blk-cgroup.c 35.8 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11
/*
 * Common Block IO controller cgroup interface
 *
 * Based on ideas and code from CFQ, CFS and BFQ:
 * Copyright (C) 2003 Jens Axboe <axboe@kernel.dk>
 *
 * Copyright (C) 2008 Fabio Checconi <fabio@gandalf.sssup.it>
 *		      Paolo Valente <paolo.valente@unimore.it>
 *
 * Copyright (C) 2009 Vivek Goyal <vgoyal@redhat.com>
 * 	              Nauman Rafique <nauman@google.com>
12 13 14 15
 *
 * For policy-specific per-blkcg data:
 * Copyright (C) 2015 Paolo Valente <paolo.valente@unimore.it>
 *                    Arianna Avanzini <avanzini.arianna@gmail.com>
16 17
 */
#include <linux/ioprio.h>
18
#include <linux/kdev_t.h>
19
#include <linux/module.h>
20
#include <linux/err.h>
21
#include <linux/blkdev.h>
22
#include <linux/backing-dev.h>
23
#include <linux/slab.h>
24
#include <linux/genhd.h>
25
#include <linux/delay.h>
T
Tejun Heo 已提交
26
#include <linux/atomic.h>
27
#include <linux/ctype.h>
28
#include <linux/blk-cgroup.h>
29
#include "blk.h"
30

31 32
#define MAX_KEY_LEN 100

33 34 35 36 37 38 39 40
/*
 * blkcg_pol_mutex protects blkcg_policy[] and policy [de]activation.
 * blkcg_pol_register_mutex nests outside of it and synchronizes entire
 * policy [un]register operations including cgroup file additions /
 * removals.  Putting cgroup file registration outside blkcg_pol_mutex
 * allows grabbing it from cgroup callbacks.
 */
static DEFINE_MUTEX(blkcg_pol_register_mutex);
41
static DEFINE_MUTEX(blkcg_pol_mutex);
42

43
struct blkcg blkcg_root;
T
Tejun Heo 已提交
44
EXPORT_SYMBOL_GPL(blkcg_root);
45

T
Tejun Heo 已提交
46 47
struct cgroup_subsys_state * const blkcg_root_css = &blkcg_root.css;

T
Tejun Heo 已提交
48
static struct blkcg_policy *blkcg_policy[BLKCG_MAX_POLS];
49

T
Tejun Heo 已提交
50 51
static LIST_HEAD(all_blkcgs);		/* protected by blkcg_pol_mutex */

52
static bool blkcg_policy_enabled(struct request_queue *q,
T
Tejun Heo 已提交
53
				 const struct blkcg_policy *pol)
54 55 56 57
{
	return pol && test_bit(pol->plid, q->blkcg_pols);
}

58 59 60 61 62 63
/**
 * blkg_free - free a blkg
 * @blkg: blkg to free
 *
 * Free @blkg which may be partially allocated.
 */
T
Tejun Heo 已提交
64
static void blkg_free(struct blkcg_gq *blkg)
65
{
66
	int i;
67 68 69 70

	if (!blkg)
		return;

71
	for (i = 0; i < BLKCG_MAX_POLS; i++)
72 73
		if (blkg->pd[i])
			blkcg_policy[i]->pd_free_fn(blkg->pd[i]);
74

75 76
	if (blkg->blkcg != &blkcg_root)
		blk_exit_rl(&blkg->rl);
77 78 79

	blkg_rwstat_exit(&blkg->stat_ios);
	blkg_rwstat_exit(&blkg->stat_bytes);
80
	kfree(blkg);
81 82 83 84 85 86
}

/**
 * blkg_alloc - allocate a blkg
 * @blkcg: block cgroup the new blkg is associated with
 * @q: request_queue the new blkg is associated with
87
 * @gfp_mask: allocation mask to use
88
 *
89
 * Allocate a new blkg assocating @blkcg and @q.
90
 */
91 92
static struct blkcg_gq *blkg_alloc(struct blkcg *blkcg, struct request_queue *q,
				   gfp_t gfp_mask)
93
{
T
Tejun Heo 已提交
94
	struct blkcg_gq *blkg;
95
	int i;
96 97

	/* alloc and init base part */
98
	blkg = kzalloc_node(sizeof(*blkg), gfp_mask, q->node);
99 100 101
	if (!blkg)
		return NULL;

102 103 104 105
	if (blkg_rwstat_init(&blkg->stat_bytes, gfp_mask) ||
	    blkg_rwstat_init(&blkg->stat_ios, gfp_mask))
		goto err_free;

T
Tejun Heo 已提交
106
	blkg->q = q;
107
	INIT_LIST_HEAD(&blkg->q_node);
108
	blkg->blkcg = blkcg;
109
	atomic_set(&blkg->refcnt, 1);
110

111 112 113 114 115 116 117
	/* root blkg uses @q->root_rl, init rl only for !root blkgs */
	if (blkcg != &blkcg_root) {
		if (blk_init_rl(&blkg->rl, q, gfp_mask))
			goto err_free;
		blkg->rl.blkg = blkg;
	}

T
Tejun Heo 已提交
118
	for (i = 0; i < BLKCG_MAX_POLS; i++) {
T
Tejun Heo 已提交
119
		struct blkcg_policy *pol = blkcg_policy[i];
120
		struct blkg_policy_data *pd;
121

122
		if (!blkcg_policy_enabled(q, pol))
123 124 125
			continue;

		/* alloc per-policy data and attach it to blkg */
126
		pd = pol->pd_alloc_fn(gfp_mask, q->node);
127 128
		if (!pd)
			goto err_free;
129

130 131
		blkg->pd[i] = pd;
		pd->blkg = blkg;
T
Tejun Heo 已提交
132
		pd->plid = i;
133 134
	}

135
	return blkg;
136 137 138 139

err_free:
	blkg_free(blkg);
	return NULL;
140 141
}

T
Tejun Heo 已提交
142 143
struct blkcg_gq *blkg_lookup_slowpath(struct blkcg *blkcg,
				      struct request_queue *q, bool update_hint)
144
{
T
Tejun Heo 已提交
145
	struct blkcg_gq *blkg;
146

147
	/*
148 149 150 151
	 * Hint didn't match.  Look up from the radix tree.  Note that the
	 * hint can only be updated under queue_lock as otherwise @blkg
	 * could have already been removed from blkg_tree.  The caller is
	 * responsible for grabbing queue_lock if @update_hint.
152 153
	 */
	blkg = radix_tree_lookup(&blkcg->blkg_tree, q->id);
154 155 156 157 158
	if (blkg && blkg->q == q) {
		if (update_hint) {
			lockdep_assert_held(q->queue_lock);
			rcu_assign_pointer(blkcg->blkg_hint, blkg);
		}
159
		return blkg;
160
	}
161

162 163
	return NULL;
}
164
EXPORT_SYMBOL_GPL(blkg_lookup_slowpath);
165

166 167
/*
 * If @new_blkg is %NULL, this function tries to allocate a new one as
168
 * necessary using %GFP_NOWAIT.  @new_blkg is always consumed on return.
169
 */
170 171 172
static struct blkcg_gq *blkg_create(struct blkcg *blkcg,
				    struct request_queue *q,
				    struct blkcg_gq *new_blkg)
173
{
T
Tejun Heo 已提交
174
	struct blkcg_gq *blkg;
175
	struct bdi_writeback_congested *wb_congested;
176
	int i, ret;
177

