blk-cgroup.c 31.3 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/blk-cgroup.h>
28
#include "blk.h"
29

30 31
#define MAX_KEY_LEN 100

32 33 34 35 36 37 38 39
/*
 * 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);
40
static DEFINE_MUTEX(blkcg_pol_mutex);
41

42
struct blkcg blkcg_root;
T
Tejun Heo 已提交
43
EXPORT_SYMBOL_GPL(blkcg_root);
44

T
Tejun Heo 已提交
45 46
struct cgroup_subsys_state * const blkcg_root_css = &blkcg_root.css;

T
Tejun Heo 已提交
47
static struct blkcg_policy *blkcg_policy[BLKCG_MAX_POLS];
48

T
Tejun Heo 已提交
49 50
static LIST_HEAD(all_blkcgs);		/* protected by blkcg_pol_mutex */

51
static bool blkcg_policy_enabled(struct request_queue *q,
T
Tejun Heo 已提交
52
				 const struct blkcg_policy *pol)
53 54 55 56
{
	return pol && test_bit(pol->plid, q->blkcg_pols);
}

57 58 59 60 61 62
/**
 * blkg_free - free a blkg
 * @blkg: blkg to free
 *
 * Free @blkg which may be partially allocated.
 */
T
Tejun Heo 已提交
63
static void blkg_free(struct blkcg_gq *blkg)
64
{
65
	int i;
66 67 68 69

	if (!blkg)
		return;

70 71
	for (i = 0; i < BLKCG_MAX_POLS; i++)
		kfree(blkg->pd[i]);
72

73 74
	if (blkg->blkcg != &blkcg_root)
		blk_exit_rl(&blkg->rl);
75
	kfree(blkg);
76 77 78 79 80 81
}

/**
 * blkg_alloc - allocate a blkg
 * @blkcg: block cgroup the new blkg is associated with
 * @q: request_queue the new blkg is associated with
82
 * @gfp_mask: allocation mask to use
83
 *
84
 * Allocate a new blkg assocating @blkcg and @q.
85
 */
86 87
static struct blkcg_gq *blkg_alloc(struct blkcg *blkcg, struct request_queue *q,
				   gfp_t gfp_mask)
88
{
T
Tejun Heo 已提交
89
	struct blkcg_gq *blkg;
90
	int i;
91 92

	/* alloc and init base part */
93
	blkg = kzalloc_node(sizeof(*blkg), gfp_mask, q->node);
94 95 96
	if (!blkg)
		return NULL;

T
Tejun Heo 已提交
97
	blkg->q = q;
98
	INIT_LIST_HEAD(&blkg->q_node);
99
	blkg->blkcg = blkcg;
100
	atomic_set(&blkg->refcnt, 1);
101

102 103 104 105 106 107 108
	/* 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
109
	for (i = 0; i < BLKCG_MAX_POLS; i++) {
T
Tejun Heo 已提交
110
		struct blkcg_policy *pol = blkcg_policy[i];
111
		struct blkg_policy_data *pd;
112

113
		if (!blkcg_policy_enabled(q, pol))
114 115 116
			continue;

		/* alloc per-policy data and attach it to blkg */
117
		pd = kzalloc_node(pol->pd_size, gfp_mask, q->node);
118 119
		if (!pd)
			goto err_free;
120

121 122
		blkg->pd[i] = pd;
		pd->blkg = blkg;
T
Tejun Heo 已提交
123
		pd->plid = i;
124 125
	}

126
	return blkg;
127 128 129 130

err_free:
	blkg_free(blkg);
	return NULL;
131 132
}

133 134 135 136 137 138 139 140 141 142 143
/**
 * __blkg_lookup - internal version of blkg_lookup()
 * @blkcg: blkcg of interest
 * @q: request_queue of interest
 * @update_hint: whether to update lookup hint with the result or not
 *
 * This is internal version and shouldn't be used by policy
 * implementations.  Looks up blkgs for the @blkcg - @q pair regardless of
 * @q's bypass state.  If @update_hint is %true, the caller should be
 * holding @q->queue_lock and lookup hint is updated on success.
 */
144 145
struct blkcg_gq *__blkg_lookup(struct blkcg *blkcg, struct request_queue *q,
			       bool update_hint)
146
{
T
Tejun Heo 已提交
147
	struct blkcg_gq *blkg;
148

149 150 151 152 153
	blkg = rcu_dereference(blkcg->blkg_hint);
	if (blkg && blkg->q == q)
		return blkg;

	/*
154 155 156 157
	 * 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.
158 159
	 */
	blkg = radix_tree_lookup(&blkcg->blkg_tree, q->id);
160 161 162 163 164
	if (blkg && blkg->q == q) {
		if (update_hint) {
			lockdep_assert_held(q->queue_lock);
			rcu_assign_pointer(blkcg->blkg_hint, blkg);
		}
165
		return blkg;
166
	}
167

