delayed-ref.c 25.8 KB
Newer Older
1
// SPDX-License-Identifier: GPL-2.0
2 3 4 5 6
/*
 * Copyright (C) 2009 Oracle.  All rights reserved.
 */

#include <linux/sched.h>
7
#include <linux/slab.h>
8 9 10 11
#include <linux/sort.h>
#include "ctree.h"
#include "delayed-ref.h"
#include "transaction.h"
12
#include "qgroup.h"
13

14 15 16 17
struct kmem_cache *btrfs_delayed_ref_head_cachep;
struct kmem_cache *btrfs_delayed_tree_ref_cachep;
struct kmem_cache *btrfs_delayed_data_ref_cachep;
struct kmem_cache *btrfs_delayed_extent_op_cachep;
18 19 20 21 22 23 24 25 26 27
/*
 * delayed back reference update tracking.  For subvolume trees
 * we queue up extent allocations and backref maintenance for
 * delayed processing.   This avoids deep call chains where we
 * add extents in the middle of btrfs_search_slot, and it allows
 * us to buffer up frequently modified backrefs in an rb tree instead
 * of hammering updates on the extent allocation tree.
 */

/*
28 29
 * compare two delayed tree backrefs with same bytenr and type
 */
J
Josef Bacik 已提交
30 31
static int comp_tree_refs(struct btrfs_delayed_tree_ref *ref1,
			  struct btrfs_delayed_tree_ref *ref2)
32
{
33
	if (ref1->node.type == BTRFS_TREE_BLOCK_REF_KEY) {
34 35 36 37 38 39 40 41 42 43
		if (ref1->root < ref2->root)
			return -1;
		if (ref1->root > ref2->root)
			return 1;
	} else {
		if (ref1->parent < ref2->parent)
			return -1;
		if (ref1->parent > ref2->parent)
			return 1;
	}
44 45 46 47 48
	return 0;
}

/*
 * compare two delayed data backrefs with same bytenr and type
49
 */
J
Josef Bacik 已提交
50 51
static int comp_data_refs(struct btrfs_delayed_data_ref *ref1,
			  struct btrfs_delayed_data_ref *ref2)
52
{
53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74
	if (ref1->node.type == BTRFS_EXTENT_DATA_REF_KEY) {
		if (ref1->root < ref2->root)
			return -1;
		if (ref1->root > ref2->root)
			return 1;
		if (ref1->objectid < ref2->objectid)
			return -1;
		if (ref1->objectid > ref2->objectid)
			return 1;
		if (ref1->offset < ref2->offset)
			return -1;
		if (ref1->offset > ref2->offset)
			return 1;
	} else {
		if (ref1->parent < ref2->parent)
			return -1;
		if (ref1->parent > ref2->parent)
			return 1;
	}
	return 0;
}

J
Josef Bacik 已提交
75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102
static int comp_refs(struct btrfs_delayed_ref_node *ref1,
		     struct btrfs_delayed_ref_node *ref2,
		     bool check_seq)
{
	int ret = 0;

	if (ref1->type < ref2->type)
		return -1;
	if (ref1->type > ref2->type)
		return 1;
	if (ref1->type == BTRFS_TREE_BLOCK_REF_KEY ||
	    ref1->type == BTRFS_SHARED_BLOCK_REF_KEY)
		ret = comp_tree_refs(btrfs_delayed_node_to_tree_ref(ref1),
				     btrfs_delayed_node_to_tree_ref(ref2));
	else
		ret = comp_data_refs(btrfs_delayed_node_to_data_ref(ref1),
				     btrfs_delayed_node_to_data_ref(ref2));
	if (ret)
		return ret;
	if (check_seq) {
		if (ref1->seq < ref2->seq)
			return -1;
		if (ref1->seq > ref2->seq)
			return 1;
	}
	return 0;
}

L
Liu Bo 已提交
103 104 105 106 107 108 109 110 111 112 113
/* insert a new ref to head ref rbtree */
static struct btrfs_delayed_ref_head *htree_insert(struct rb_root *root,
						   struct rb_node *node)
{
	struct rb_node **p = &root->rb_node;
	struct rb_node *parent_node = NULL;
	struct btrfs_delayed_ref_head *entry;
	struct btrfs_delayed_ref_head *ins;
	u64 bytenr;

	ins = rb_entry(node, struct btrfs_delayed_ref_head, href_node);
114
	bytenr = ins->bytenr;
L
Liu Bo 已提交
115 116 117 118 119
	while (*p) {
		parent_node = *p;
		entry = rb_entry(parent_node, struct btrfs_delayed_ref_head,
				 href_node);