178 179 180
	WARN_ON_ONCE(!rcu_read_lock_held());
	lockdep_assert_held(q->queue_lock);

181
	/* blkg holds a reference to blkcg */
182
	if (!css_tryget_online(&blkcg->css)) {
183
		ret = -ENODEV;
184
		goto err_free_blkg;
185
	}
186

187
	wb_congested = wb_congested_get_create(&q->backing_dev_info,
188
					       blkcg->css.id, GFP_NOWAIT);
189 190 191 192 193
	if (!wb_congested) {
		ret = -ENOMEM;
		goto err_put_css;
	}

194
	/* allocate */
195
	if (!new_blkg) {
196
		new_blkg = blkg_alloc(blkcg, q, GFP_NOWAIT);
197
		if (unlikely(!new_blkg)) {
198
			ret = -ENOMEM;
199
			goto err_put_congested;
200 201 202
		}
	}
	blkg = new_blkg;
203
	blkg->wb_congested = wb_congested;
204

205
	/* link parent */
T
Tejun Heo 已提交
206 207 208
	if (blkcg_parent(blkcg)) {
		blkg->parent = __blkg_lookup(blkcg_parent(blkcg), q, false);
		if (WARN_ON_ONCE(!blkg->parent)) {
209
			ret = -ENODEV;
210
			goto err_put_congested;
T
Tejun Heo 已提交
211 212 213 214
		}
		blkg_get(blkg->parent);
	}

215 216 217 218 219
	/* invoke per-policy init */
	for (i = 0; i < BLKCG_MAX_POLS; i++) {
		struct blkcg_policy *pol = blkcg_policy[i];

		if (blkg->pd[i] && pol->pd_init_fn)
220
			pol->pd_init_fn(blkg->pd[i]);
221 222 223
	}

	/* insert */
224
	spin_lock(&blkcg->lock);
225 226 227 228
	ret = radix_tree_insert(&blkcg->blkg_tree, q->id, blkg);
	if (likely(!ret)) {
		hlist_add_head_rcu(&blkg->blkcg_node, &blkcg->blkg_list);
		list_add(&blkg->q_node, &q->blkg_list);
229 230 231 232 233

		for (i = 0; i < BLKCG_MAX_POLS; i++) {
			struct blkcg_policy *pol = blkcg_policy[i];

			if (blkg->pd[i] && pol->pd_online_fn)
234
				pol->pd_online_fn(blkg->pd[i]);
235
		}
236
	}
237
	blkg->online = true;
238
	spin_unlock(&blkcg->lock);
239

240
	if (!ret)
241
		return blkg;
242

T
Tejun Heo 已提交
243 244 245 246
	/* @blkg failed fully initialized, use the usual release path */
	blkg_put(blkg);
	return ERR_PTR(ret);

247 248
err_put_congested:
	wb_congested_put(wb_congested);
249
err_put_css:
250
	css_put(&blkcg->css);
251
err_free_blkg:
252
	blkg_free(new_blkg);
253
	return ERR_PTR(ret);
254
}
255

256 257 258 259 260 261
/**
 * blkg_lookup_create - lookup blkg, try to create one if not there
 * @blkcg: blkcg of interest
 * @q: request_queue of interest
 *
 * Lookup blkg for the @blkcg - @q pair.  If it doesn't exist, try to
T
Tejun Heo 已提交
262 263 264
 * create one.  blkg creation is performed recursively from blkcg_root such
 * that all non-root blkg's have access to the parent blkg.  This function
 * should be called under RCU read lock and @q->queue_lock.
265 266 267 268 269
 *
 * Returns pointer to the looked up or created blkg on success, ERR_PTR()
 * value on error.  If @q is dead, returns ERR_PTR(-EINVAL).  If @q is not
 * dead and bypassing, returns ERR_PTR(-EBUSY).
 */
T
Tejun Heo 已提交
270 271
struct blkcg_gq *blkg_lookup_create(struct blkcg *blkcg,
				    struct request_queue *q)
272
{
273 274 275 276 277
	struct blkcg_gq *blkg;

	WARN_ON_ONCE(!rcu_read_lock_held());
	lockdep_assert_held(q->queue_lock);

278 279 280 281 282
	/*
	 * This could be the first entry point of blkcg implementation and
	 * we shouldn't allow anything to go through for a bypassing queue.
	 */
	if (unlikely(blk_queue_bypass(q)))
283
		return ERR_PTR(blk_queue_dying(q) ? -ENODEV : -EBUSY);
284 285 286 287 288

	blkg = __blkg_lookup(blkcg, q, true);
	if (blkg)
		return blkg;

T
Tejun Heo 已提交
289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305
	/*
	 * Create blkgs walking down from blkcg_root to @blkcg, so that all
	 * non-root blkgs have access to their parents.
	 */
	while (true) {
		struct blkcg *pos = blkcg;
		struct blkcg *parent = blkcg_parent(blkcg);

		while (parent && !__blkg_lookup(parent, q, false)) {
			pos = parent;
			parent = blkcg_parent(parent);
		}

		blkg = blkg_create(pos, q, NULL);
		if (pos == blkcg || IS_ERR(blkg))
			return blkg;
	}
306
}
307

T
Tejun Heo 已提交
308
static void blkg_destroy(struct blkcg_gq *blkg)
309
{
T
Tejun Heo 已提交
310
	struct blkcg *blkcg = blkg->blkcg;
311
	struct blkcg_gq *parent = blkg->parent;
312
	int i;
313

314
	lockdep_assert_held(blkg->q->queue_lock);
315
	lockdep_assert_held(&blkcg->lock);
316 317

	/* Something wrong if we are trying to remove same group twice */
318
	WARN_ON_ONCE(list_empty(&blkg->q_node));
319
	WARN_ON_ONCE(hlist_unhashed(&blkg->blkcg_node));
320

321 322 323 324
	for (i = 0; i < BLKCG_MAX_POLS; i++) {
		struct blkcg_policy *pol = blkcg_policy[i];

		if (blkg->pd[i] && pol->pd_offline_fn)
325
			pol->pd_offline_fn(blkg->pd[i]);
326
	}
327 328 329 330 331 332

	if (parent) {
		blkg_rwstat_add_aux(&parent->stat_bytes, &blkg->stat_bytes);
		blkg_rwstat_add_aux(&parent->stat_ios, &blkg->stat_ios);
	}

333 334
	blkg->online = false;

335
	radix_tree_delete(&blkcg->blkg_tree, blkg->q->id);
336
	list_del_init(&blkg->q_node);
337
	hlist_del_init_rcu(&blkg->blkcg_node);
338

339 340 341 342 343
	/*
	 * Both setting lookup hint to and clearing it from @blkg are done
	 * under queue_lock.  If it's not pointing to @blkg now, it never
	 * will.  Hint assignment itself can race safely.
	 */
344
	if (rcu_access_pointer(blkcg->blkg_hint) == blkg)
345 346
		rcu_assign_pointer(blkcg->blkg_hint, NULL);

347 348 349 350 351 352 353
	/*
	 * Put the reference taken at the time of creation so that when all
	 * queues are gone, group can be destroyed.
	 */
	blkg_put(blkg);
}

