delayed-ref.c 26.7 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
/*
 * Copyright (C) 2009 Oracle.  All rights reserved.
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public
 * License v2 as published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * General Public License for more details.
 *
 * You should have received a copy of the GNU General Public
 * License along with this program; if not, write to the
 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
 * Boston, MA 021110-1307, USA.
 */

#include <linux/sched.h>
20
#include <linux/slab.h>
21 22 23 24
#include <linux/sort.h>
#include "ctree.h"
#include "delayed-ref.h"
#include "transaction.h"
25
#include "qgroup.h"
26

27 28 29 30
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;
31 32 33 34 35 36 37 38 39 40
/*
 * 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.
 */

/*
41 42 43
 * compare two delayed tree backrefs with same bytenr and type
 */
static int comp_tree_refs(struct btrfs_delayed_tree_ref *ref2,
44
			  struct btrfs_delayed_tree_ref *ref1, int type)
45
{
46 47 48 49 50 51 52 53 54 55 56
	if (type == BTRFS_TREE_BLOCK_REF_KEY) {
		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;
	}
57 58 59 60 61
	return 0;
}

/*
 * compare two delayed data backrefs with same bytenr and type
62
 */
63 64
static int comp_data_refs(struct btrfs_delayed_data_ref *ref2,
			  struct btrfs_delayed_data_ref *ref1)
65
{
66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87
	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;
}

L
Liu Bo 已提交
88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117
/* 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);
	bytenr = ins->node.bytenr;
	while (*p) {
		parent_node = *p;
		entry = rb_entry(parent_node, struct btrfs_delayed_ref_head,
				 href_node);

		if (bytenr < entry->node.bytenr)
			p = &(*p)->rb_left;
		else if (bytenr > entry->node.bytenr)
			p = &(*p)->rb_right;
		else
			return entry;
	}

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

118
/*
119
 * find an head entry based on bytenr. This returns the delayed ref
120 121 122
 * 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.
123
 */
L
Liu Bo 已提交
124 125
static struct btrfs_delayed_ref_head *
find_ref_head(struct rb_root *root, u64 bytenr,
126
	      int return_bigger)
127
{
128
	struct rb_node *n;
L
Liu Bo 已提交
129
	struct btrfs_delayed_ref_head *entry;
130

131 132
	n = root->rb_node;
	entry = NULL;
133
	while (n) {
L
Liu Bo 已提交
134
		entry = rb_entry(n, struct btrfs_delayed_ref_head, href_node);
135

L
Liu Bo 已提交
136
		if (bytenr < entry->node.bytenr)
137
			n = n->rb_left;
138
		else if (bytenr > entry->node.bytenr)
139 140 141 142
			n = n->rb_right;
		else
			return entry;
	}
143
	if (entry && return_bigger) {
144
		if (bytenr > entry->node.bytenr) {
L
Liu Bo 已提交
145
			n = rb_next(&entry->href_node);
146 147
			if (!n)
				n = rb_first(root);
L
Liu Bo 已提交
148 149
			entry = rb_entry(n, struct btrfs_delayed_ref_head,
					 href_node);
150
			return entry;
151 152 153
		}
		return entry;
	}
154 155 156
	return NULL;
}

157 158
int btrfs_delayed_ref_lock(struct btrfs_trans_handle *trans,
			   struct btrfs_delayed_ref_head *head)
159
{
160 161 162 163 164 165 166
	struct btrfs_delayed_ref_root *delayed_refs;

	delayed_refs = &trans->transaction->delayed_refs;
	assert_spin_locked(&delayed_refs->lock);
	if (mutex_trylock(&head->mutex))
		return 0;

167
	refcount_inc(&head->node.refs);
168 169 170 171 172 173 174 175 176 177 178 179 180
	spin_unlock(&delayed_refs->lock);

	mutex_lock(&head->mutex);
	spin_lock(&delayed_refs->lock);
	if (!head->node.in_tree) {
		mutex_unlock(&head->mutex);
		btrfs_put_delayed_ref(&head->node);
		return -EAGAIN;
	}
	btrfs_put_delayed_ref(&head->node);
	return 0;
}

