backref.c 60.3 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
/*
 * Copyright (C) 2011 STRATO.  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.
 */

19
#include <linux/mm.h>
20
#include <linux/rbtree.h>
21 22 23
#include "ctree.h"
#include "disk-io.h"
#include "backref.h"
24 25 26
#include "ulist.h"
#include "transaction.h"
#include "delayed-ref.h"
27
#include "locking.h"
28

29 30 31 32 33
enum merge_mode {
	MERGE_IDENTICAL_KEYS = 1,
	MERGE_IDENTICAL_PARENTS,
};

34 35 36
/* Just an arbitrary number so we can be sure this happened */
#define BACKREF_FOUND_SHARED 6

37 38 39 40 41 42
struct extent_inode_elem {
	u64 inum;
	u64 offset;
	struct extent_inode_elem *next;
};

43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301
/*
 * ref_root is used as the root of the ref tree that hold a collection
 * of unique references.
 */
struct ref_root {
	struct rb_root rb_root;

	/*
	 * The unique_refs represents the number of ref_nodes with a positive
	 * count stored in the tree. Even if a ref_node (the count is greater
	 * than one) is added, the unique_refs will only increase by one.
	 */
	unsigned int unique_refs;
};

/* ref_node is used to store a unique reference to the ref tree. */
struct ref_node {
	struct rb_node rb_node;

	/* For NORMAL_REF, otherwise all these fields should be set to 0 */
	u64 root_id;
	u64 object_id;
	u64 offset;

	/* For SHARED_REF, otherwise parent field should be set to 0 */
	u64 parent;

	/* Ref to the ref_mod of btrfs_delayed_ref_node */
	int ref_mod;
};

/* Dynamically allocate and initialize a ref_root */
static struct ref_root *ref_root_alloc(void)
{
	struct ref_root *ref_tree;

	ref_tree = kmalloc(sizeof(*ref_tree), GFP_NOFS);
	if (!ref_tree)
		return NULL;

	ref_tree->rb_root = RB_ROOT;
	ref_tree->unique_refs = 0;

	return ref_tree;
}

/* Free all nodes in the ref tree, and reinit ref_root */
static void ref_root_fini(struct ref_root *ref_tree)
{
	struct ref_node *node;
	struct rb_node *next;

	while ((next = rb_first(&ref_tree->rb_root)) != NULL) {
		node = rb_entry(next, struct ref_node, rb_node);
		rb_erase(next, &ref_tree->rb_root);
		kfree(node);
	}

	ref_tree->rb_root = RB_ROOT;
	ref_tree->unique_refs = 0;
}

static void ref_root_free(struct ref_root *ref_tree)
{
	if (!ref_tree)
		return;

	ref_root_fini(ref_tree);
	kfree(ref_tree);
}

/*
 * Compare ref_node with (root_id, object_id, offset, parent)
 *
 * The function compares two ref_node a and b. It returns an integer less
 * than, equal to, or greater than zero , respectively, to be less than, to
 * equal, or be greater than b.
 */
static int ref_node_cmp(struct ref_node *a, struct ref_node *b)
{
	if (a->root_id < b->root_id)
		return -1;
	else if (a->root_id > b->root_id)
		return 1;

	if (a->object_id < b->object_id)
		return -1;
	else if (a->object_id > b->object_id)
		return 1;

	if (a->offset < b->offset)
		return -1;
	else if (a->offset > b->offset)
		return 1;

	if (a->parent < b->parent)
		return -1;
	else if (a->parent > b->parent)
		return 1;

	return 0;
}

/*
 * Search ref_node with (root_id, object_id, offset, parent) in the tree
 *
 * if found, the pointer of the ref_node will be returned;
 * if not found, NULL will be returned and pos will point to the rb_node for
 * insert, pos_parent will point to pos'parent for insert;
*/
static struct ref_node *__ref_tree_search(struct ref_root *ref_tree,
					  struct rb_node ***pos,
					  struct rb_node **pos_parent,
					  u64 root_id, u64 object_id,
					  u64 offset, u64 parent)
{
	struct ref_node *cur = NULL;
	struct ref_node entry;
	int ret;

	entry.root_id = root_id;
	entry.object_id = object_id;
	entry.offset = offset;
	entry.parent = parent;