354 355 356 357
/**
 * blkg_destroy_all - destroy all blkgs associated with a request_queue
 * @q: request_queue of interest
 *
358
 * Destroy all blkgs associated with @q.
359
 */
360
static void blkg_destroy_all(struct request_queue *q)
361
{
T
Tejun Heo 已提交
362
	struct blkcg_gq *blkg, *n;
363

364
	lockdep_assert_held(q->queue_lock);
365

366
	list_for_each_entry_safe(blkg, n, &q->blkg_list, q_node) {
T
Tejun Heo 已提交
367
		struct blkcg *blkcg = blkg->blkcg;
368

369 370 371
		spin_lock(&blkcg->lock);
		blkg_destroy(blkg);
		spin_unlock(&blkcg->lock);
372
	}
373 374 375

	q->root_blkg = NULL;
	q->root_rl.blkg = NULL;
376 377
}

378 379 380 381 382 383 384 385 386
/*
 * A group is RCU protected, but having an rcu lock does not mean that one
 * can access all the fields of blkg and assume these are valid.  For
 * example, don't try to follow throtl_data and request queue links.
 *
 * Having a reference to blkg under an rcu allows accesses to only values
 * local to groups like group stats and group rate limits.
 */
void __blkg_release_rcu(struct rcu_head *rcu_head)
T
Tejun Heo 已提交
387
{
388
	struct blkcg_gq *blkg = container_of(rcu_head, struct blkcg_gq, rcu_head);
389

T
Tejun Heo 已提交
390
	/* release the blkcg and parent blkg refs this blkg has been holding */
T
Tejun Heo 已提交
391
	css_put(&blkg->blkcg->css);
392
	if (blkg->parent)
T
Tejun Heo 已提交
393
		blkg_put(blkg->parent);
T
Tejun Heo 已提交
394

395 396
	wb_congested_put(blkg->wb_congested);

397
	blkg_free(blkg);
T
Tejun Heo 已提交
398
}
399
EXPORT_SYMBOL_GPL(__blkg_release_rcu);
T
Tejun Heo 已提交
400

401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416
/*
 * The next function used by blk_queue_for_each_rl().  It's a bit tricky
 * because the root blkg uses @q->root_rl instead of its own rl.
 */
struct request_list *__blk_queue_next_rl(struct request_list *rl,
					 struct request_queue *q)
{
	struct list_head *ent;
	struct blkcg_gq *blkg;

	/*
	 * Determine the current blkg list_head.  The first entry is
	 * root_rl which is off @q->blkg_list and mapped to the head.
	 */
	if (rl == &q->root_rl) {
		ent = &q->blkg_list;
417 418 419
		/* There are no more block groups, hence no request lists */
		if (list_empty(ent))
			return NULL;
420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435
	} else {
		blkg = container_of(rl, struct blkcg_gq, rl);
		ent = &blkg->q_node;
	}

	/* walk to the next list_head, skip root blkcg */
	ent = ent->next;
	if (ent == &q->root_blkg->q_node)
		ent = ent->next;
	if (ent == &q->blkg_list)
		return NULL;

	blkg = container_of(ent, struct blkcg_gq, q_node);
	return &blkg->rl;
}

436 437
static int blkcg_reset_stats(struct cgroup_subsys_state *css,
			     struct cftype *cftype, u64 val)
438
{
439
	struct blkcg *blkcg = css_to_blkcg(css);
T
Tejun Heo 已提交
440
	struct blkcg_gq *blkg;
441
	int i;
442

443
	mutex_lock(&blkcg_pol_mutex);
444
	spin_lock_irq(&blkcg->lock);
T
Tejun Heo 已提交
445 446 447 448 449 450

	/*
	 * Note that stat reset is racy - it doesn't synchronize against
	 * stat updates.  This is a debug feature which shouldn't exist
	 * anyway.  If you get hit by a race, retry.
	 */
451
	hlist_for_each_entry(blkg, &blkcg->blkg_list, blkcg_node) {
452 453 454
		blkg_rwstat_reset(&blkg->stat_bytes);
		blkg_rwstat_reset(&blkg->stat_ios);

T
Tejun Heo 已提交
455
		for (i = 0; i < BLKCG_MAX_POLS; i++) {
T
Tejun Heo 已提交
456
			struct blkcg_policy *pol = blkcg_policy[i];
457

458 459
			if (blkg->pd[i] && pol->pd_reset_stats_fn)
				pol->pd_reset_stats_fn(blkg->pd[i]);
460
		}
461
	}
462

463
	spin_unlock_irq(&blkcg->lock);
464
	mutex_unlock(&blkcg_pol_mutex);
465 466 467
	return 0;
}

468
const char *blkg_dev_name(struct blkcg_gq *blkg)
469
{
470 471 472 473
	/* some drivers (floppy) instantiate a queue w/o disk registered */
	if (blkg->q->backing_dev_info.dev)
		return dev_name(blkg->q->backing_dev_info.dev);
	return NULL;
474
}
475
EXPORT_SYMBOL_GPL(blkg_dev_name);
476

477 478 479 480 481 482 483 484 485 486 487
/**
 * blkcg_print_blkgs - helper for printing per-blkg data
 * @sf: seq_file to print to
 * @blkcg: blkcg of interest
 * @prfill: fill function to print out a blkg
 * @pol: policy in question
 * @data: data to be passed to @prfill
 * @show_total: to print out sum of prfill return values or not
 *
 * This function invokes @prfill on each blkg of @blkcg if pd for the
 * policy specified by @pol exists.  @prfill is invoked with @sf, the
488 489 490
 * policy data and @data and the matching queue lock held.  If @show_total
 * is %true, the sum of the return values from @prfill is printed with
 * "Total" label at the end.
491 492 493 494
 *
 * This is to be used to construct print functions for
 * cftype->read_seq_string method.
 */
T
Tejun Heo 已提交
495
void blkcg_print_blkgs(struct seq_file *sf, struct blkcg *blkcg,
496 497
		       u64 (*prfill)(struct seq_file *,
				     struct blkg_policy_data *, int),
T
Tejun Heo 已提交
498
		       const struct blkcg_policy *pol, int data,
499
		       bool show_total)
500
{
T
Tejun Heo 已提交
501
	struct blkcg_gq *blkg;
502
	u64 total = 0;
503

504
	rcu_read_lock();
505
	hlist_for_each_entry_rcu(blkg, &blkcg->blkg_list, blkcg_node) {
506
		spin_lock_irq(blkg->q->queue_lock);
507
		if (blkcg_policy_enabled(blkg->q, pol))
508
			total += prfill(sf, blkg->pd[pol->plid], data);
509 510 511
		spin_unlock_irq(blkg->q->queue_lock);
	}
	rcu_read_unlock();
512 513 514 515

	if (show_total)
		seq_printf(sf, "Total %llu\n", (unsigned long long)total);
}
516
EXPORT_SYMBOL_GPL(blkcg_print_blkgs);
517 518 519 520

/**
 * __blkg_prfill_u64 - prfill helper for a single u64 value
 * @sf: seq_file to print to
521
 * @pd: policy private data of interest
522 523
 * @v: value to print
 *
524
 * Print @v to @sf for the device assocaited with @pd.
525
 */
526
u64 __blkg_prfill_u64(struct seq_file *sf, struct blkg_policy_data *pd, u64 v)
527
{
528
	const char *dname = blkg_dev_name(pd->blkg);
529 530 531 532 533 534 535

	if (!dname)
		return 0;

	seq_printf(sf, "%s %llu\n", dname, (unsigned long long)v);
	return v;
}
536
EXPORT_SYMBOL_GPL(__blkg_prfill_u64);
537 538 539 540