120
		if (bytenr < entry->bytenr)
L
Liu Bo 已提交
121
			p = &(*p)->rb_left;
122
		else if (bytenr > entry->bytenr)
L
Liu Bo 已提交
123 124 125 126 127 128 129 130 131 132
			p = &(*p)->rb_right;
		else
			return entry;
	}

	rb_link_node(node, parent_node, p);
	rb_insert_color(node, root);
	return NULL;
}

133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160
static struct btrfs_delayed_ref_node* tree_insert(struct rb_root *root,
		struct btrfs_delayed_ref_node *ins)
{
	struct rb_node **p = &root->rb_node;
	struct rb_node *node = &ins->ref_node;
	struct rb_node *parent_node = NULL;
	struct btrfs_delayed_ref_node *entry;

	while (*p) {
		int comp;

		parent_node = *p;
		entry = rb_entry(parent_node, struct btrfs_delayed_ref_node,
				 ref_node);
		comp = comp_refs(ins, entry, true);
		if (comp < 0)
			p = &(*p)->rb_left;
		else if (comp > 0)
			p = &(*p)->rb_right;
		else
			return entry;
	}

	rb_link_node(node, parent_node, p);
	rb_insert_color(node, root);
	return NULL;
}

161
/*
162
 * find an head entry based on bytenr. This returns the delayed ref
163 164 165
 * head if it was able to find one, or NULL if nothing was in that spot.
 * If return_bigger is given, the next bigger entry is returned if no exact
 * match is found.
166
 */
L
Liu Bo 已提交
167 168
static struct btrfs_delayed_ref_head *
find_ref_head(struct rb_root *root, u64 bytenr,
169
	      int return_bigger)
170
{
171
	struct rb_node *n;
L
Liu Bo 已提交
172
	struct btrfs_delayed_ref_head *entry;
173

174 175
	n = root->rb_node;
	entry = NULL;
176
	while (n) {
L
Liu Bo 已提交
177
		entry = rb_entry(n, struct btrfs_delayed_ref_head, href_node);
178

179
		if (bytenr < entry->bytenr)
180
			n = n->rb_left;
181
		else if (bytenr > entry->bytenr)
182 183 184 185
			n = n->rb_right;
		else
			return entry;
	}
186
	if (entry && return_bigger) {
187
		if (bytenr > entry->bytenr) {
L
Liu Bo 已提交
188
			n = rb_next(&entry->href_node);
189 190
			if (!n)
				n = rb_first(root);
L
Liu Bo 已提交
191 192
			entry = rb_entry(n, struct btrfs_delayed_ref_head,
					 href_node);
193
			return entry;
194 195 196
		}
		return entry;
	}
197 198 199
	return NULL;
}

200 201
int btrfs_delayed_ref_lock(struct btrfs_trans_handle *trans,
			   struct btrfs_delayed_ref_head *head)
202
{
203 204 205
	struct btrfs_delayed_ref_root *delayed_refs;

	delayed_refs = &trans->transaction->delayed_refs;
206
	lockdep_assert_held(&delayed_refs->lock);
207 208 209
	if (mutex_trylock(&head->mutex))
		return 0;

210
	refcount_inc(&head->refs);
211 212 213 214
	spin_unlock(&delayed_refs->lock);

	mutex_lock(&head->mutex);
	spin_lock(&delayed_refs->lock);
215
	if (RB_EMPTY_NODE(&head->href_node)) {
216
		mutex_unlock(&head->mutex);
217
		btrfs_put_delayed_ref_head(head);
218 219
		return -EAGAIN;
	}
220
	btrfs_put_delayed_ref_head(head);
221 222 223
	return 0;
}

224
static inline void drop_delayed_ref(struct btrfs_trans_handle *trans,
225
				    struct btrfs_delayed_ref_root *delayed_refs,
226
				    struct btrfs_delayed_ref_head *head,
227 228
				    struct btrfs_delayed_ref_node *ref)
{
229
	lockdep_assert_held(&head->lock);
230 231
	rb_erase(&ref->ref_node, &head->ref_tree);
	RB_CLEAR_NODE(&ref->ref_node);
232 233
	if (!list_empty(&ref->add_list))
		list_del(&ref->add_list);
234 235
	ref->in_tree = 0;
	btrfs_put_delayed_ref(ref);
236
	atomic_dec(&delayed_refs->num_entries);
237 238 239 240
	if (trans->delayed_ref_updates)
		trans->delayed_ref_updates--;
}

241 242 243 244 245 246 247
static bool merge_ref(struct btrfs_trans_handle *trans,
		      struct btrfs_delayed_ref_root *delayed_refs,
		      struct btrfs_delayed_ref_head *head,
		      struct btrfs_delayed_ref_node *ref,
		      u64 seq)
{
	struct btrfs_delayed_ref_node *next;
248
	struct rb_node *node = rb_next(&ref->ref_node);
249 250
	bool done = false;

251
	while (!done && node) {
252 253
		int mod;