181
static inline void drop_delayed_ref(struct btrfs_trans_handle *trans,
182
				    struct btrfs_delayed_ref_root *delayed_refs,
183
				    struct btrfs_delayed_ref_head *head,
184 185
				    struct btrfs_delayed_ref_node *ref)
{
L
Liu Bo 已提交
186 187 188
	if (btrfs_delayed_ref_is_head(ref)) {
		head = btrfs_delayed_node_to_head(ref);
		rb_erase(&head->href_node, &delayed_refs->href_root);
189 190
	} else {
		assert_spin_locked(&head->lock);
191
		list_del(&ref->list);
192 193
		if (!list_empty(&ref->add_list))
			list_del(&ref->add_list);
L
Liu Bo 已提交
194
	}
195 196
	ref->in_tree = 0;
	btrfs_put_delayed_ref(ref);
197
	atomic_dec(&delayed_refs->num_entries);
198 199 200 201
	if (trans->delayed_ref_updates)
		trans->delayed_ref_updates--;
}

202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224
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;
	bool done = false;

	next = list_first_entry(&head->ref_list, struct btrfs_delayed_ref_node,
				list);
	while (!done && &next->list != &head->ref_list) {
		int mod;
		struct btrfs_delayed_ref_node *next2;

		next2 = list_next_entry(next, list);

		if (next == ref)
			goto next;

		if (seq && next->seq >= seq)
			goto next;

225
		if (next->type != ref->type)
226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314
			goto next;

		if ((ref->type == BTRFS_TREE_BLOCK_REF_KEY ||
		     ref->type == BTRFS_SHARED_BLOCK_REF_KEY) &&
		    comp_tree_refs(btrfs_delayed_node_to_tree_ref(ref),
				   btrfs_delayed_node_to_tree_ref(next),
				   ref->type))
			goto next;
		if ((ref->type == BTRFS_EXTENT_DATA_REF_KEY ||
		     ref->type == BTRFS_SHARED_DATA_REF_KEY) &&
		    comp_data_refs(btrfs_delayed_node_to_data_ref(ref),
				   btrfs_delayed_node_to_data_ref(next)))
			goto next;

		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);
		}
next:
		next = next2;
	}

	return done;
}

void btrfs_merge_delayed_refs(struct btrfs_trans_handle *trans,
			      struct btrfs_fs_info *fs_info,
			      struct btrfs_delayed_ref_root *delayed_refs,
			      struct btrfs_delayed_ref_head *head)
{
	struct btrfs_delayed_ref_node *ref;
	u64 seq = 0;

	assert_spin_locked(&head->lock);

	if (list_empty(&head->ref_list))
		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);

	ref = list_first_entry(&head->ref_list, struct btrfs_delayed_ref_node,
			       list);
	while (&ref->list != &head->ref_list) {
		if (seq && ref->seq >= seq)
			goto next;

		if (merge_ref(trans, delayed_refs, head, ref, seq)) {
			if (list_empty(&head->ref_list))
				break;
			ref = list_first_entry(&head->ref_list,
					       struct btrfs_delayed_ref_node,
					       list);
			continue;
		}
next:
		ref = list_next_entry(ref, list);
	}
}

315 316
int btrfs_check_delayed_seq(struct btrfs_fs_info *fs_info,
			    struct btrfs_delayed_ref_root *delayed_refs,
317 318 319
			    u64 seq)
{
	struct seq_list *elem;
320 321 322 323 324 325 326
	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) {
327 328 329 330 331
			btrfs_debug(fs_info,
				"holding back delayed_ref %#x.%x, lowest is %#x.%x (%p)",
				(u32)(seq >> 32), (u32)seq,
				(u32)(elem->seq >> 32), (u32)elem->seq,
				delayed_refs);
332 333
			ret = 1;
		}
334
	}
335 336 337

	spin_unlock(&fs_info->tree_mod_seq_lock);
	return ret;
338 339
}

340 341
struct btrfs_delayed_ref_head *
btrfs_select_ref_head(struct btrfs_trans_handle *trans)
342 343
{
	struct btrfs_delayed_ref_root *delayed_refs;
344 345 346
	struct btrfs_delayed_ref_head *head;
	u64 start;
	bool loop = false;
347