/**
 * __blkg_prfill_rwstat - prfill helper for a blkg_rwstat
 * @sf: seq_file to print to
541
 * @pd: policy private data of interest
542 543
 * @rwstat: rwstat to print
 *
544
 * Print @rwstat to @sf for the device assocaited with @pd.
545
 */
546
u64 __blkg_prfill_rwstat(struct seq_file *sf, struct blkg_policy_data *pd,
547
			 const struct blkg_rwstat *rwstat)
548 549 550 551 552 553 554
{
	static const char *rwstr[] = {
		[BLKG_RWSTAT_READ]	= "Read",
		[BLKG_RWSTAT_WRITE]	= "Write",
		[BLKG_RWSTAT_SYNC]	= "Sync",
		[BLKG_RWSTAT_ASYNC]	= "Async",
	};
555
	const char *dname = blkg_dev_name(pd->blkg);
556 557 558 559 560 561 562 563
	u64 v;
	int i;

	if (!dname)
		return 0;

	for (i = 0; i < BLKG_RWSTAT_NR; i++)
		seq_printf(sf, "%s %s %llu\n", dname, rwstr[i],
T
Tejun Heo 已提交
564
			   (unsigned long long)atomic64_read(&rwstat->aux_cnt[i]));
565

T
Tejun Heo 已提交
566 567
	v = atomic64_read(&rwstat->aux_cnt[BLKG_RWSTAT_READ]) +
		atomic64_read(&rwstat->aux_cnt[BLKG_RWSTAT_WRITE]);
568 569 570
	seq_printf(sf, "%s Total %llu\n", dname, (unsigned long long)v);
	return v;
}
T
Tejun Heo 已提交
571
EXPORT_SYMBOL_GPL(__blkg_prfill_rwstat);
572

573 574 575
/**
 * blkg_prfill_stat - prfill callback for blkg_stat
 * @sf: seq_file to print to
576 577
 * @pd: policy private data of interest
 * @off: offset to the blkg_stat in @pd
578 579 580
 *
 * prfill callback for printing a blkg_stat.
 */
581
u64 blkg_prfill_stat(struct seq_file *sf, struct blkg_policy_data *pd, int off)
582
{
583
	return __blkg_prfill_u64(sf, pd, blkg_stat_read((void *)pd + off));
584
}
585
EXPORT_SYMBOL_GPL(blkg_prfill_stat);
586

587 588 589
/**
 * blkg_prfill_rwstat - prfill callback for blkg_rwstat
 * @sf: seq_file to print to
590 591
 * @pd: policy private data of interest
 * @off: offset to the blkg_rwstat in @pd
592 593 594
 *
 * prfill callback for printing a blkg_rwstat.
 */
595 596
u64 blkg_prfill_rwstat(struct seq_file *sf, struct blkg_policy_data *pd,
		       int off)
597
{
598
	struct blkg_rwstat rwstat = blkg_rwstat_read((void *)pd + off);
599

600
	return __blkg_prfill_rwstat(sf, pd, &rwstat);
601
}
602
EXPORT_SYMBOL_GPL(blkg_prfill_rwstat);
603

604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684
static u64 blkg_prfill_rwstat_field(struct seq_file *sf,
				    struct blkg_policy_data *pd, int off)
{
	struct blkg_rwstat rwstat = blkg_rwstat_read((void *)pd->blkg + off);

	return __blkg_prfill_rwstat(sf, pd, &rwstat);
}

/**
 * blkg_print_stat_bytes - seq_show callback for blkg->stat_bytes
 * @sf: seq_file to print to
 * @v: unused
 *
 * To be used as cftype->seq_show to print blkg->stat_bytes.
 * cftype->private must be set to the blkcg_policy.
 */
int blkg_print_stat_bytes(struct seq_file *sf, void *v)
{
	blkcg_print_blkgs(sf, css_to_blkcg(seq_css(sf)),
			  blkg_prfill_rwstat_field, (void *)seq_cft(sf)->private,
			  offsetof(struct blkcg_gq, stat_bytes), true);
	return 0;
}
EXPORT_SYMBOL_GPL(blkg_print_stat_bytes);

/**
 * blkg_print_stat_bytes - seq_show callback for blkg->stat_ios
 * @sf: seq_file to print to
 * @v: unused
 *
 * To be used as cftype->seq_show to print blkg->stat_ios.  cftype->private
 * must be set to the blkcg_policy.
 */
int blkg_print_stat_ios(struct seq_file *sf, void *v)
{
	blkcg_print_blkgs(sf, css_to_blkcg(seq_css(sf)),
			  blkg_prfill_rwstat_field, (void *)seq_cft(sf)->private,
			  offsetof(struct blkcg_gq, stat_ios), true);
	return 0;
}
EXPORT_SYMBOL_GPL(blkg_print_stat_ios);

static u64 blkg_prfill_rwstat_field_recursive(struct seq_file *sf,
					      struct blkg_policy_data *pd,
					      int off)
{
	struct blkg_rwstat rwstat = blkg_rwstat_recursive_sum(pd->blkg,
							      NULL, off);
	return __blkg_prfill_rwstat(sf, pd, &rwstat);
}

/**
 * blkg_print_stat_bytes_recursive - recursive version of blkg_print_stat_bytes
 * @sf: seq_file to print to
 * @v: unused
 */
int blkg_print_stat_bytes_recursive(struct seq_file *sf, void *v)
{
	blkcg_print_blkgs(sf, css_to_blkcg(seq_css(sf)),
			  blkg_prfill_rwstat_field_recursive,
			  (void *)seq_cft(sf)->private,
			  offsetof(struct blkcg_gq, stat_bytes), true);
	return 0;
}
EXPORT_SYMBOL_GPL(blkg_print_stat_bytes_recursive);

/**
 * blkg_print_stat_ios_recursive - recursive version of blkg_print_stat_ios
 * @sf: seq_file to print to
 * @v: unused
 */
int blkg_print_stat_ios_recursive(struct seq_file *sf, void *v)
{
	blkcg_print_blkgs(sf, css_to_blkcg(seq_css(sf)),
			  blkg_prfill_rwstat_field_recursive,
			  (void *)seq_cft(sf)->private,
			  offsetof(struct blkcg_gq, stat_ios), true);
	return 0;
}
EXPORT_SYMBOL_GPL(blkg_print_stat_ios_recursive);

685 686
/**
 * blkg_stat_recursive_sum - collect hierarchical blkg_stat
687 688 689
 * @blkg: blkg of interest
 * @pol: blkcg_policy which contains the blkg_stat
 * @off: offset to the blkg_stat in blkg_policy_data or @blkg
690
 *
691 692 693 694 695 696
 * Collect the blkg_stat specified by @blkg, @pol and @off and all its
 * online descendants and their aux counts.  The caller must be holding the
 * queue lock for online tests.
 *
 * If @pol is NULL, blkg_stat is at @off bytes into @blkg; otherwise, it is
 * at @off bytes into @blkg's blkg_policy_data of the policy.
697
 */
698 699
u64 blkg_stat_recursive_sum(struct blkcg_gq *blkg,
			    struct blkcg_policy *pol, int off)
700 701
{
	struct blkcg_gq *pos_blkg;
702
	struct cgroup_subsys_state *pos_css;
703
	u64 sum = 0;
704