254 255
		next = rb_entry(node, struct btrfs_delayed_ref_node, ref_node);
		node = rb_next(node);
256
		if (seq && next->seq >= seq)
257
			break;
J
Josef Bacik 已提交
258
		if (comp_refs(ref, next, false))
259
			break;
260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291

		if (ref->action == next->action) {
			mod = next->ref_mod;
		} else {
			if (ref->ref_mod < next->ref_mod) {
				swap(ref, next);
				done = true;
			}
			mod = -next->ref_mod;
		}

		drop_delayed_ref(trans, delayed_refs, head, next);
		ref->ref_mod += mod;
		if (ref->ref_mod == 0) {
			drop_delayed_ref(trans, delayed_refs, head, ref);
			done = true;
		} else {
			/*
			 * Can't have multiples of the same ref on a tree block.
			 */
			WARN_ON(ref->type == BTRFS_TREE_BLOCK_REF_KEY ||
				ref->type == BTRFS_SHARED_BLOCK_REF_KEY);
		}
	}

	return done;
}

void btrfs_merge_delayed_refs(struct btrfs_trans_handle *trans,
			      struct btrfs_delayed_ref_root *delayed_refs,
			      struct btrfs_delayed_ref_head *head)
{
292
	struct btrfs_fs_info *fs_info = trans->fs_info;
293
	struct btrfs_delayed_ref_node *ref;
294
	struct rb_node *node;
295 296
	u64 seq = 0;

297
	lockdep_assert_held(&head->lock);
298

299
	if (RB_EMPTY_ROOT(&head->ref_tree))
300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315
		return;

	/* We don't have too many refs to merge for data. */
	if (head->is_data)
		return;

	spin_lock(&fs_info->tree_mod_seq_lock);
	if (!list_empty(&fs_info->tree_mod_seq_list)) {
		struct seq_list *elem;

		elem = list_first_entry(&fs_info->tree_mod_seq_list,
					struct seq_list, list);
		seq = elem->seq;
	}
	spin_unlock(&fs_info->tree_mod_seq_lock);

316 317 318
again:
	for (node = rb_first(&head->ref_tree); node; node = rb_next(node)) {
		ref = rb_entry(node, struct btrfs_delayed_ref_node, ref_node);
319 320
		if (seq && ref->seq >= seq)
			continue;
321 322
		if (merge_ref(trans, delayed_refs, head, ref, seq))
			goto again;
323 324 325
	}
}

326
int btrfs_check_delayed_seq(struct btrfs_fs_info *fs_info, u64 seq)
327 328
{
	struct seq_list *elem;
329 330 331 332 333 334 335
	int ret = 0;

	spin_lock(&fs_info->tree_mod_seq_lock);
	if (!list_empty(&fs_info->tree_mod_seq_list)) {
		elem = list_first_entry(&fs_info->tree_mod_seq_list,
					struct seq_list, list);
		if (seq >= elem->seq) {
336
			btrfs_debug(fs_info,
337
				"holding back delayed_ref %#x.%x, lowest is %#x.%x",
338
				(u32)(seq >> 32), (u32)seq,
339
				(u32)(elem->seq >> 32), (u32)elem->seq);
340 341
			ret = 1;
		}
342
	}
343 344 345

	spin_unlock(&fs_info->tree_mod_seq_lock);
	return ret;
346 347
}

348 349
struct btrfs_delayed_ref_head *
btrfs_select_ref_head(struct btrfs_trans_handle *trans)
350 351
{
	struct btrfs_delayed_ref_root *delayed_refs;
352 353 354
	struct btrfs_delayed_ref_head *head;
	u64 start;
	bool loop = false;
355

356
	delayed_refs = &trans->transaction->delayed_refs;
L
Liu Bo 已提交
357

358
again:
359
	start = delayed_refs->run_delayed_start;
360
	head = find_ref_head(&delayed_refs->href_root, start, 1);
361 362
	if (!head && !loop) {
		delayed_refs->run_delayed_start = 0;
363
		start = 0;
364
		loop = true;
365
		head = find_ref_head(&delayed_refs->href_root, start, 1);
366 367 368 369
		if (!head)
			return NULL;
	} else if (!head && loop) {
		return NULL;
370
	}
371

372 373 374 375 376 377 378 379 380 381 382 383 384 385 386
	while (head->processing) {
		struct rb_node *node;

		node = rb_next(&head->href_node);
		if (!node) {
			if (loop)
				return NULL;
			delayed_refs->run_delayed_start = 0;
			start = 0;
			loop = true;
			goto again;
		}
		head = rb_entry(node, struct btrfs_delayed_ref_head,
				href_node);
	}
387

388 389 390
	head->processing = 1;
	WARN_ON(delayed_refs->num_heads_ready == 0);
	delayed_refs->num_heads_ready--;
391 392
	delayed_refs->run_delayed_start = head->bytenr +
		head->num_bytes;
393
	return head;
394 395
}