348
	delayed_refs = &trans->transaction->delayed_refs;
L
Liu Bo 已提交
349

350
again:
351
	start = delayed_refs->run_delayed_start;
352
	head = find_ref_head(&delayed_refs->href_root, start, 1);
353 354
	if (!head && !loop) {
		delayed_refs->run_delayed_start = 0;
355
		start = 0;
356
		loop = true;
357
		head = find_ref_head(&delayed_refs->href_root, start, 1);
358 359 360 361
		if (!head)
			return NULL;
	} else if (!head && loop) {
		return NULL;
362
	}
363

364 365 366 367 368 369 370 371 372 373 374 375 376 377 378
	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);
	}
379

380 381 382 383 384 385
	head->processing = 1;
	WARN_ON(delayed_refs->num_heads_ready == 0);
	delayed_refs->num_heads_ready--;
	delayed_refs->run_delayed_start = head->node.bytenr +
		head->node.num_bytes;
	return head;
386 387
}

388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410
/*
 * Helper to insert the ref_node to the tail or merge with tail.
 *
 * Return 0 for insert.
 * Return >0 for merge.
 */
static int
add_delayed_ref_tail_merge(struct btrfs_trans_handle *trans,
			   struct btrfs_delayed_ref_root *root,
			   struct btrfs_delayed_ref_head *href,
			   struct btrfs_delayed_ref_node *ref)
{
	struct btrfs_delayed_ref_node *exist;
	int mod;
	int ret = 0;

	spin_lock(&href->lock);
	/* Check whether we can merge the tail node with ref */
	if (list_empty(&href->ref_list))
		goto add_tail;
	exist = list_entry(href->ref_list.prev, struct btrfs_delayed_ref_node,
			   list);
	/* No need to compare bytenr nor is_head */
411
	if (exist->type != ref->type || exist->seq != ref->seq)
412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435
		goto add_tail;

	if ((exist->type == BTRFS_TREE_BLOCK_REF_KEY ||
	     exist->type == BTRFS_SHARED_BLOCK_REF_KEY) &&
	    comp_tree_refs(btrfs_delayed_node_to_tree_ref(exist),
			   btrfs_delayed_node_to_tree_ref(ref),
			   ref->type))
		goto add_tail;
	if ((exist->type == BTRFS_EXTENT_DATA_REF_KEY ||
	     exist->type == BTRFS_SHARED_DATA_REF_KEY) &&
	    comp_data_refs(btrfs_delayed_node_to_data_ref(exist),
			   btrfs_delayed_node_to_data_ref(ref)))
		goto add_tail;

	/* 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;
436 437 438 439 440 441 442 443 444
			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);
			}
445 446 447 448 449 450 451 452 453 454 455 456 457
		} 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;

add_tail:
	list_add_tail(&ref->list, &href->ref_list);
458 459
	if (ref->action == BTRFS_ADD_DELAYED_REF)
		list_add_tail(&ref->add_list, &href->ref_add_list);
460 461 462 463 464 465
	atomic_inc(&root->num_entries);
	trans->delayed_ref_updates++;
	spin_unlock(&href->lock);
	return ret;
}

466 467 468 469 470
/*
 * helper function to update the accounting in the head ref
 * existing and update must have the same bytenr
 */
static noinline void
471 472
update_existing_head_ref(struct btrfs_delayed_ref_root *delayed_refs,
			 struct btrfs_delayed_ref_node *existing,
473 474
			 struct btrfs_delayed_ref_node *update,
			 int *old_ref_mod_ret)
475 476 477
{
	struct btrfs_delayed_ref_head *existing_ref;
	struct btrfs_delayed_ref_head *ref;
478
	int old_ref_mod;
479 480 481

	existing_ref = btrfs_delayed_node_to_head(existing);
	ref = btrfs_delayed_node_to_head(update);
482
	BUG_ON(existing_ref->is_data != ref->is_data);
483

484
	spin_lock(&existing_ref->lock);
485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502
	if (ref->must_insert_reserved) {
		/* 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
		 */
		existing_ref->must_insert_reserved = ref->must_insert_reserved;