705
	lockdep_assert_held(blkg->q->queue_lock);
706 707

	rcu_read_lock();
708 709 710 711 712
	blkg_for_each_descendant_pre(pos_blkg, pos_css, blkg) {
		struct blkg_stat *stat;

		if (!pos_blkg->online)
			continue;
713

714 715 716 717 718 719
		if (pol)
			stat = (void *)blkg_to_pd(pos_blkg, pol) + off;
		else
			stat = (void *)blkg + off;

		sum += blkg_stat_read(stat) + atomic64_read(&stat->aux_cnt);
720 721 722 723 724 725 726 727 728
	}
	rcu_read_unlock();

	return sum;
}
EXPORT_SYMBOL_GPL(blkg_stat_recursive_sum);

/**
 * blkg_rwstat_recursive_sum - collect hierarchical blkg_rwstat
729 730 731
 * @blkg: blkg of interest
 * @pol: blkcg_policy which contains the blkg_rwstat
 * @off: offset to the blkg_rwstat in blkg_policy_data or @blkg
732
 *
733 734 735 736 737 738
 * Collect the blkg_rwstat specified by @blkg, @pol and @off and all its
 * online descendants and their aux counts.  The caller must be holding the
 * queue lock for online tests.
 *
 * If @pol is NULL, blkg_rwstat is at @off bytes into @blkg; otherwise, it
 * is at @off bytes into @blkg's blkg_policy_data of the policy.
739
 */
740 741
struct blkg_rwstat blkg_rwstat_recursive_sum(struct blkcg_gq *blkg,
					     struct blkcg_policy *pol, int off)
742 743
{
	struct blkcg_gq *pos_blkg;
744
	struct cgroup_subsys_state *pos_css;
745
	struct blkg_rwstat sum = { };
746 747
	int i;

748
	lockdep_assert_held(blkg->q->queue_lock);
749 750

	rcu_read_lock();
751
	blkg_for_each_descendant_pre(pos_blkg, pos_css, blkg) {
752
		struct blkg_rwstat *rwstat;
753 754 755 756

		if (!pos_blkg->online)
			continue;

757 758 759 760 761
		if (pol)
			rwstat = (void *)blkg_to_pd(pos_blkg, pol) + off;
		else
			rwstat = (void *)pos_blkg + off;

762
		for (i = 0; i < BLKG_RWSTAT_NR; i++)
763 764 765
			atomic64_add(atomic64_read(&rwstat->aux_cnt[i]) +
				percpu_counter_sum_positive(&rwstat->cpu_cnt[i]),
				&sum.aux_cnt[i]);
766 767 768 769 770 771 772
	}
	rcu_read_unlock();

	return sum;
}
EXPORT_SYMBOL_GPL(blkg_rwstat_recursive_sum);

773 774 775
/**
 * blkg_conf_prep - parse and prepare for per-blkg config update
 * @blkcg: target block cgroup
776
 * @pol: target policy
777 778 779 780
 * @input: input string
 * @ctx: blkg_conf_ctx to be filled
 *
 * Parse per-blkg config update from @input and initialize @ctx with the
781 782 783
 * result.  @ctx->blkg points to the blkg to be updated and @ctx->body the
 * part of @input following MAJ:MIN.  This function returns with RCU read
 * lock and queue lock held and must be paired with blkg_conf_finish().
784
 */
T
Tejun Heo 已提交
785
int blkg_conf_prep(struct blkcg *blkcg, const struct blkcg_policy *pol,
786
		   char *input, struct blkg_conf_ctx *ctx)
787
	__acquires(rcu) __acquires(disk->queue->queue_lock)
788
{
789
	struct gendisk *disk;
T
Tejun Heo 已提交
790
	struct blkcg_gq *blkg;
791
	struct module *owner;
T
Tejun Heo 已提交
792
	unsigned int major, minor;
793 794
	int key_len, part, ret;
	char *body;
795

796
	if (sscanf(input, "%u:%u%n", &major, &minor, &key_len) != 2)
T
Tejun Heo 已提交
797
		return -EINVAL;
798

799 800 801 802 803
	body = input + key_len;
	if (!isspace(*body))
		return -EINVAL;
	body = skip_spaces(body);

T
Tejun Heo 已提交
804
	disk = get_gendisk(MKDEV(major, minor), &part);
805
	if (!disk)
806
		return -ENODEV;
807
	if (part) {
808
		owner = disk->fops->owner;
809
		put_disk(disk);
810
		module_put(owner);
811
		return -ENODEV;
812
	}
813 814

	rcu_read_lock();
T
Tejun Heo 已提交
815
	spin_lock_irq(disk->queue->queue_lock);
816

817
	if (blkcg_policy_enabled(disk->queue, pol))
818
		blkg = blkg_lookup_create(blkcg, disk->queue);
819
	else
820
		blkg = ERR_PTR(-EOPNOTSUPP);
821

T
Tejun Heo 已提交
822 823
	if (IS_ERR(blkg)) {
		ret = PTR_ERR(blkg);
824
		rcu_read_unlock();
825
		spin_unlock_irq(disk->queue->queue_lock);
826
		owner = disk->fops->owner;
827
		put_disk(disk);
828
		module_put(owner);
829 830 831 832 833 834 835 836 837
		/*
		 * If queue was bypassing, we should retry.  Do so after a
		 * short msleep().  It isn't strictly necessary but queue
		 * can be bypassing for some time and it's always nice to
		 * avoid busy looping.
		 */
		if (ret == -EBUSY) {
			msleep(10);
			ret = restart_syscall();
838
		}
T
Tejun Heo 已提交
839
		return ret;
840
	}
841 842 843

	ctx->disk = disk;
	ctx->blkg = blkg;
844
	ctx->body = body;
T
Tejun Heo 已提交
845
	return 0;
846
}
847
EXPORT_SYMBOL_GPL(blkg_conf_prep);
848

849 850 851 852 853 854 855
/**
 * blkg_conf_finish - finish up per-blkg config update
 * @ctx: blkg_conf_ctx intiailized by blkg_conf_prep()
 *
 * Finish up after per-blkg config update.  This function must be paired
 * with blkg_conf_prep().
 */
856
void blkg_conf_finish(struct blkg_conf_ctx *ctx)
857
	__releases(ctx->disk->queue->queue_lock) __releases(rcu)
858
{
859 860
	struct module *owner;

861
	spin_unlock_irq(ctx->disk->queue->queue_lock);
862
	rcu_read_unlock();
863
	owner = ctx->disk->fops->owner;
864
	put_disk(ctx->disk);
865
	module_put(owner);
866
}
867
EXPORT_SYMBOL_GPL(blkg_conf_finish);
868