168 169 170 171 172 173 174 175 176 177 178 179
	return NULL;
}

/**
 * blkg_lookup - lookup blkg for the specified blkcg - q pair
 * @blkcg: blkcg of interest
 * @q: request_queue of interest
 *
 * Lookup blkg for the @blkcg - @q pair.  This function should be called
 * under RCU read lock and is guaranteed to return %NULL if @q is bypassing
 * - see blk_queue_bypass_start() for details.
 */
T
Tejun Heo 已提交
180
struct blkcg_gq *blkg_lookup(struct blkcg *blkcg, struct request_queue *q)
181 182 183 184 185
{
	WARN_ON_ONCE(!rcu_read_lock_held());

	if (unlikely(blk_queue_bypass(q)))
		return NULL;
186
	return __blkg_lookup(blkcg, q, false);
187 188 189
}
EXPORT_SYMBOL_GPL(blkg_lookup);

190 191
/*
 * If @new_blkg is %NULL, this function tries to allocate a new one as
192
 * necessary using %GFP_NOWAIT.  @new_blkg is always consumed on return.
193
 */
194 195 196
static struct blkcg_gq *blkg_create(struct blkcg *blkcg,
				    struct request_queue *q,
				    struct blkcg_gq *new_blkg)
197
{
T
Tejun Heo 已提交
198
	struct blkcg_gq *blkg;
199
	struct bdi_writeback_congested *wb_congested;
200
	int i, ret;
201

202 203 204
	WARN_ON_ONCE(!rcu_read_lock_held());
	lockdep_assert_held(q->queue_lock);

205
	/* blkg holds a reference to blkcg */
206
	if (!css_tryget_online(&blkcg->css)) {
207 208
		ret = -EINVAL;
		goto err_free_blkg;
209
	}
210

211
	wb_congested = wb_congested_get_create(&q->backing_dev_info,
212
					       blkcg->css.id, GFP_NOWAIT);
213 214 215 216 217
	if (!wb_congested) {
		ret = -ENOMEM;
		goto err_put_css;
	}

218
	/* allocate */
219
	if (!new_blkg) {
220
		new_blkg = blkg_alloc(blkcg, q, GFP_NOWAIT);
221
		if (unlikely(!new_blkg)) {
222
			ret = -ENOMEM;
223
			goto err_put_congested;
224 225 226
		}
	}
	blkg = new_blkg;
227
	blkg->wb_congested = wb_congested;
228

229
	/* link parent */
T
Tejun Heo 已提交
230 231 232
	if (blkcg_parent(blkcg)) {
		blkg->parent = __blkg_lookup(blkcg_parent(blkcg), q, false);
		if (WARN_ON_ONCE(!blkg->parent)) {
233
			ret = -EINVAL;
234
			goto err_put_congested;
T
Tejun Heo 已提交
235 236 237 238
		}
		blkg_get(blkg->parent);
	}

239 240 241 242 243 244 245 246 247
	/* 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)
			pol->pd_init_fn(blkg);
	}

	/* insert */
248
	spin_lock(&blkcg->lock);
249 250 251 252
	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);
253 254 255 256 257 258 259

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

			if (blkg->pd[i] && pol->pd_online_fn)
				pol->pd_online_fn(blkg);
		}
260
	}
261
	blkg->online = true;
262
	spin_unlock(&blkcg->lock);
263

264
	if (!ret)
265
		return blkg;
266

T
Tejun Heo 已提交
267 268 269 270
	/* @blkg failed fully initialized, use the usual release path */
	blkg_put(blkg);
	return ERR_PTR(ret);

271 272
err_put_congested:
	wb_congested_put(wb_congested);
273
err_put_css:
274
	css_put(&blkcg->css);
275
err_free_blkg:
276
	blkg_free(new_blkg);
277
	return ERR_PTR(ret);
278
}
279

280 281 282 283 284 285
/**
 * 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
286 287 288
 * 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.
289 290 291 292 293
 *
 * 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
294 295
struct blkcg_gq *blkg_lookup_create(struct blkcg *blkcg,
				    struct request_queue *q)
296
{
297 298 299 300 301
	struct blkcg_gq *blkg;

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

302 303 304 305 306
	/*
	 * 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)))
B
Bart Van Assche 已提交
307
		return ERR_PTR(blk_queue_dying(q) ? -EINVAL : -EBUSY);
308 309 310 311 312

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

T
Tejun Heo 已提交
313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329
	/*
	 * 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;
	}
330
}
331
EXPORT_SYMBOL_GPL(blkg_lookup_create);
332

T
Tejun Heo 已提交
333
static void blkg_destroy(struct blkcg_gq *blkg)
334
{
T
Tejun Heo 已提交
335
	struct blkcg *blkcg = blkg->blkcg;
336
	int i;
337