396 397 398 399 400 401
/*
 * Helper to insert the ref_node to the tail or merge with tail.
 *
 * Return 0 for insert.
 * Return >0 for merge.
 */
402 403 404 405
static int insert_delayed_ref(struct btrfs_trans_handle *trans,
			      struct btrfs_delayed_ref_root *root,
			      struct btrfs_delayed_ref_head *href,
			      struct btrfs_delayed_ref_node *ref)
406 407 408 409 410 411
{
	struct btrfs_delayed_ref_node *exist;
	int mod;
	int ret = 0;

	spin_lock(&href->lock);
412 413 414
	exist = tree_insert(&href->ref_tree, ref);
	if (!exist)
		goto inserted;
415 416 417 418 419 420 421 422 423 424 425

	/* Now we are sure we can merge */
	ret = 1;
	if (exist->action == ref->action) {
		mod = ref->ref_mod;
	} else {
		/* Need to change action */
		if (exist->ref_mod < ref->ref_mod) {
			exist->action = ref->action;
			mod = -exist->ref_mod;
			exist->ref_mod = ref->ref_mod;
426 427 428 429 430 431 432 433 434
			if (ref->action == BTRFS_ADD_DELAYED_REF)
				list_add_tail(&exist->add_list,
					      &href->ref_add_list);
			else if (ref->action == BTRFS_DROP_DELAYED_REF) {
				ASSERT(!list_empty(&exist->add_list));
				list_del(&exist->add_list);
			} else {
				ASSERT(0);
			}
435 436 437 438 439 440 441 442 443 444
		} else
			mod = -ref->ref_mod;
	}
	exist->ref_mod += mod;

	/* remove existing tail if its ref_mod is zero */
	if (exist->ref_mod == 0)
		drop_delayed_ref(trans, root, href, exist);
	spin_unlock(&href->lock);
	return ret;
445
inserted:
446 447
	if (ref->action == BTRFS_ADD_DELAYED_REF)
		list_add_tail(&ref->add_list, &href->ref_add_list);
448 449 450 451 452 453
	atomic_inc(&root->num_entries);
	trans->delayed_ref_updates++;
	spin_unlock(&href->lock);
	return ret;
}

454 455 456 457 458
/*
 * helper function to update the accounting in the head ref
 * existing and update must have the same bytenr
 */
static noinline void
459
update_existing_head_ref(struct btrfs_delayed_ref_root *delayed_refs,
460 461
			 struct btrfs_delayed_ref_head *existing,
			 struct btrfs_delayed_ref_head *update,
462
			 int *old_ref_mod_ret)
463
{
464
	int old_ref_mod;
465

466
	BUG_ON(existing->is_data != update->is_data);
467

468 469
	spin_lock(&existing->lock);
	if (update->must_insert_reserved) {
470 471 472 473 474 475 476
		/* if the extent was freed and then
		 * reallocated before the delayed ref
		 * entries were processed, we can end up
		 * with an existing head ref without
		 * the must_insert_reserved flag set.
		 * Set it again here
		 */
477
		existing->must_insert_reserved = update->must_insert_reserved;
478 479 480 481 482 483 484 485 486

		/*
		 * update the num_bytes so we make sure the accounting
		 * is done correctly
		 */
		existing->num_bytes = update->num_bytes;

	}

487 488 489
	if (update->extent_op) {
		if (!existing->extent_op) {
			existing->extent_op = update->extent_op;
490
		} else {
491 492 493 494 495
			if (update->extent_op->update_key) {
				memcpy(&existing->extent_op->key,
				       &update->extent_op->key,
				       sizeof(update->extent_op->key));
				existing->extent_op->update_key = true;
496
			}
497 498 499 500
			if (update->extent_op->update_flags) {
				existing->extent_op->flags_to_set |=
					update->extent_op->flags_to_set;
				existing->extent_op->update_flags = true;
501
			}
502
			btrfs_free_delayed_extent_op(update->extent_op);
503 504
		}
	}
505
	/*
506 507 508
	 * update the reference mod on the head to reflect this new operation,
	 * only need the lock for this case cause we could be processing it
	 * currently, for refs we just added we know we're a-ok.
509
	 */
510
	old_ref_mod = existing->total_ref_mod;
511 512
	if (old_ref_mod_ret)
		*old_ref_mod_ret = old_ref_mod;
513
	existing->ref_mod += update->ref_mod;
514
	existing->total_ref_mod += update->ref_mod;
515 516 517 518 519