869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907
static int blkcg_print_stat(struct seq_file *sf, void *v)
{
	struct blkcg *blkcg = css_to_blkcg(seq_css(sf));
	struct blkcg_gq *blkg;

	rcu_read_lock();

	hlist_for_each_entry_rcu(blkg, &blkcg->blkg_list, blkcg_node) {
		const char *dname;
		struct blkg_rwstat rwstat;
		u64 rbytes, wbytes, rios, wios;

		dname = blkg_dev_name(blkg);
		if (!dname)
			continue;

		spin_lock_irq(blkg->q->queue_lock);

		rwstat = blkg_rwstat_recursive_sum(blkg, NULL,
					offsetof(struct blkcg_gq, stat_bytes));
		rbytes = atomic64_read(&rwstat.aux_cnt[BLKG_RWSTAT_READ]);
		wbytes = atomic64_read(&rwstat.aux_cnt[BLKG_RWSTAT_WRITE]);

		rwstat = blkg_rwstat_recursive_sum(blkg, NULL,
					offsetof(struct blkcg_gq, stat_ios));
		rios = atomic64_read(&rwstat.aux_cnt[BLKG_RWSTAT_READ]);
		wios = atomic64_read(&rwstat.aux_cnt[BLKG_RWSTAT_WRITE]);

		spin_unlock_irq(blkg->q->queue_lock);

		if (rbytes || wbytes || rios || wios)
			seq_printf(sf, "%s rbytes=%llu wbytes=%llu rios=%llu wios=%llu\n",
				   dname, rbytes, wbytes, rios, wios);
	}

	rcu_read_unlock();
	return 0;
}

908
static struct cftype blkcg_files[] = {
909 910
	{
		.name = "stat",
911
		.flags = CFTYPE_NOT_ON_ROOT,
912 913 914 915 916
		.seq_show = blkcg_print_stat,
	},
	{ }	/* terminate */
};

917
static struct cftype blkcg_legacy_files[] = {
918 919
	{
		.name = "reset_stats",
T
Tejun Heo 已提交
920
		.write_u64 = blkcg_reset_stats,
921
	},
922
	{ }	/* terminate */
923 924
};

925
/**
926
 * blkcg_css_offline - cgroup css_offline callback
927
 * @css: css of interest
928
 *
929 930
 * This function is called when @css is about to go away and responsible
 * for shooting down all blkgs associated with @css.  blkgs should be
931 932 933 934 935
 * removed while holding both q and blkcg locks.  As blkcg lock is nested
 * inside q lock, this function performs reverse double lock dancing.
 *
 * This is the blkcg counterpart of ioc_release_fn().
 */
936
static void blkcg_css_offline(struct cgroup_subsys_state *css)
937
{
938
	struct blkcg *blkcg = css_to_blkcg(css);
939

940
	spin_lock_irq(&blkcg->lock);
941

942
	while (!hlist_empty(&blkcg->blkg_list)) {
T
Tejun Heo 已提交
943 944
		struct blkcg_gq *blkg = hlist_entry(blkcg->blkg_list.first,
						struct blkcg_gq, blkcg_node);
T
Tejun Heo 已提交
945
		struct request_queue *q = blkg->q;
946

947 948 949 950 951 952
		if (spin_trylock(q->queue_lock)) {
			blkg_destroy(blkg);
			spin_unlock(q->queue_lock);
		} else {
			spin_unlock_irq(&blkcg->lock);
			cpu_relax();
953
			spin_lock_irq(&blkcg->lock);
954
		}
955
	}
956

957
	spin_unlock_irq(&blkcg->lock);
958 959

	wb_blkcg_offline(blkcg);
960 961
}

962
static void blkcg_css_free(struct cgroup_subsys_state *css)
963
{
964
	struct blkcg *blkcg = css_to_blkcg(css);
965
	int i;
966

T
Tejun Heo 已提交
967
	mutex_lock(&blkcg_pol_mutex);
968

T
Tejun Heo 已提交
969 970
	list_del(&blkcg->all_blkcgs_node);

971
	for (i = 0; i < BLKCG_MAX_POLS; i++)
972 973 974 975 976
		if (blkcg->cpd[i])
			blkcg_policy[i]->cpd_free_fn(blkcg->cpd[i]);

	mutex_unlock(&blkcg_pol_mutex);

977
	kfree(blkcg);
978 979
}

980 981
static struct cgroup_subsys_state *
blkcg_css_alloc(struct cgroup_subsys_state *parent_css)
982
{
T
Tejun Heo 已提交
983
	struct blkcg *blkcg;
984 985
	struct cgroup_subsys_state *ret;
	int i;
986

T
Tejun Heo 已提交
987 988
	mutex_lock(&blkcg_pol_mutex);

989
	if (!parent_css) {
T
Tejun Heo 已提交
990
		blkcg = &blkcg_root;
991 992 993 994 995 996
	} else {
		blkcg = kzalloc(sizeof(*blkcg), GFP_KERNEL);
		if (!blkcg) {
			ret = ERR_PTR(-ENOMEM);
			goto free_blkcg;
		}
997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008
	}

	for (i = 0; i < BLKCG_MAX_POLS ; i++) {
		struct blkcg_policy *pol = blkcg_policy[i];
		struct blkcg_policy_data *cpd;

		/*
		 * If the policy hasn't been attached yet, wait for it
		 * to be attached before doing anything else. Otherwise,
		 * check if the policy requires any specific per-cgroup
		 * data: if it does, allocate and initialize it.
		 */
1009
		if (!pol || !pol->cpd_alloc_fn)
1010 1011
			continue;

1012
		cpd = pol->cpd_alloc_fn(GFP_KERNEL);
1013 1014 1015 1016
		if (!cpd) {
			ret = ERR_PTR(-ENOMEM);
			goto free_pd_blkcg;
		}
1017 1018
		blkcg->cpd[i] = cpd;
		cpd->blkcg = blkcg;
1019
		cpd->plid = i;
1020 1021
		if (pol->cpd_init_fn)
			pol->cpd_init_fn(cpd);
1022
	}
1023 1024

	spin_lock_init(&blkcg->lock);
1025
	INIT_RADIX_TREE(&blkcg->blkg_tree, GFP_NOWAIT);
1026
	INIT_HLIST_HEAD(&blkcg->blkg_list);
1027 1028 1029
#ifdef CONFIG_CGROUP_WRITEBACK
	INIT_LIST_HEAD(&blkcg->cgwb_list);
#endif
T
Tejun Heo 已提交
1030 1031 1032
	list_add_tail(&blkcg->all_blkcgs_node, &all_blkcgs);

	mutex_unlock(&blkcg_pol_mutex);
1033
	return &blkcg->css;
1034 1035 1036

free_pd_blkcg:
	for (i--; i >= 0; i--)
1037 1038
		if (blkcg->cpd[i])
			blkcg_policy[i]->cpd_free_fn(blkcg->cpd[i]);
1039 1040
free_blkcg:
	kfree(blkcg);
T
Tejun Heo 已提交
1041
	mutex_unlock(&blkcg_pol_mutex);
1042
	return ret;
1043 1044
}