338
	lockdep_assert_held(blkg->q->queue_lock);
339
	lockdep_assert_held(&blkcg->lock);
340 341

	/* Something wrong if we are trying to remove same group twice */
342
	WARN_ON_ONCE(list_empty(&blkg->q_node));
343
	WARN_ON_ONCE(hlist_unhashed(&blkg->blkcg_node));
344

345 346 347 348 349 350 351 352
	for (i = 0; i < BLKCG_MAX_POLS; i++) {
		struct blkcg_policy *pol = blkcg_policy[i];

		if (blkg->pd[i] && pol->pd_offline_fn)
			pol->pd_offline_fn(blkg);
	}
	blkg->online = false;

353
	radix_tree_delete(&blkcg->blkg_tree, blkg->q->id);
354
	list_del_init(&blkg->q_node);
355
	hlist_del_init_rcu(&blkg->blkcg_node);
356

357 358 359 360 361
	/*
	 * 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.
	 */
362
	if (rcu_access_pointer(blkcg->blkg_hint) == blkg)
363 364
		rcu_assign_pointer(blkcg->blkg_hint, NULL);

365 366 367 368 369 370 371
	/*
	 * Put the reference taken at the time of creation so that when all
	 * queues are gone, group can be destroyed.
	 */
	blkg_put(blkg);
}

372 373 374 375
/**
 * blkg_destroy_all - destroy all blkgs associated with a request_queue
 * @q: request_queue of interest
 *
376
 * Destroy all blkgs associated with @q.
377
 */
378
static void blkg_destroy_all(struct request_queue *q)
379
{
T
Tejun Heo 已提交
380
	struct blkcg_gq *blkg, *n;
381

382
	lockdep_assert_held(q->queue_lock);
383

384
	list_for_each_entry_safe(blkg, n, &q->blkg_list, q_node) {
T
Tejun Heo 已提交
385
		struct blkcg *blkcg = blkg->blkcg;
386

387 388 389
		spin_lock(&blkcg->lock);
		blkg_destroy(blkg);
		spin_unlock(&blkcg->lock);
390 391 392
	}
}

393 394 395 396 397 398 399 400 401
/*
 * 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
402
{
403
	struct blkcg_gq *blkg = container_of(rcu_head, struct blkcg_gq, rcu_head);
404 405 406 407 408 409 410 411 412 413
	int i;

	/* tell policies that this one is being freed */
	for (i = 0; i < BLKCG_MAX_POLS; i++) {
		struct blkcg_policy *pol = blkcg_policy[i];

		if (blkg->pd[i] && pol->pd_exit_fn)
			pol->pd_exit_fn(blkg);
	}

T
Tejun Heo 已提交
414
	/* release the blkcg and parent blkg refs this blkg has been holding */
T
Tejun Heo 已提交
415
	css_put(&blkg->blkcg->css);
416
	if (blkg->parent)
T
Tejun Heo 已提交
417
		blkg_put(blkg->parent);
T
Tejun Heo 已提交
418

419 420
	wb_congested_put(blkg->wb_congested);

421
	blkg_free(blkg);
T
Tejun Heo 已提交
422
}
423
EXPORT_SYMBOL_GPL(__blkg_release_rcu);
T
Tejun Heo 已提交
424

425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440
/*
 * 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;
441 442 443
		/* There are no more block groups, hence no request lists */
		if (list_empty(ent))
			return NULL;
444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459
	} 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;
}

460 461
static int blkcg_reset_stats(struct cgroup_subsys_state *css,
			     struct cftype *cftype, u64 val)
462
{
463
	struct blkcg *blkcg = css_to_blkcg(css);
T
Tejun Heo 已提交
464
	struct blkcg_gq *blkg;
465
	int i;
466

467
	mutex_lock(&blkcg_pol_mutex);
468
	spin_lock_irq(&blkcg->lock);
T
Tejun Heo 已提交
469 470 471 472 473 474

	/*
	 * 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.
	 */
475
	hlist_for_each_entry(blkg, &blkcg->blkg_list, blkcg_node) {
T
Tejun Heo 已提交
476
		for (i = 0; i < BLKCG_MAX_POLS; i++) {
T
Tejun Heo 已提交
477
			struct blkcg_policy *pol = blkcg_policy[i];
478

479
			if (blkcg_policy_enabled(blkg->q, pol) &&
480 481
			    pol->pd_reset_stats_fn)
				pol->pd_reset_stats_fn(blkg);
482
		}
483
	}
484

485
	spin_unlock_irq(&blkcg->lock);
486
	mutex_unlock(&blkcg_pol_mutex);
487 488 489
	return 0;
}

T
Tejun Heo 已提交
490
static const char *blkg_dev_name(struct blkcg_gq *blkg)
491
{
492 493 494 495
	/* 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;
496 497
}

