1465 |
throtl_service_queue_init(&td->service_queue); |
e43473b7f
|
1466 |
|
cd1604fab
|
1467 |
q->td = td; |
29b125892
|
1468 |
td->queue = q; |
02977e4af
|
1469 |
|
a2b1693ba
|
1470 |
/* activate policy */ |
3c798398e
|
1471 |
ret = blkcg_activate_policy(q, &blkcg_policy_throtl); |
a2b1693ba
|
1472 |
if (ret) |
f51b802c1
|
1473 |
kfree(td); |
a2b1693ba
|
1474 |
return ret; |
e43473b7f
|
1475 1476 1477 1478 |
} void blk_throtl_exit(struct request_queue *q) { |
c875f4d02
|
1479 |
BUG_ON(!q->td); |
da5277700
|
1480 |
throtl_shutdown_wq(q); |
3c798398e
|
1481 |
blkcg_deactivate_policy(q, &blkcg_policy_throtl); |
c9a929dde
|
1482 |
kfree(q->td); |
e43473b7f
|
1483 1484 1485 1486 |
} static int __init throtl_init(void) { |
450adcbe5
|
1487 1488 1489 1490 |
kthrotld_workqueue = alloc_workqueue("kthrotld", WQ_MEM_RECLAIM, 0); if (!kthrotld_workqueue) panic("Failed to create kthrotld "); |
3c798398e
|
1491 |
return blkcg_policy_register(&blkcg_policy_throtl); |
e43473b7f
|
1492 1493 1494 |
} module_init(throtl_init); |