1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056
/**
 * blkcg_init_queue - initialize blkcg part of request queue
 * @q: request_queue to initialize
 *
 * Called from blk_alloc_queue_node(). Responsible for initializing blkcg
 * part of new request_queue @q.
 *
 * RETURNS:
 * 0 on success, -errno on failure.
 */
int blkcg_init_queue(struct request_queue *q)
{
1057 1058 1059 1060 1061 1062 1063 1064 1065
	struct blkcg_gq *new_blkg, *blkg;
	bool preloaded;
	int ret;

	new_blkg = blkg_alloc(&blkcg_root, q, GFP_KERNEL);
	if (!new_blkg)
		return -ENOMEM;

	preloaded = !radix_tree_preload(GFP_KERNEL);
1066

1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081
	/*
	 * Make sure the root blkg exists and count the existing blkgs.  As
	 * @q is bypassing at this point, blkg_lookup_create() can't be
	 * used.  Open code insertion.
	 */
	rcu_read_lock();
	spin_lock_irq(q->queue_lock);
	blkg = blkg_create(&blkcg_root, q, new_blkg);
	spin_unlock_irq(q->queue_lock);
	rcu_read_unlock();

	if (preloaded)
		radix_tree_preload_end();

	if (IS_ERR(blkg)) {
1082
		blkg_free(new_blkg);
1083 1084 1085 1086 1087
		return PTR_ERR(blkg);
	}

	q->root_blkg = blkg;
	q->root_rl.blkg = blkg;
1088

1089 1090 1091 1092 1093 1094 1095
	ret = blk_throtl_init(q);
	if (ret) {
		spin_lock_irq(q->queue_lock);
		blkg_destroy_all(q);
		spin_unlock_irq(q->queue_lock);
	}
	return ret;
1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107
}

/**
 * blkcg_drain_queue - drain blkcg part of request_queue
 * @q: request_queue to drain
 *
 * Called from blk_drain_queue().  Responsible for draining blkcg part.
 */
void blkcg_drain_queue(struct request_queue *q)
{
	lockdep_assert_held(q->queue_lock);

1108 1109 1110 1111 1112 1113 1114
	/*
	 * @q could be exiting and already have destroyed all blkgs as
	 * indicated by NULL root_blkg.  If so, don't confuse policies.
	 */
	if (!q->root_blkg)
		return;

1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125
	blk_throtl_drain(q);
}

/**
 * blkcg_exit_queue - exit and release blkcg part of request_queue
 * @q: request_queue being released
 *
 * Called from blk_release_queue().  Responsible for exiting blkcg part.
 */
void blkcg_exit_queue(struct request_queue *q)
{
1126
	spin_lock_irq(q->queue_lock);
1127
	blkg_destroy_all(q);
1128 1129
	spin_unlock_irq(q->queue_lock);

1130 1131 1132
	blk_throtl_exit(q);
}

1133 1134 1135 1136 1137 1138
/*
 * We cannot support shared io contexts, as we have no mean to support
 * two tasks with the same ioc in two different groups without major rework
 * of the main cic data structures.  For now we allow a task to change
 * its cgroup only if it's the only owner of its ioc.
 */
1139
static int blkcg_can_attach(struct cgroup_taskset *tset)
1140
{
1141
	struct task_struct *task;
1142
	struct cgroup_subsys_state *dst_css;
1143 1144 1145 1146
	struct io_context *ioc;
	int ret = 0;

	/* task_lock() is needed to avoid races with exit_io_context() */
1147
	cgroup_taskset_for_each(task, dst_css, tset) {
1148 1149 1150 1151 1152 1153 1154 1155
		task_lock(task);
		ioc = task->io_context;
		if (ioc && atomic_read(&ioc->nr_tasks) > 1)
			ret = -EINVAL;
		task_unlock(task);
		if (ret)
			break;
	}
1156 1157 1158
	return ret;
}

1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178
static void blkcg_bind(struct cgroup_subsys_state *root_css)
{
	int i;

	mutex_lock(&blkcg_pol_mutex);

	for (i = 0; i < BLKCG_MAX_POLS; i++) {
		struct blkcg_policy *pol = blkcg_policy[i];
		struct blkcg *blkcg;

		if (!pol || !pol->cpd_bind_fn)
			continue;

		list_for_each_entry(blkcg, &all_blkcgs, all_blkcgs_node)
			if (blkcg->cpd[pol->plid])
				pol->cpd_bind_fn(blkcg->cpd[pol->plid]);
	}
	mutex_unlock(&blkcg_pol_mutex);
}

1179
struct cgroup_subsys io_cgrp_subsys = {
1180 1181 1182
	.css_alloc = blkcg_css_alloc,
	.css_offline = blkcg_css_offline,
	.css_free = blkcg_css_free,
T
Tejun Heo 已提交
1183
	.can_attach = blkcg_can_attach,
1184
	.bind = blkcg_bind,
1185
	.dfl_cftypes = blkcg_files,
1186
	.legacy_cftypes = blkcg_legacy_files,
1187
	.legacy_name = "blkio",
1188 1189 1190 1191 1192 1193 1194 1195
#ifdef CONFIG_MEMCG
	/*
	 * This ensures that, if available, memcg is automatically enabled
	 * together on the default hierarchy so that the owner cgroup can
	 * be retrieved from writeback pages.
	 */
	.depends_on = 1 << memory_cgrp_id,
#endif
1196
};
1197
EXPORT_SYMBOL_GPL(io_cgrp_subsys);
1198

1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215
/**
 * blkcg_activate_policy - activate a blkcg policy on a request_queue
 * @q: request_queue of interest
 * @pol: blkcg policy to activate
 *
 * Activate @pol on @q.  Requires %GFP_KERNEL context.  @q goes through
 * bypass mode to populate its blkgs with policy_data for @pol.
 *
 * Activation happens with @q bypassed, so nobody would be accessing blkgs
 * from IO path.  Update of each blkg is protected by both queue and blkcg
 * locks so that holding either lock and testing blkcg_policy_enabled() is
 * always enough for dereferencing policy data.
 *
 * The caller is responsible for synchronizing [de]activations and policy
 * [un]registerations.  Returns 0 on success, -errno on failure.
 */
int blkcg_activate_policy(struct request_queue *q,
T
Tejun Heo 已提交
1216
			  const struct blkcg_policy *pol)
1217
{
1218
	struct blkg_policy_data *pd_prealloc = NULL;
1219
	struct blkcg_gq *blkg;
1220
	int ret;
1221 1222 1223 1224 1225

	if (blkcg_policy_enabled(q, pol))
		return 0;

	blk_queue_bypass_start(q);
1226 1227
pd_prealloc:
	if (!pd_prealloc) {
1228
		pd_prealloc = pol->pd_alloc_fn(GFP_KERNEL, q->node);
1229
		if (!pd_prealloc) {
1230
			ret = -ENOMEM;
1231
			goto out_bypass_end;
1232 1233 1234 1235 1236 1237
		}
	}

	spin_lock_irq(q->queue_lock);

	list_for_each_entry(blkg, &q->blkg_list, q_node) {
1238 1239 1240 1241
		struct blkg_policy_data *pd;

		if (blkg->pd[pol->plid])
			continue;
1242

1243
		pd = pol->pd_alloc_fn(GFP_NOWAIT, q->node);
1244 1245 1246 1247 1248 1249
		if (!pd)
			swap(pd, pd_prealloc);
		if (!pd) {
			spin_unlock_irq(q->queue_lock);
			goto pd_prealloc;
		}
1250 1251 1252

		blkg->pd[pol->plid] = pd;
		pd->blkg = blkg;
T
Tejun Heo 已提交
1253
		pd->plid = pol->plid;
1254
		if (pol->pd_init_fn)
1255
			pol->pd_init_fn(pd);
1256 1257 1258 1259
	}

	__set_bit(pol->plid, q->blkcg_pols);
	ret = 0;
1260

1261
	spin_unlock_irq(q->queue_lock);
1262
out_bypass_end:
1263
	blk_queue_bypass_end(q);
1264 1265
	if (pd_prealloc)
		pol->pd_free_fn(pd_prealloc);
1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278
	return ret;
}
EXPORT_SYMBOL_GPL(blkcg_activate_policy);