498 499 500 501 502 503 504 505 506 507 508
/**
 * 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
509 510 511
 * 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.
512 513 514 515
 *
 * This is to be used to construct print functions for
 * cftype->read_seq_string method.
 */
T
Tejun Heo 已提交
516
void blkcg_print_blkgs(struct seq_file *sf, struct blkcg *blkcg,
517 518
		       u64 (*prfill)(struct seq_file *,
				     struct blkg_policy_data *, int),
T
Tejun Heo 已提交
519
		       const struct blkcg_policy *pol, int data,
520
		       bool show_total)
521
{
T
Tejun Heo 已提交
522
	struct blkcg_gq *blkg;
523
	u64 total = 0;
524

525
	rcu_read_lock();
526
	hlist_for_each_entry_rcu(blkg, &blkcg->blkg_list, blkcg_node) {
527
		spin_lock_irq(blkg->q->queue_lock);
528
		if (blkcg_policy_enabled(blkg->q, pol))
529
			total += prfill(sf, blkg->pd[pol->plid], data);
530 531 532
		spin_unlock_irq(blkg->q->queue_lock);
	}
	rcu_read_unlock();
533 534 535 536

	if (show_total)
		seq_printf(sf, "Total %llu\n", (unsigned long long)total);
}
537
EXPORT_SYMBOL_GPL(blkcg_print_blkgs);
538 539 540 541

/**
 * __blkg_prfill_u64 - prfill helper for a single u64 value
 * @sf: seq_file to print to
542
 * @pd: policy private data of interest
543 544
 * @v: value to print
 *
545
 * Print @v to @sf for the device assocaited with @pd.
546
 */
547
u64 __blkg_prfill_u64(struct seq_file *sf, struct blkg_policy_data *pd, u64 v)
548
{
549
	const char *dname = blkg_dev_name(pd->blkg);
550 551 552 553 554 555 556

	if (!dname)
		return 0;

	seq_printf(sf, "%s %llu\n", dname, (unsigned long long)v);
	return v;
}
557
EXPORT_SYMBOL_GPL(__blkg_prfill_u64);
558 559 560 561

/**
 * __blkg_prfill_rwstat - prfill helper for a blkg_rwstat
 * @sf: seq_file to print to
562
 * @pd: policy private data of interest
563 564
 * @rwstat: rwstat to print
 *
565
 * Print @rwstat to @sf for the device assocaited with @pd.
566
 */
567
u64 __blkg_prfill_rwstat(struct seq_file *sf, struct blkg_policy_data *pd,
568
			 const struct blkg_rwstat *rwstat)
569 570 571 572 573 574 575
{
	static const char *rwstr[] = {
		[BLKG_RWSTAT_READ]	= "Read",
		[BLKG_RWSTAT_WRITE]	= "Write",
		[BLKG_RWSTAT_SYNC]	= "Sync",
		[BLKG_RWSTAT_ASYNC]	= "Async",
	};
576
	const char *dname = blkg_dev_name(pd->blkg);
577 578 579 580 581 582 583 584 585 586 587 588 589 590
	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],
			   (unsigned long long)rwstat->cnt[i]);

	v = rwstat->cnt[BLKG_RWSTAT_READ] + rwstat->cnt[BLKG_RWSTAT_WRITE];
	seq_printf(sf, "%s Total %llu\n", dname, (unsigned long long)v);
	return v;
}
T
Tejun Heo 已提交
591
EXPORT_SYMBOL_GPL(__blkg_prfill_rwstat);
592

593 594 595
/**
 * blkg_prfill_stat - prfill callback for blkg_stat
 * @sf: seq_file to print to
596 597
 * @pd: policy private data of interest
 * @off: offset to the blkg_stat in @pd
598 599 600
 *
 * prfill callback for printing a blkg_stat.
 */
601
u64 blkg_prfill_stat(struct seq_file *sf, struct blkg_policy_data *pd, int off)
602
{
603
	return __blkg_prfill_u64(sf, pd, blkg_stat_read((void *)pd + off));
604
}
605
EXPORT_SYMBOL_GPL(blkg_prfill_stat);
606

607 608 609
/**
 * blkg_prfill_rwstat - prfill callback for blkg_rwstat
 * @sf: seq_file to print to
610 611
 * @pd: policy private data of interest
 * @off: offset to the blkg_rwstat in @pd
612 613 614
 *
 * prfill callback for printing a blkg_rwstat.
 */
615 616
u64 blkg_prfill_rwstat(struct seq_file *sf, struct blkg_policy_data *pd,
		       int off)
617
{
618
	struct blkg_rwstat rwstat = blkg_rwstat_read((void *)pd + off);
619

620
	return __blkg_prfill_rwstat(sf, pd, &rwstat);
621
}
622
EXPORT_SYMBOL_GPL(blkg_prfill_rwstat);
623