	/*
	 * If we are going to from a positive ref mod to a negative or vice
	 * versa we need to make sure to adjust pending_csums accordingly.
	 */
520 521
	if (existing->is_data) {
		if (existing->total_ref_mod >= 0 && old_ref_mod < 0)
522
			delayed_refs->pending_csums -= existing->num_bytes;
523
		if (existing->total_ref_mod < 0 && old_ref_mod >= 0)
524 525
			delayed_refs->pending_csums += existing->num_bytes;
	}
526
	spin_unlock(&existing->lock);
527 528 529
}

/*
530
 * helper function to actually insert a head node into the rbtree.
531
 * this does all the dirty work in terms of maintaining the correct
532
 * overall modification count.
533
 */
534
static noinline struct btrfs_delayed_ref_head *
535
add_delayed_ref_head(struct btrfs_trans_handle *trans,
536
		     struct btrfs_delayed_ref_head *head_ref,
537
		     struct btrfs_qgroup_extent_record *qrecord,
538
		     u64 bytenr, u64 num_bytes, u64 ref_root, u64 reserved,
539 540
		     int action, int is_data, int is_system,
		     int *qrecord_inserted_ret,
541
		     int *old_ref_mod, int *new_ref_mod)
542

543
{
544
	struct btrfs_delayed_ref_head *existing;
545 546 547
	struct btrfs_delayed_ref_root *delayed_refs;
	int count_mod = 1;
	int must_insert_reserved = 0;
548
	int qrecord_inserted = 0;
549

550 551 552
	/* If reserved is provided, it must be a data extent. */
	BUG_ON(!is_data && reserved);

553 554 555 556
	/*
	 * the head node stores the sum of all the mods, so dropping a ref
	 * should drop the sum in the head node by one.
	 */
557 558 559 560
	if (action == BTRFS_UPDATE_DELAYED_HEAD)
		count_mod = 0;
	else if (action == BTRFS_DROP_DELAYED_REF)
		count_mod = -1;
561 562 563 564 565 566 567 568 569 570 571 572

	/*
	 * BTRFS_ADD_DELAYED_EXTENT means that we need to update
	 * the reserved accounting when the extent is finally added, or
	 * if a later modification deletes the delayed ref without ever
	 * inserting the extent into the extent allocation tree.
	 * ref->must_insert_reserved is the flag used to record
	 * that accounting mods are required.
	 *
	 * Once we record must_insert_reserved, switch the action to
	 * BTRFS_ADD_DELAYED_REF because other special casing is not required.
	 */
573
	if (action == BTRFS_ADD_DELAYED_EXTENT)
574
		must_insert_reserved = 1;
575
	else
576 577 578 579
		must_insert_reserved = 0;

	delayed_refs = &trans->transaction->delayed_refs;

580 581 582 583
	refcount_set(&head_ref->refs, 1);
	head_ref->bytenr = bytenr;
	head_ref->num_bytes = num_bytes;
	head_ref->ref_mod = count_mod;
584 585
	head_ref->must_insert_reserved = must_insert_reserved;
	head_ref->is_data = is_data;
586
	head_ref->is_system = is_system;
587
	head_ref->ref_tree = RB_ROOT;
588
	INIT_LIST_HEAD(&head_ref->ref_add_list);
589
	RB_CLEAR_NODE(&head_ref->href_node);
590
	head_ref->processing = 0;
591
	head_ref->total_ref_mod = count_mod;
592 593
	head_ref->qgroup_reserved = 0;
	head_ref->qgroup_ref_root = 0;
594 595
	spin_lock_init(&head_ref->lock);
	mutex_init(&head_ref->mutex);
596

597 598
	/* Record qgroup extent info if provided */
	if (qrecord) {
599 600 601 602 603
		if (ref_root && reserved) {
			head_ref->qgroup_ref_root = ref_root;
			head_ref->qgroup_reserved = reserved;
		}

604 605 606 607
		qrecord->bytenr = bytenr;
		qrecord->num_bytes = num_bytes;
		qrecord->old_roots = NULL;

608
		if(btrfs_qgroup_trace_extent_nolock(trans->fs_info,
609
					delayed_refs, qrecord))
610
			kfree(qrecord);
611 612
		else
			qrecord_inserted = 1;
613 614
	}

615
	trace_add_delayed_ref_head(trans->fs_info, head_ref, action);
616

617 618
	existing = htree_insert(&delayed_refs->href_root,
				&head_ref->href_node);
619
	if (existing) {
620 621
		WARN_ON(ref_root && reserved && existing->qgroup_ref_root
			&& existing->qgroup_reserved);
622
		update_existing_head_ref(delayed_refs, existing, head_ref,
623
					 old_ref_mod);
624 625 626 627
		/*
		 * we've updated the existing ref, free the newly
		 * allocated ref
		 */
628
		kmem_cache_free(btrfs_delayed_ref_head_cachep, head_ref);
629
		head_ref = existing;
630
	} else {
631 632
		if (old_ref_mod)
			*old_ref_mod = 0;
633 634
		if (is_data && count_mod < 0)
			delayed_refs->pending_csums += num_bytes;
635 636
		delayed_refs->num_heads++;
		delayed_refs->num_heads_ready++;
637
		atomic_inc(&delayed_refs->num_entries);
638 639
		trans->delayed_ref_updates++;
	}
640 641
	if (qrecord_inserted_ret)
		*qrecord_inserted_ret = qrecord_inserted;
642 643
	if (new_ref_mod)
		*new_ref_mod = head_ref->total_ref_mod;
644
	return head_ref;
645 646
}