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

	}

503 504 505 506 507 508 509 510
	if (ref->extent_op) {
		if (!existing_ref->extent_op) {
			existing_ref->extent_op = ref->extent_op;
		} else {
			if (ref->extent_op->update_key) {
				memcpy(&existing_ref->extent_op->key,
				       &ref->extent_op->key,
				       sizeof(ref->extent_op->key));
511
				existing_ref->extent_op->update_key = true;
512 513 514 515
			}
			if (ref->extent_op->update_flags) {
				existing_ref->extent_op->flags_to_set |=
					ref->extent_op->flags_to_set;
516
				existing_ref->extent_op->update_flags = true;
517
			}
518
			btrfs_free_delayed_extent_op(ref->extent_op);
519 520
		}
	}
521
	/*
522 523 524
	 * 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.
525
	 */
526
	old_ref_mod = existing_ref->total_ref_mod;
527 528
	if (old_ref_mod_ret)
		*old_ref_mod_ret = old_ref_mod;
529
	existing->ref_mod += update->ref_mod;
530 531 532 533 534 535 536 537 538 539 540 541
	existing_ref->total_ref_mod += update->ref_mod;

	/*
	 * 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.
	 */
	if (existing_ref->is_data) {
		if (existing_ref->total_ref_mod >= 0 && old_ref_mod < 0)
			delayed_refs->pending_csums -= existing->num_bytes;
		if (existing_ref->total_ref_mod < 0 && old_ref_mod >= 0)
			delayed_refs->pending_csums += existing->num_bytes;
	}
542
	spin_unlock(&existing_ref->lock);
543 544 545
}

/*
546
 * helper function to actually insert a head node into the rbtree.
547
 * this does all the dirty work in terms of maintaining the correct
548
 * overall modification count.
549
 */
550 551 552
static noinline struct btrfs_delayed_ref_head *
add_delayed_ref_head(struct btrfs_fs_info *fs_info,
		     struct btrfs_trans_handle *trans,
553 554
		     struct btrfs_delayed_ref_node *ref,
		     struct btrfs_qgroup_extent_record *qrecord,
555
		     u64 bytenr, u64 num_bytes, u64 ref_root, u64 reserved,
556 557
		     int action, int is_data, int *qrecord_inserted_ret,
		     int *old_ref_mod, int *new_ref_mod)
558
{
559
	struct btrfs_delayed_ref_head *existing;
560
	struct btrfs_delayed_ref_head *head_ref = NULL;
561 562 563
	struct btrfs_delayed_ref_root *delayed_refs;
	int count_mod = 1;
	int must_insert_reserved = 0;
564
	int qrecord_inserted = 0;
565

566 567 568
	/* If reserved is provided, it must be a data extent. */
	BUG_ON(!is_data && reserved);

569 570 571 572
	/*
	 * the head node stores the sum of all the mods, so dropping a ref
	 * should drop the sum in the head node by one.
	 */
573 574 575 576
	if (action == BTRFS_UPDATE_DELAYED_HEAD)
		count_mod = 0;
	else if (action == BTRFS_DROP_DELAYED_REF)
		count_mod = -1;
577 578 579 580 581 582 583 584 585 586 587 588

	/*
	 * 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.
	 */
589
	if (action == BTRFS_ADD_DELAYED_EXTENT)
590
		must_insert_reserved = 1;
591
	else
592 593 594 595 596
		must_insert_reserved = 0;

	delayed_refs = &trans->transaction->delayed_refs;

	/* first set the basic ref node struct up */
597
	refcount_set(&ref->refs, 1);
598
	ref->bytenr = bytenr;
599
	ref->num_bytes = num_bytes;
600
	ref->ref_mod = count_mod;
601 602 603
	ref->type  = 0;
	ref->action  = 0;
	ref->is_head = 1;
604
	ref->in_tree = 1;
605
	ref->seq = 0;
606 607 608 609

	head_ref = btrfs_delayed_node_to_head(ref);
	head_ref->must_insert_reserved = must_insert_reserved;
	head_ref->is_data = is_data;
610
	INIT_LIST_HEAD(&head_ref->ref_list);
611
	INIT_LIST_HEAD(&head_ref->ref_add_list);
612
	head_ref->processing = 0;
613
	head_ref->total_ref_mod = count_mod;
614 615
	head_ref->qgroup_reserved = 0;
	head_ref->qgroup_ref_root = 0;
616

617 618
	/* Record qgroup extent info if provided */
	if (qrecord) {
619 620 621 622 623
		if (ref_root && reserved) {
			head_ref->qgroup_ref_root = ref_root;
			head_ref->qgroup_reserved = reserved;
		}

624 625 626 627
		qrecord->bytenr = bytenr;
		qrecord->num_bytes = num_bytes;
		qrecord->old_roots = NULL;