624 625 626 627 628 629 630 631 632 633 634 635 636
/**
 * blkg_stat_recursive_sum - collect hierarchical blkg_stat
 * @pd: policy private data of interest
 * @off: offset to the blkg_stat in @pd
 *
 * Collect the blkg_stat specified by @off from @pd and all its online
 * descendants and return the sum.  The caller must be holding the queue
 * lock for online tests.
 */
u64 blkg_stat_recursive_sum(struct blkg_policy_data *pd, int off)
{
	struct blkcg_policy *pol = blkcg_policy[pd->plid];
	struct blkcg_gq *pos_blkg;
637
	struct cgroup_subsys_state *pos_css;
638
	u64 sum = 0;
639 640 641 642

	lockdep_assert_held(pd->blkg->q->queue_lock);

	rcu_read_lock();
643
	blkg_for_each_descendant_pre(pos_blkg, pos_css, pd_to_blkg(pd)) {
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
		struct blkg_policy_data *pos_pd = blkg_to_pd(pos_blkg, pol);
		struct blkg_stat *stat = (void *)pos_pd + off;

		if (pos_blkg->online)
			sum += blkg_stat_read(stat);
	}
	rcu_read_unlock();

	return sum;
}
EXPORT_SYMBOL_GPL(blkg_stat_recursive_sum);

/**
 * blkg_rwstat_recursive_sum - collect hierarchical blkg_rwstat
 * @pd: policy private data of interest
 * @off: offset to the blkg_stat in @pd
 *
 * Collect the blkg_rwstat specified by @off from @pd and all its online
 * descendants and return the sum.  The caller must be holding the queue
 * lock for online tests.
 */
struct blkg_rwstat blkg_rwstat_recursive_sum(struct blkg_policy_data *pd,
					     int off)
{
	struct blkcg_policy *pol = blkcg_policy[pd->plid];
	struct blkcg_gq *pos_blkg;
670
	struct cgroup_subsys_state *pos_css;
671
	struct blkg_rwstat sum = { };
672 673 674 675 676
	int i;

	lockdep_assert_held(pd->blkg->q->queue_lock);

	rcu_read_lock();
677
	blkg_for_each_descendant_pre(pos_blkg, pos_css, pd_to_blkg(pd)) {
678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695
		struct blkg_policy_data *pos_pd = blkg_to_pd(pos_blkg, pol);
		struct blkg_rwstat *rwstat = (void *)pos_pd + off;
		struct blkg_rwstat tmp;

		if (!pos_blkg->online)
			continue;

		tmp = blkg_rwstat_read(rwstat);

		for (i = 0; i < BLKG_RWSTAT_NR; i++)
			sum.cnt[i] += tmp.cnt[i];
	}
	rcu_read_unlock();

	return sum;
}
EXPORT_SYMBOL_GPL(blkg_rwstat_recursive_sum);

696 697 698
/**
 * blkg_conf_prep - parse and prepare for per-blkg config update
 * @blkcg: target block cgroup
699
 * @pol: target policy
700 701 702 703 704
 * @input: input string
 * @ctx: blkg_conf_ctx to be filled
 *
 * Parse per-blkg config update from @input and initialize @ctx with the
 * result.  @ctx->blkg points to the blkg to be updated and @ctx->v the new
705 706
 * value.  This function returns with RCU read lock and queue lock held and
 * must be paired with blkg_conf_finish().
707
 */
T
Tejun Heo 已提交
708 709
int blkg_conf_prep(struct blkcg *blkcg, const struct blkcg_policy *pol,
		   const char *input, struct blkg_conf_ctx *ctx)
710
	__acquires(rcu) __acquires(disk->queue->queue_lock)
711
{
712
	struct gendisk *disk;
T
Tejun Heo 已提交
713
	struct blkcg_gq *blkg;
T
Tejun Heo 已提交
714 715 716
	unsigned int major, minor;
	unsigned long long v;
	int part, ret;
717

T
Tejun Heo 已提交
718 719
	if (sscanf(input, "%u:%u %llu", &major, &minor, &v) != 3)
		return -EINVAL;
720

T
Tejun Heo 已提交
721
	disk = get_gendisk(MKDEV(major, minor), &part);
722
	if (!disk)
T
Tejun Heo 已提交
723
		return -EINVAL;
724 725 726 727
	if (part) {
		put_disk(disk);
		return -EINVAL;
	}
728 729

	rcu_read_lock();
T
Tejun Heo 已提交
730
	spin_lock_irq(disk->queue->queue_lock);
731

732
	if (blkcg_policy_enabled(disk->queue, pol))
733
		blkg = blkg_lookup_create(blkcg, disk->queue);
734 735
	else
		blkg = ERR_PTR(-EINVAL);
736