/**
 * blkcg_deactivate_policy - deactivate a blkcg policy on a request_queue
 * @q: request_queue of interest
 * @pol: blkcg policy to deactivate
 *
 * Deactivate @pol on @q.  Follows the same synchronization rules as
 * blkcg_activate_policy().
 */
void blkcg_deactivate_policy(struct request_queue *q,
T
Tejun Heo 已提交
1279
			     const struct blkcg_policy *pol)
1280
{
T
Tejun Heo 已提交
1281
	struct blkcg_gq *blkg;
1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294

	if (!blkcg_policy_enabled(q, pol))
		return;

	blk_queue_bypass_start(q);
	spin_lock_irq(q->queue_lock);

	__clear_bit(pol->plid, q->blkcg_pols);

	list_for_each_entry(blkg, &q->blkg_list, q_node) {
		/* grab blkcg lock too while removing @pd from @blkg */
		spin_lock(&blkg->blkcg->lock);

1295
		if (blkg->pd[pol->plid]) {
1296 1297
			if (pol->pd_offline_fn)
				pol->pd_offline_fn(blkg->pd[pol->plid]);
1298 1299 1300
			pol->pd_free_fn(blkg->pd[pol->plid]);
			blkg->pd[pol->plid] = NULL;
		}
1301 1302 1303 1304 1305 1306 1307 1308 1309

		spin_unlock(&blkg->blkcg->lock);
	}

	spin_unlock_irq(q->queue_lock);
	blk_queue_bypass_end(q);
}
EXPORT_SYMBOL_GPL(blkcg_deactivate_policy);

T
Tejun Heo 已提交
1310
/**
T
Tejun Heo 已提交
1311 1312
 * blkcg_policy_register - register a blkcg policy
 * @pol: blkcg policy to register
T
Tejun Heo 已提交
1313
 *
T
Tejun Heo 已提交
1314 1315
 * Register @pol with blkcg core.  Might sleep and @pol may be modified on
 * successful registration.  Returns 0 on success and -errno on failure.
T
Tejun Heo 已提交
1316
 */
1317
int blkcg_policy_register(struct blkcg_policy *pol)
1318
{
1319
	struct blkcg *blkcg;
T
Tejun Heo 已提交
1320
	int i, ret;
1321

1322
	mutex_lock(&blkcg_pol_register_mutex);
1323 1324
	mutex_lock(&blkcg_pol_mutex);

T
Tejun Heo 已提交
1325 1326 1327
	/* find an empty slot */
	ret = -ENOSPC;
	for (i = 0; i < BLKCG_MAX_POLS; i++)
T
Tejun Heo 已提交
1328
		if (!blkcg_policy[i])
T
Tejun Heo 已提交
1329 1330
			break;
	if (i >= BLKCG_MAX_POLS)
1331
		goto err_unlock;
1332

1333
	/* register @pol */
T
Tejun Heo 已提交
1334
	pol->plid = i;
1335 1336 1337
	blkcg_policy[pol->plid] = pol;

	/* allocate and install cpd's */
1338
	if (pol->cpd_alloc_fn) {
1339 1340 1341
		list_for_each_entry(blkcg, &all_blkcgs, all_blkcgs_node) {
			struct blkcg_policy_data *cpd;

1342
			cpd = pol->cpd_alloc_fn(GFP_KERNEL);
1343
			if (!cpd)
1344 1345
				goto err_free_cpds;

1346 1347
			blkcg->cpd[pol->plid] = cpd;
			cpd->blkcg = blkcg;
1348
			cpd->plid = pol->plid;
1349
			pol->cpd_init_fn(cpd);
1350 1351 1352
		}
	}

1353
	mutex_unlock(&blkcg_pol_mutex);
T
Tejun Heo 已提交
1354 1355

	/* everything is in place, add intf files for the new policy */
1356 1357 1358
	if (pol->dfl_cftypes)
		WARN_ON(cgroup_add_dfl_cftypes(&io_cgrp_subsys,
					       pol->dfl_cftypes));
1359
	if (pol->legacy_cftypes)
1360
		WARN_ON(cgroup_add_legacy_cftypes(&io_cgrp_subsys,
1361
						  pol->legacy_cftypes));
1362 1363 1364
	mutex_unlock(&blkcg_pol_register_mutex);
	return 0;

1365
err_free_cpds:
1366
	if (pol->cpd_alloc_fn) {
1367
		list_for_each_entry(blkcg, &all_blkcgs, all_blkcgs_node) {
1368 1369 1370 1371
			if (blkcg->cpd[pol->plid]) {
				pol->cpd_free_fn(blkcg->cpd[pol->plid]);
				blkcg->cpd[pol->plid] = NULL;
			}
1372 1373 1374
		}
	}
	blkcg_policy[pol->plid] = NULL;
1375
err_unlock:
1376
	mutex_unlock(&blkcg_pol_mutex);
1377
	mutex_unlock(&blkcg_pol_register_mutex);
T
Tejun Heo 已提交
1378
	return ret;
1379
}
T
Tejun Heo 已提交
1380
EXPORT_SYMBOL_GPL(blkcg_policy_register);
1381

T
Tejun Heo 已提交
1382
/**
T
Tejun Heo 已提交
1383 1384
 * blkcg_policy_unregister - unregister a blkcg policy
 * @pol: blkcg policy to unregister
T
Tejun Heo 已提交
1385
 *
T
Tejun Heo 已提交
1386
 * Undo blkcg_policy_register(@pol).  Might sleep.
T
Tejun Heo 已提交
1387
 */
T
Tejun Heo 已提交
1388
void blkcg_policy_unregister(struct blkcg_policy *pol)
1389
{
1390 1391
	struct blkcg *blkcg;

1392
	mutex_lock(&blkcg_pol_register_mutex);
1393

T
Tejun Heo 已提交
1394
	if (WARN_ON(blkcg_policy[pol->plid] != pol))
T
Tejun Heo 已提交
1395 1396 1397
		goto out_unlock;

	/* kill the intf files first */
1398 1399
	if (pol->dfl_cftypes)
		cgroup_rm_cftypes(pol->dfl_cftypes);
1400 1401
	if (pol->legacy_cftypes)
		cgroup_rm_cftypes(pol->legacy_cftypes);
1402

1403
	/* remove cpds and unregister */
1404
	mutex_lock(&blkcg_pol_mutex);
1405

1406
	if (pol->cpd_alloc_fn) {
1407
		list_for_each_entry(blkcg, &all_blkcgs, all_blkcgs_node) {
1408 1409 1410 1411
			if (blkcg->cpd[pol->plid]) {
				pol->cpd_free_fn(blkcg->cpd[pol->plid]);
				blkcg->cpd[pol->plid] = NULL;
			}
1412 1413
		}
	}
T
Tejun Heo 已提交
1414
	blkcg_policy[pol->plid] = NULL;
1415

1416
	mutex_unlock(&blkcg_pol_mutex);
1417 1418
out_unlock:
	mutex_unlock(&blkcg_pol_register_mutex);
1419
}
T
Tejun Heo 已提交
1420
EXPORT_SYMBOL_GPL(blkcg_policy_unregister);