628
		if(btrfs_qgroup_trace_extent_nolock(fs_info,
629
					delayed_refs, qrecord))
630
			kfree(qrecord);
631 632
		else
			qrecord_inserted = 1;
633 634
	}

635
	spin_lock_init(&head_ref->lock);
636 637
	mutex_init(&head_ref->mutex);

638
	trace_add_delayed_ref_head(fs_info, ref, head_ref, action);
639

640 641
	existing = htree_insert(&delayed_refs->href_root,
				&head_ref->href_node);
642
	if (existing) {
643 644
		WARN_ON(ref_root && reserved && existing->qgroup_ref_root
			&& existing->qgroup_reserved);
645 646
		update_existing_head_ref(delayed_refs, &existing->node, ref,
					 old_ref_mod);
647 648 649 650
		/*
		 * we've updated the existing ref, free the newly
		 * allocated ref
		 */
651
		kmem_cache_free(btrfs_delayed_ref_head_cachep, head_ref);
652
		head_ref = existing;
653
	} else {
654 655
		if (old_ref_mod)
			*old_ref_mod = 0;
656 657
		if (is_data && count_mod < 0)
			delayed_refs->pending_csums += num_bytes;
658 659
		delayed_refs->num_heads++;
		delayed_refs->num_heads_ready++;
660
		atomic_inc(&delayed_refs->num_entries);
661 662
		trans->delayed_ref_updates++;
	}
663 664
	if (qrecord_inserted_ret)
		*qrecord_inserted_ret = qrecord_inserted;
665 666
	if (new_ref_mod)
		*new_ref_mod = head_ref->total_ref_mod;
667
	return head_ref;
668 669 670 671 672
}

/*
 * helper to insert a delayed tree ref into the rbtree.
 */
673 674 675 676 677 678
static noinline void
add_delayed_tree_ref(struct btrfs_fs_info *fs_info,
		     struct btrfs_trans_handle *trans,
		     struct btrfs_delayed_ref_head *head_ref,
		     struct btrfs_delayed_ref_node *ref, u64 bytenr,
		     u64 num_bytes, u64 parent, u64 ref_root, int level,
679
		     int action)
680 681 682
{
	struct btrfs_delayed_tree_ref *full_ref;
	struct btrfs_delayed_ref_root *delayed_refs;
683
	u64 seq = 0;
684
	int ret;
685 686 687 688

	if (action == BTRFS_ADD_DELAYED_EXTENT)
		action = BTRFS_ADD_DELAYED_REF;

J
Josef Bacik 已提交
689 690
	if (is_fstree(ref_root))
		seq = atomic64_read(&fs_info->tree_mod_seq);
691 692 693
	delayed_refs = &trans->transaction->delayed_refs;

	/* first set the basic ref node struct up */
694
	refcount_set(&ref->refs, 1);
695
	ref->bytenr = bytenr;
696
	ref->num_bytes = num_bytes;
697 698 699 700
	ref->ref_mod = 1;
	ref->action = action;
	ref->is_head = 0;
	ref->in_tree = 1;
701
	ref->seq = seq;
702 703
	INIT_LIST_HEAD(&ref->list);
	INIT_LIST_HEAD(&ref->add_list);
704

705
	full_ref = btrfs_delayed_node_to_tree_ref(ref);
706 707 708
	full_ref->parent = parent;
	full_ref->root = ref_root;
	if (parent)
709
		ref->type = BTRFS_SHARED_BLOCK_REF_KEY;
710
	else
711 712
		ref->type = BTRFS_TREE_BLOCK_REF_KEY;
	full_ref->level = level;
713

714
	trace_add_delayed_tree_ref(fs_info, ref, full_ref, action);
715

716 717 718 719 720 721 722
	ret = add_delayed_ref_tail_merge(trans, delayed_refs, head_ref, ref);

	/*
	 * XXX: memory should be freed at the same level allocated.
	 * But bad practice is anywhere... Follow it now. Need cleanup.
	 */
	if (ret > 0)
723
		kmem_cache_free(btrfs_delayed_tree_ref_cachep, full_ref);
724 725 726 727 728
}