647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697
/*
 * init_delayed_ref_common - Initialize the structure which represents a
 *			     modification to a an extent.
 *
 * @fs_info:    Internal to the mounted filesystem mount structure.
 *
 * @ref:	The structure which is going to be initialized.
 *
 * @bytenr:	The logical address of the extent for which a modification is
 *		going to be recorded.
 *
 * @num_bytes:  Size of the extent whose modification is being recorded.
 *
 * @ref_root:	The id of the root where this modification has originated, this
 *		can be either one of the well-known metadata trees or the
 *		subvolume id which references this extent.
 *
 * @action:	Can be one of BTRFS_ADD_DELAYED_REF/BTRFS_DROP_DELAYED_REF or
 *		BTRFS_ADD_DELAYED_EXTENT
 *
 * @ref_type:	Holds the type of the extent which is being recorded, can be
 *		one of BTRFS_SHARED_BLOCK_REF_KEY/BTRFS_TREE_BLOCK_REF_KEY
 *		when recording a metadata extent or BTRFS_SHARED_DATA_REF_KEY/
 *		BTRFS_EXTENT_DATA_REF_KEY when recording data extent
 */
static void init_delayed_ref_common(struct btrfs_fs_info *fs_info,
				    struct btrfs_delayed_ref_node *ref,
				    u64 bytenr, u64 num_bytes, u64 ref_root,
				    int action, u8 ref_type)
{
	u64 seq = 0;

	if (action == BTRFS_ADD_DELAYED_EXTENT)
		action = BTRFS_ADD_DELAYED_REF;

	if (is_fstree(ref_root))
		seq = atomic64_read(&fs_info->tree_mod_seq);

	refcount_set(&ref->refs, 1);
	ref->bytenr = bytenr;
	ref->num_bytes = num_bytes;
	ref->ref_mod = 1;
	ref->action = action;
	ref->is_head = 0;
	ref->in_tree = 1;
	ref->seq = seq;
	ref->type = ref_type;
	RB_CLEAR_NODE(&ref->ref_node);
	INIT_LIST_HEAD(&ref->add_list);
}

698 699 700
/*
 * helper to insert a delayed data ref into the rbtree.
 */
701
static noinline void
702
add_delayed_data_ref(struct btrfs_trans_handle *trans,
703 704 705
		     struct btrfs_delayed_ref_head *head_ref,
		     struct btrfs_delayed_ref_node *ref, u64 bytenr,
		     u64 num_bytes, u64 parent, u64 ref_root, u64 owner,
706
		     u64 offset, int action)
707 708 709
{
	struct btrfs_delayed_data_ref *full_ref;
	struct btrfs_delayed_ref_root *delayed_refs;
710
	u8 ref_type;
711
	int ret;
712 713 714

	delayed_refs = &trans->transaction->delayed_refs;
	full_ref = btrfs_delayed_node_to_data_ref(ref);
715
	if (parent)
716
	        ref_type = BTRFS_SHARED_DATA_REF_KEY;
717
	else
718
	        ref_type = BTRFS_EXTENT_DATA_REF_KEY;
A
Arne Jansen 已提交
719

720 721 722 723
	init_delayed_ref_common(trans->fs_info, ref, bytenr, num_bytes,
				ref_root, action, ref_type);
	full_ref->root = ref_root;
	full_ref->parent = parent;
724 725
	full_ref->objectid = owner;
	full_ref->offset = offset;
726

727 728 729
	trace_add_delayed_data_ref(trans->fs_info, ref, full_ref,
				   action == BTRFS_ADD_DELAYED_EXTENT ?
				   BTRFS_ADD_DELAYED_REF : action);
730

731
	ret = insert_delayed_ref(trans, delayed_refs, head_ref, ref);
732
	if (ret > 0)
733
		kmem_cache_free(btrfs_delayed_data_ref_cachep, full_ref);
734 735 736
}