T
Tejun Heo 已提交
737 738
	if (IS_ERR(blkg)) {
		ret = PTR_ERR(blkg);
739
		rcu_read_unlock();
740
		spin_unlock_irq(disk->queue->queue_lock);
741 742 743 744 745 746 747 748 749 750
		put_disk(disk);
		/*
		 * 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();
751
		}
T
Tejun Heo 已提交
752
		return ret;
753
	}
754 755 756

	ctx->disk = disk;
	ctx->blkg = blkg;
T
Tejun Heo 已提交
757 758
	ctx->v = v;
	return 0;
759
}
760
EXPORT_SYMBOL_GPL(blkg_conf_prep);
761

762 763 764 765 766 767 768
/**
 * 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().
 */
769
void blkg_conf_finish(struct blkg_conf_ctx *ctx)
770
	__releases(ctx->disk->queue->queue_lock) __releases(rcu)
771
{
772
	spin_unlock_irq(ctx->disk->queue->queue_lock);
773 774
	rcu_read_unlock();
	put_disk(ctx->disk);
775
}
776
EXPORT_SYMBOL_GPL(blkg_conf_finish);
777

T
Tejun Heo 已提交
778
struct cftype blkcg_files[] = {
779 780
	{
		.name = "reset_stats",
T
Tejun Heo 已提交
781
		.write_u64 = blkcg_reset_stats,
782
	},
783
	{ }	/* terminate */
784 785
};

786
/**
787
 * blkcg_css_offline - cgroup css_offline callback
788
 * @css: css of interest
789
 *
790 791
 * This function is called when @css is about to go away and responsible
 * for shooting down all blkgs associated with @css.  blkgs should be
792 793 794 795 796
 * 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().
 */
797
static void blkcg_css_offline(struct cgroup_subsys_state *css)
798
{
799
	struct blkcg *blkcg = css_to_blkcg(css);
800

801
	spin_lock_irq(&blkcg->lock);
802

803
	while (!hlist_empty(&blkcg->blkg_list)) {
T
Tejun Heo 已提交
804 805
		struct blkcg_gq *blkg = hlist_entry(blkcg->blkg_list.first,
						struct blkcg_gq, blkcg_node);
T
Tejun Heo 已提交
806
		struct request_queue *q = blkg->q;
807

808 809 810 811 812 813
		if (spin_trylock(q->queue_lock)) {
			blkg_destroy(blkg);
			spin_unlock(q->queue_lock);
		} else {
			spin_unlock_irq(&blkcg->lock);
			cpu_relax();
814
			spin_lock_irq(&blkcg->lock);
815
		}
816
	}
817

818
	spin_unlock_irq(&blkcg->lock);
819 820

	wb_blkcg_offline(blkcg);
821 822
}

823
static void blkcg_css_free(struct cgroup_subsys_state *css)
824
{
825
	struct blkcg *blkcg = css_to_blkcg(css);
826
	int i;
827

T
Tejun Heo 已提交
828 829 830 831
	mutex_lock(&blkcg_pol_mutex);
	list_del(&blkcg->all_blkcgs_node);
	mutex_unlock(&blkcg_pol_mutex);

832 833 834
	for (i = 0; i < BLKCG_MAX_POLS; i++)
		kfree(blkcg->pd[i]);
	kfree(blkcg);
835 836
}

837 838
static struct cgroup_subsys_state *
blkcg_css_alloc(struct cgroup_subsys_state *parent_css)
839
{
T
Tejun Heo 已提交
840
	struct blkcg *blkcg;
841 842
	struct cgroup_subsys_state *ret;
	int i;
843

T
Tejun Heo 已提交
844 845
	mutex_lock(&blkcg_pol_mutex);

846
	if (!parent_css) {
T
Tejun Heo 已提交
847
		blkcg = &blkcg_root;
848 849 850 851 852 853
	} else {
		blkcg = kzalloc(sizeof(*blkcg), GFP_KERNEL);
		if (!blkcg) {
			ret = ERR_PTR(-ENOMEM);
			goto free_blkcg;
		}
854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878
	}

	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.
		 */
		if (!pol || !pol->cpd_size)
			continue;

		BUG_ON(blkcg->pd[i]);
		cpd = kzalloc(pol->cpd_size, GFP_KERNEL);
		if (!cpd) {
			ret = ERR_PTR(-ENOMEM);
			goto free_pd_blkcg;
		}
		blkcg->pd[i] = cpd;
		cpd->plid = i;
		pol->cpd_init_fn(blkcg);
	}
879 880

	spin_lock_init(&blkcg->lock);
881
	INIT_RADIX_TREE(&blkcg->blkg_tree, GFP_NOWAIT);
882
	INIT_HLIST_HEAD(&blkcg->blkg_list);
883 884 885
#ifdef CONFIG_CGROUP_WRITEBACK
	INIT_LIST_HEAD(&blkcg->cgwb_list);
#endif
T
Tejun Heo 已提交
886 887 888
	list_add_tail(&blkcg->all_blkcgs_node, &all_blkcgs);

	mutex_unlock(&blkcg_pol_mutex);
889
	return &blkcg->css;
890 891 892 893 894 895

free_pd_blkcg:
	for (i--; i >= 0; i--)
		kfree(blkcg->pd[i]);
free_blkcg:
	kfree(blkcg);
T
Tejun Heo 已提交
896
	mutex_unlock(&blkcg_pol_mutex);
897
	return ret;
898 899
}