/*
 * helper to insert a delayed data ref into the rbtree.
 */
729 730 731 732 733 734
static noinline void
add_delayed_data_ref(struct btrfs_fs_info *fs_info,
		     struct btrfs_trans_handle *trans,
		     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,
735
		     u64 offset, int action)
736 737 738
{
	struct btrfs_delayed_data_ref *full_ref;
	struct btrfs_delayed_ref_root *delayed_refs;
739
	u64 seq = 0;
740
	int ret;
741 742 743 744 745 746

	if (action == BTRFS_ADD_DELAYED_EXTENT)
		action = BTRFS_ADD_DELAYED_REF;

	delayed_refs = &trans->transaction->delayed_refs;

J
Josef Bacik 已提交
747 748 749
	if (is_fstree(ref_root))
		seq = atomic64_read(&fs_info->tree_mod_seq);

750
	/* first set the basic ref node struct up */
751
	refcount_set(&ref->refs, 1);
752 753 754 755 756 757
	ref->bytenr = bytenr;
	ref->num_bytes = num_bytes;
	ref->ref_mod = 1;
	ref->action = action;
	ref->is_head = 0;
	ref->in_tree = 1;
758
	ref->seq = seq;
759 760
	INIT_LIST_HEAD(&ref->list);
	INIT_LIST_HEAD(&ref->add_list);
761

762
	full_ref = btrfs_delayed_node_to_data_ref(ref);
763 764 765
	full_ref->parent = parent;
	full_ref->root = ref_root;
	if (parent)
766
		ref->type = BTRFS_SHARED_DATA_REF_KEY;
767
	else
768
		ref->type = BTRFS_EXTENT_DATA_REF_KEY;
A
Arne Jansen 已提交
769

770 771
	full_ref->objectid = owner;
	full_ref->offset = offset;
772

773
	trace_add_delayed_data_ref(fs_info, ref, full_ref, action);
774

775 776 777
	ret = add_delayed_ref_tail_merge(trans, delayed_refs, head_ref, ref);

	if (ret > 0)
778
		kmem_cache_free(btrfs_delayed_data_ref_cachep, full_ref);
779 780 781
}

/*
782
 * add a delayed tree ref.  This does all of the accounting required
783 784 785
 * to make sure the delayed ref is eventually processed before this
 * transaction commits.
 */
A
Arne Jansen 已提交
786 787
int btrfs_add_delayed_tree_ref(struct btrfs_fs_info *fs_info,
			       struct btrfs_trans_handle *trans,
788 789
			       u64 bytenr, u64 num_bytes, u64 parent,
			       u64 ref_root,  int level, int action,
790 791
			       struct btrfs_delayed_extent_op *extent_op,
			       int *old_ref_mod, int *new_ref_mod)
792
{
793
	struct btrfs_delayed_tree_ref *ref;
794 795
	struct btrfs_delayed_ref_head *head_ref;
	struct btrfs_delayed_ref_root *delayed_refs;
796
	struct btrfs_qgroup_extent_record *record = NULL;
797
	int qrecord_inserted;
798

799
	BUG_ON(extent_op && extent_op->is_data);
800
	ref = kmem_cache_alloc(btrfs_delayed_tree_ref_cachep, GFP_NOFS);
801 802 803
	if (!ref)
		return -ENOMEM;

804
	head_ref = kmem_cache_alloc(btrfs_delayed_ref_head_cachep, GFP_NOFS);
805 806
	if (!head_ref)
		goto free_ref;
807

808 809
	if (test_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags) &&
	    is_fstree(ref_root)) {
810
		record = kmalloc(sizeof(*record), GFP_NOFS);
811 812
		if (!record)
			goto free_head_ref;
813 814
	}

815 816 817 818 819
	head_ref->extent_op = extent_op;

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

820
	/*
821 822
	 * insert both the head node and the new ref without dropping
	 * the spin lock
823
	 */
824
	head_ref = add_delayed_ref_head(fs_info, trans, &head_ref->node, record,
825
					bytenr, num_bytes, 0, 0, action, 0,
826 827
					&qrecord_inserted, old_ref_mod,
					new_ref_mod);
828

829
	add_delayed_tree_ref(fs_info, trans, head_ref, &ref->node, bytenr,
830
			     num_bytes, parent, ref_root, level, action);
831
	spin_unlock(&delayed_refs->lock);
832

833 834
	if (qrecord_inserted)
		return btrfs_qgroup_trace_extent_post(fs_info, record);
835
	return 0;
836 837 838 839 840 841 842