/*
737
 * add a delayed tree ref.  This does all of the accounting required
738 739 740
 * to make sure the delayed ref is eventually processed before this
 * transaction commits.
 */
A
Arne Jansen 已提交
741 742
int btrfs_add_delayed_tree_ref(struct btrfs_fs_info *fs_info,
			       struct btrfs_trans_handle *trans,
743 744
			       u64 bytenr, u64 num_bytes, u64 parent,
			       u64 ref_root,  int level, int action,
745 746
			       struct btrfs_delayed_extent_op *extent_op,
			       int *old_ref_mod, int *new_ref_mod)
747
{
748
	struct btrfs_delayed_tree_ref *ref;
749 750
	struct btrfs_delayed_ref_head *head_ref;
	struct btrfs_delayed_ref_root *delayed_refs;
751
	struct btrfs_qgroup_extent_record *record = NULL;
752
	int qrecord_inserted;
753
	int is_system = (ref_root == BTRFS_CHUNK_TREE_OBJECTID);
754 755
	int ret;
	u8 ref_type;
756

757
	BUG_ON(extent_op && extent_op->is_data);
758
	ref = kmem_cache_alloc(btrfs_delayed_tree_ref_cachep, GFP_NOFS);
759 760 761
	if (!ref)
		return -ENOMEM;

762 763 764 765 766 767 768 769 770 771
	if (parent)
		ref_type = BTRFS_SHARED_BLOCK_REF_KEY;
	else
		ref_type = BTRFS_TREE_BLOCK_REF_KEY;
	init_delayed_ref_common(fs_info, &ref->node, bytenr, num_bytes,
				ref_root, action, ref_type);
	ref->root = ref_root;
	ref->parent = parent;
	ref->level = level;

772
	head_ref = kmem_cache_alloc(btrfs_delayed_ref_head_cachep, GFP_NOFS);
773 774
	if (!head_ref)
		goto free_ref;
775

776 777
	if (test_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags) &&
	    is_fstree(ref_root)) {
778
		record = kmalloc(sizeof(*record), GFP_NOFS);
779 780
		if (!record)
			goto free_head_ref;
781 782
	}

783 784 785 786 787
	head_ref->extent_op = extent_op;

	delayed_refs = &trans->transaction->delayed_refs;
	spin_lock(&delayed_refs->lock);

788
	/*
789 790
	 * insert both the head node and the new ref without dropping
	 * the spin lock
791
	 */
792 793
	head_ref = add_delayed_ref_head(trans, head_ref, record, bytenr,
					num_bytes, 0, 0, action, 0,
794 795
					is_system, &qrecord_inserted,
					old_ref_mod, new_ref_mod);
796

797 798

	ret = insert_delayed_ref(trans, delayed_refs, head_ref, &ref->node);
799
	spin_unlock(&delayed_refs->lock);
800

801 802 803 804 805 806
	trace_add_delayed_tree_ref(fs_info, &ref->node, ref,
				   action == BTRFS_ADD_DELAYED_EXTENT ?
				   BTRFS_ADD_DELAYED_REF : action);
	if (ret > 0)
		kmem_cache_free(btrfs_delayed_tree_ref_cachep, ref);

807
	if (qrecord_inserted)
808 809
		btrfs_qgroup_trace_extent_post(fs_info, record);

810
	return 0;
811 812 813 814 815 816 817

free_head_ref:
	kmem_cache_free(btrfs_delayed_ref_head_cachep, head_ref);
free_ref:
	kmem_cache_free(btrfs_delayed_tree_ref_cachep, ref);

	return -ENOMEM;
818 819 820 821 822
}

/*
 * add a delayed data ref. it's similar to btrfs_add_delayed_tree_ref.
 */
A
Arne Jansen 已提交
823 824
int btrfs_add_delayed_data_ref(struct btrfs_fs_info *fs_info,
			       struct btrfs_trans_handle *trans,
825 826
			       u64 bytenr, u64 num_bytes,
			       u64 parent, u64 ref_root,
827 828
			       u64 owner, u64 offset, u64 reserved, int action,
			       int *old_ref_mod, int *new_ref_mod)
829 830 831 832
{
	struct btrfs_delayed_data_ref *ref;
	struct btrfs_delayed_ref_head *head_ref;
	struct btrfs_delayed_ref_root *delayed_refs;
833
	struct btrfs_qgroup_extent_record *record = NULL;
834
	int qrecord_inserted;
835

836
	ref = kmem_cache_alloc(btrfs_delayed_data_ref_cachep, GFP_NOFS);
837 838
	if (!ref)
		return -ENOMEM;
839

840
	head_ref = kmem_cache_alloc(btrfs_delayed_ref_head_cachep, GFP_NOFS);
841
	if (!head_ref) {
842
		kmem_cache_free(btrfs_delayed_data_ref_cachep, ref);
843 844
		return -ENOMEM;
	}
845

846 847
	if (test_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags) &&
	    is_fstree(ref_root)) {
848 849 850 851 852 853 854 855 856
		record = kmalloc(sizeof(*record), GFP_NOFS);
		if (!record) {
			kmem_cache_free(btrfs_delayed_data_ref_cachep, ref);
			kmem_cache_free(btrfs_delayed_ref_head_cachep,
					head_ref);
			return -ENOMEM;
		}
	}