900 901 902 903 904 905 906 907 908 909 910 911
/**
 * 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)
{
912 913 914 915 916 917 918 919 920
	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);
921

922 923 924 925 926 927 928 929 930 931 932 933 934 935 936
	/*
	 * 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)) {
937
		blkg_free(new_blkg);
938 939 940 941 942
		return PTR_ERR(blkg);
	}

	q->root_blkg = blkg;
	q->root_rl.blkg = blkg;
943

944 945 946 947 948 949 950
	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;
951 952 953 954 955 956 957 958 959 960 961 962
}

/**
 * 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);

963 964 965 966 967 968 969
	/*
	 * @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;

970 971 972 973 974 975 976 977 978 979 980
	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)
{
981
	spin_lock_irq(q->queue_lock);
982
	blkg_destroy_all(q);
983 984
	spin_unlock_irq(q->queue_lock);

985 986 987
	blk_throtl_exit(q);
}

988 989 990 991 992 993
/*
 * 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.
 */
994 995
static int blkcg_can_attach(struct cgroup_subsys_state *css,
			    struct cgroup_taskset *tset)
996
{
997
	struct task_struct *task;
998 999 1000 1001
	struct io_context *ioc;
	int ret = 0;

	/* task_lock() is needed to avoid races with exit_io_context() */
1002
	cgroup_taskset_for_each(task, tset) {
1003 1004 1005 1006 1007 1008 1009 1010
		task_lock(task);
		ioc = task->io_context;
		if (ioc && atomic_read(&ioc->nr_tasks) > 1)
			ret = -EINVAL;
		task_unlock(task);
		if (ret)
			break;
	}
1011 1012 1013
	return ret;
}

1014
struct cgroup_subsys blkio_cgrp_subsys = {
1015 1016 1017
	.css_alloc = blkcg_css_alloc,
	.css_offline = blkcg_css_offline,
	.css_free = blkcg_css_free,
T
Tejun Heo 已提交
1018
	.can_attach = blkcg_can_attach,
1019
	.legacy_cftypes = blkcg_files,
1020 1021 1022 1023 1024 1025 1026 1027
#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
1028
};
1029
EXPORT_SYMBOL_GPL(blkio_cgrp_subsys);
1030

1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047
/**
 * 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
1048
			  const struct blkcg_policy *pol)
1049
{
1050
	struct blkg_policy_data *pd_prealloc = NULL;
1051
	struct blkcg_gq *blkg;
1052
	int ret;
1053 1054 1055 1056 1057

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

	blk_queue_bypass_start(q);
1058 1059 1060 1061
pd_prealloc:
	if (!pd_prealloc) {
		pd_prealloc = kzalloc_node(pol->pd_size, GFP_KERNEL, q->node);
		if (!pd_prealloc) {
1062
			ret = -ENOMEM;
1063
			goto out_bypass_end;
1064 1065 1066 1067 1068 1069
		}
	}

	spin_lock_irq(q->queue_lock);

	list_for_each_entry(blkg, &q->blkg_list, q_node) {
1070 1071 1072 1073
		struct blkg_policy_data *pd;

		if (blkg->pd[pol->plid])
			continue;
1074

1075 1076 1077 1078 1079 1080 1081
		pd = kzalloc_node(pol->pd_size, GFP_NOWAIT, q->node);
		if (!pd)
			swap(pd, pd_prealloc);
		if (!pd) {
			spin_unlock_irq(q->queue_lock);
			goto pd_prealloc;
		}
1082 1083 1084

		blkg->pd[pol->plid] = pd;
		pd->blkg = blkg;
T
Tejun Heo 已提交
1085
		pd->plid = pol->plid;
1086 1087
		if (pol->pd_init_fn)
			pol->pd_init_fn(blkg);
1088 1089 1090 1091
	}

	__set_bit(pol->plid, q->blkcg_pols);
	ret = 0;
1092

1093
	spin_unlock_irq(q->queue_lock);
1094
out_bypass_end:
1095
	blk_queue_bypass_end(q);
1096
	kfree(pd_prealloc);
1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109
	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
1110
			     const struct blkcg_policy *pol)
1111
{
T
Tejun Heo 已提交
1112
	struct blkcg_gq *blkg;
1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125

	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);

1126 1127
		if (pol->pd_offline_fn)
			pol->pd_offline_fn(blkg);
1128 1129
		if (pol->pd_exit_fn)
			pol->pd_exit_fn(blkg);
1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141

		kfree(blkg->pd[pol->plid]);
		blkg->pd[pol->plid] = NULL;