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;
843 844 845 846 847
}

/*
 * add a delayed data ref. it's similar to btrfs_add_delayed_tree_ref.
 */
A
Arne Jansen 已提交
848 849
int btrfs_add_delayed_data_ref(struct btrfs_fs_info *fs_info,
			       struct btrfs_trans_handle *trans,
850 851
			       u64 bytenr, u64 num_bytes,
			       u64 parent, u64 ref_root,
852 853
			       u64 owner, u64 offset, u64 reserved, int action,
			       int *old_ref_mod, int *new_ref_mod)
854 855 856 857
{
	struct btrfs_delayed_data_ref *ref;
	struct btrfs_delayed_ref_head *head_ref;
	struct btrfs_delayed_ref_root *delayed_refs;
858
	struct btrfs_qgroup_extent_record *record = NULL;
859
	int qrecord_inserted;
860

861
	ref = kmem_cache_alloc(btrfs_delayed_data_ref_cachep, GFP_NOFS);
862 863
	if (!ref)
		return -ENOMEM;
864

865
	head_ref = kmem_cache_alloc(btrfs_delayed_ref_head_cachep, GFP_NOFS);
866
	if (!head_ref) {
867
		kmem_cache_free(btrfs_delayed_data_ref_cachep, ref);
868 869
		return -ENOMEM;
	}
870

871 872
	if (test_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags) &&
	    is_fstree(ref_root)) {
873 874 875 876 877 878 879 880 881
		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;
		}
	}

882
	head_ref->extent_op = NULL;
883

884 885 886 887 888 889 890
	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
	 */
891
	head_ref = add_delayed_ref_head(fs_info, trans, &head_ref->node, record,
892
					bytenr, num_bytes, ref_root, reserved,
893 894
					action, 1, &qrecord_inserted,
					old_ref_mod, new_ref_mod);
895

896
	add_delayed_data_ref(fs_info, trans, head_ref, &ref->node, bytenr,
A
Arne Jansen 已提交
897
				   num_bytes, parent, ref_root, owner, offset,
898
				   action);
899
	spin_unlock(&delayed_refs->lock);
900

901 902
	if (qrecord_inserted)
		return btrfs_qgroup_trace_extent_post(fs_info, record);
903 904 905
	return 0;
}

A
Arne Jansen 已提交
906 907
int btrfs_add_delayed_extent_op(struct btrfs_fs_info *fs_info,
				struct btrfs_trans_handle *trans,
908 909 910 911 912 913
				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;

914
	head_ref = kmem_cache_alloc(btrfs_delayed_ref_head_cachep, GFP_NOFS);
915 916 917 918 919 920 921 922
	if (!head_ref)
		return -ENOMEM;

	head_ref->extent_op = extent_op;

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

923
	add_delayed_ref_head(fs_info, trans, &head_ref->node, NULL, bytenr,
924
			     num_bytes, 0, 0, BTRFS_UPDATE_DELAYED_HEAD,
925
			     extent_op->is_data, NULL, NULL, NULL);
926

927 928 929 930
	spin_unlock(&delayed_refs->lock);
	return 0;
}

931 932 933 934 935 936
/*
 * 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 *
937
btrfs_find_delayed_ref_head(struct btrfs_delayed_ref_root *delayed_refs, u64 bytenr)
938
{
939
	return find_ref_head(&delayed_refs->href_root, bytenr, 0);
940
}
941 942 943

void btrfs_delayed_ref_exit(void)
{
944 945 946 947
	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);
948 949 950 951 952 953 954
}

int btrfs_delayed_ref_init(void)
{
	btrfs_delayed_ref_head_cachep = kmem_cache_create(
				"btrfs_delayed_ref_head",
				sizeof(struct btrfs_delayed_ref_head), 0,
955
				SLAB_MEM_SPREAD, NULL);
956 957 958 959 960 961
	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,
962
				SLAB_MEM_SPREAD, NULL);
963 964 965 966 967 968
	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,
969
				SLAB_MEM_SPREAD, NULL);
970 971 972 973 974 975
	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,
976
				SLAB_MEM_SPREAD, NULL);
977 978 979 980 981 982 983 984
	if (!btrfs_delayed_extent_op_cachep)
		goto fail;

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