857
	head_ref->extent_op = NULL;
858

859 860 861 862 863 864 865
	delayed_refs = &trans->transaction->delayed_refs;
	spin_lock(&delayed_refs->lock);

	/*
	 * insert both the head node and the new ref without dropping
	 * the spin lock
	 */
866 867
	head_ref = add_delayed_ref_head(trans, head_ref, record, bytenr,
					num_bytes, ref_root, reserved,
868
					action, 1, 0, &qrecord_inserted,
869
					old_ref_mod, new_ref_mod);
870

871 872
	add_delayed_data_ref(trans, head_ref, &ref->node, bytenr, num_bytes,
			     parent, ref_root, owner, offset, action);
873
	spin_unlock(&delayed_refs->lock);
874

875 876
	if (qrecord_inserted)
		return btrfs_qgroup_trace_extent_post(fs_info, record);
877 878 879
	return 0;
}

A
Arne Jansen 已提交
880 881
int btrfs_add_delayed_extent_op(struct btrfs_fs_info *fs_info,
				struct btrfs_trans_handle *trans,
882 883 884 885 886 887
				u64 bytenr, u64 num_bytes,
				struct btrfs_delayed_extent_op *extent_op)
{
	struct btrfs_delayed_ref_head *head_ref;
	struct btrfs_delayed_ref_root *delayed_refs;

888
	head_ref = kmem_cache_alloc(btrfs_delayed_ref_head_cachep, GFP_NOFS);
889 890 891 892 893 894 895 896
	if (!head_ref)
		return -ENOMEM;

	head_ref->extent_op = extent_op;

	delayed_refs = &trans->transaction->delayed_refs;
	spin_lock(&delayed_refs->lock);

897 898 899 900 901
	/*
	 * extent_ops just modify the flags of an extent and they don't result
	 * in ref count changes, hence it's safe to pass false/0 for is_system
	 * argument
	 */
902 903 904
	add_delayed_ref_head(trans, head_ref, NULL, bytenr, num_bytes, 0, 0,
			     BTRFS_UPDATE_DELAYED_HEAD, extent_op->is_data,
			     0, NULL, NULL, NULL);
905

906 907 908 909
	spin_unlock(&delayed_refs->lock);
	return 0;
}

910 911 912 913 914 915
/*
 * this does a simple search for the head node for a given extent.
 * It must be called with the delayed ref spinlock held, and it returns
 * the head node if any where found, or NULL if not.
 */
struct btrfs_delayed_ref_head *
916
btrfs_find_delayed_ref_head(struct btrfs_delayed_ref_root *delayed_refs, u64 bytenr)
917
{
918
	return find_ref_head(&delayed_refs->href_root, bytenr, 0);
919
}
920

921
void __cold btrfs_delayed_ref_exit(void)
922
{
923 924 925 926
	kmem_cache_destroy(btrfs_delayed_ref_head_cachep);
	kmem_cache_destroy(btrfs_delayed_tree_ref_cachep);
	kmem_cache_destroy(btrfs_delayed_data_ref_cachep);
	kmem_cache_destroy(btrfs_delayed_extent_op_cachep);
927 928
}

929
int __init btrfs_delayed_ref_init(void)
930 931 932 933
{
	btrfs_delayed_ref_head_cachep = kmem_cache_create(
				"btrfs_delayed_ref_head",
				sizeof(struct btrfs_delayed_ref_head), 0,
934
				SLAB_MEM_SPREAD, NULL);
935 936 937 938 939 940
	if (!btrfs_delayed_ref_head_cachep)
		goto fail;

	btrfs_delayed_tree_ref_cachep = kmem_cache_create(
				"btrfs_delayed_tree_ref",
				sizeof(struct btrfs_delayed_tree_ref), 0,
941
				SLAB_MEM_SPREAD, NULL);
942 943 944 945 946 947
	if (!btrfs_delayed_tree_ref_cachep)
		goto fail;

	btrfs_delayed_data_ref_cachep = kmem_cache_create(
				"btrfs_delayed_data_ref",
				sizeof(struct btrfs_delayed_data_ref), 0,
948
				SLAB_MEM_SPREAD, NULL);
949 950 951 952 953 954
	if (!btrfs_delayed_data_ref_cachep)
		goto fail;

	btrfs_delayed_extent_op_cachep = kmem_cache_create(
				"btrfs_delayed_extent_op",
				sizeof(struct btrfs_delayed_extent_op), 0,
955
				SLAB_MEM_SPREAD, NULL);
956 957 958 959 960 961 962 963
	if (!btrfs_delayed_extent_op_cachep)
		goto fail;

	return 0;
fail:
	btrfs_delayed_ref_exit();
	return -ENOMEM;
}