		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
1142
/**
T
Tejun Heo 已提交
1143 1144
 * blkcg_policy_register - register a blkcg policy
 * @pol: blkcg policy to register
T
Tejun Heo 已提交
1145
 *
T
Tejun Heo 已提交
1146 1147
 * 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
1148
 */
1149
int blkcg_policy_register(struct blkcg_policy *pol)
1150
{
1151
	struct blkcg *blkcg;
T
Tejun Heo 已提交
1152
	int i, ret;
1153

1154 1155 1156
	if (WARN_ON(pol->pd_size < sizeof(struct blkg_policy_data)))
		return -EINVAL;

1157
	mutex_lock(&blkcg_pol_register_mutex);
1158 1159
	mutex_lock(&blkcg_pol_mutex);

T
Tejun Heo 已提交
1160 1161 1162
	/* find an empty slot */
	ret = -ENOSPC;
	for (i = 0; i < BLKCG_MAX_POLS; i++)
T
Tejun Heo 已提交
1163
		if (!blkcg_policy[i])
T
Tejun Heo 已提交
1164 1165
			break;
	if (i >= BLKCG_MAX_POLS)
1166
		goto err_unlock;
1167

1168
	/* register @pol */
T
Tejun Heo 已提交
1169
	pol->plid = i;
1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188
	blkcg_policy[pol->plid] = pol;

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

			cpd = kzalloc(pol->cpd_size, GFP_KERNEL);
			if (!cpd) {
				mutex_unlock(&blkcg_pol_mutex);
				goto err_free_cpds;
			}

			blkcg->pd[pol->plid] = cpd;
			cpd->plid = pol->plid;
			pol->cpd_init_fn(blkcg);
		}
	}

1189
	mutex_unlock(&blkcg_pol_mutex);
T
Tejun Heo 已提交
1190 1191

	/* everything is in place, add intf files for the new policy */
T
Tejun Heo 已提交
1192
	if (pol->cftypes)
1193 1194
		WARN_ON(cgroup_add_legacy_cftypes(&blkio_cgrp_subsys,
						  pol->cftypes));
1195 1196 1197
	mutex_unlock(&blkcg_pol_register_mutex);
	return 0;

1198 1199 1200 1201 1202 1203 1204 1205
err_free_cpds:
	if (pol->cpd_size) {
		list_for_each_entry(blkcg, &all_blkcgs, all_blkcgs_node) {
			kfree(blkcg->pd[pol->plid]);
			blkcg->pd[pol->plid] = NULL;
		}
	}
	blkcg_policy[pol->plid] = NULL;
1206
err_unlock:
1207
	mutex_unlock(&blkcg_pol_mutex);
1208
	mutex_unlock(&blkcg_pol_register_mutex);
T
Tejun Heo 已提交
1209
	return ret;
1210
}
T
Tejun Heo 已提交
1211
EXPORT_SYMBOL_GPL(blkcg_policy_register);
1212

T
Tejun Heo 已提交
1213
/**
T
Tejun Heo 已提交
1214 1215
 * blkcg_policy_unregister - unregister a blkcg policy
 * @pol: blkcg policy to unregister
T
Tejun Heo 已提交
1216
 *
T
Tejun Heo 已提交
1217
 * Undo blkcg_policy_register(@pol).  Might sleep.
T
Tejun Heo 已提交
1218
 */
T
Tejun Heo 已提交
1219
void blkcg_policy_unregister(struct blkcg_policy *pol)
1220
{
1221 1222
	struct blkcg *blkcg;

1223
	mutex_lock(&blkcg_pol_register_mutex);
1224

T
Tejun Heo 已提交
1225
	if (WARN_ON(blkcg_policy[pol->plid] != pol))
T
Tejun Heo 已提交
1226 1227 1228
		goto out_unlock;

	/* kill the intf files first */
T
Tejun Heo 已提交
1229
	if (pol->cftypes)
1230
		cgroup_rm_cftypes(pol->cftypes);
1231

1232
	/* remove cpds and unregister */
1233
	mutex_lock(&blkcg_pol_mutex);
1234 1235 1236 1237 1238 1239 1240

	if (pol->cpd_size) {
		list_for_each_entry(blkcg, &all_blkcgs, all_blkcgs_node) {
			kfree(blkcg->pd[pol->plid]);
			blkcg->pd[pol->plid] = NULL;
		}
	}
T
Tejun Heo 已提交
1241
	blkcg_policy[pol->plid] = NULL;
1242

1243
	mutex_unlock(&blkcg_pol_mutex);
1244 1245
out_unlock:
	mutex_unlock(&blkcg_pol_register_mutex);
1246
}
T
Tejun Heo 已提交
1247
EXPORT_SYMBOL_GPL(blkcg_policy_unregister);