	*pos = &ref_tree->rb_root.rb_node;

	while (**pos) {
		*pos_parent = **pos;
		cur = rb_entry(*pos_parent, struct ref_node, rb_node);

		ret = ref_node_cmp(cur, &entry);
		if (ret > 0)
			*pos = &(**pos)->rb_left;
		else if (ret < 0)
			*pos = &(**pos)->rb_right;
		else
			return cur;
	}

	return NULL;
}

/*
 * Insert a ref_node to the ref tree
 * @pos used for specifiy the position to insert
 * @pos_parent for specifiy pos's parent
 *
 * success, return 0;
 * ref_node already exists, return -EEXIST;
*/
static int ref_tree_insert(struct ref_root *ref_tree, struct rb_node **pos,
			   struct rb_node *pos_parent, struct ref_node *ins)
{
	struct rb_node **p = NULL;
	struct rb_node *parent = NULL;
	struct ref_node *cur = NULL;

	if (!pos) {
		cur = __ref_tree_search(ref_tree, &p, &parent, ins->root_id,
					ins->object_id, ins->offset,
					ins->parent);
		if (cur)
			return -EEXIST;
	} else {
		p = pos;
		parent = pos_parent;
	}

	rb_link_node(&ins->rb_node, parent, p);
	rb_insert_color(&ins->rb_node, &ref_tree->rb_root);

	return 0;
}

/* Erase and free ref_node, caller should update ref_root->unique_refs */
static void ref_tree_remove(struct ref_root *ref_tree, struct ref_node *node)
{
	rb_erase(&node->rb_node, &ref_tree->rb_root);
	kfree(node);
}

/*
 * Update ref_root->unique_refs
 *
 * Call __ref_tree_search
 *	1. if ref_node doesn't exist, ref_tree_insert this node, and update
 *	ref_root->unique_refs:
 *		if ref_node->ref_mod > 0, ref_root->unique_refs++;
 *		if ref_node->ref_mod < 0, do noting;
 *
 *	2. if ref_node is found, then get origin ref_node->ref_mod, and update
 *	ref_node->ref_mod.
 *		if ref_node->ref_mod is equal to 0,then call ref_tree_remove
 *
 *		according to origin_mod and new_mod, update ref_root->items
 *		+----------------+--------------+-------------+
 *		|		 |new_count <= 0|new_count > 0|
 *		+----------------+--------------+-------------+
 *		|origin_count < 0|       0      |      1      |
 *		+----------------+--------------+-------------+
 *		|origin_count > 0|      -1      |      0      |
 *		+----------------+--------------+-------------+
 *
 * In case of allocation failure, -ENOMEM is returned and the ref_tree stays
 * unaltered.
 * Success, return 0
 */
static int ref_tree_add(struct ref_root *ref_tree, u64 root_id, u64 object_id,
			u64 offset, u64 parent, int count)
{
	struct ref_node *node = NULL;
	struct rb_node **pos = NULL;
	struct rb_node *pos_parent = NULL;
	int origin_count;
	int ret;

	if (!count)
		return 0;

	node = __ref_tree_search(ref_tree, &pos, &pos_parent, root_id,
				 object_id, offset, parent);
	if (node == NULL) {
		node = kmalloc(sizeof(*node), GFP_NOFS);
		if (!node)
			return -ENOMEM;

		node->root_id = root_id;
		node->object_id = object_id;
		node->offset = offset;
		node->parent = parent;
		node->ref_mod = count;

		ret = ref_tree_insert(ref_tree, pos, pos_parent, node);
		ASSERT(!ret);
		if (ret) {
			kfree(node);
			return ret;
		}

		ref_tree->unique_refs += node->ref_mod > 0 ? 1 : 0;

		return 0;
	}

	origin_count = node->ref_mod;
	node->ref_mod += count;

	if (node->ref_mod > 0)
		ref_tree->unique_refs += origin_count > 0 ? 0 : 1;
	else if (node->ref_mod <= 0)
		ref_tree->unique_refs += origin_count > 0 ? -1 : 0;

	if (!node->ref_mod)
		ref_tree_remove(ref_tree, node);

	return 0;
}

302 303 304 305 306
static int check_extent_in_eb(const struct btrfs_key *key,
			      const struct extent_buffer *eb,
			      const struct btrfs_file_extent_item *fi,
			      u64 extent_item_pos,
			      struct extent_inode_elem **eie)
307
{
308
	u64 offset = 0;
309 310
	struct extent_inode_elem *e;

311 312 313 314 315
	if (!btrfs_file_extent_compression(eb, fi) &&
	    !btrfs_file_extent_encryption(eb, fi) &&
	    !btrfs_file_extent_other_encoding(eb, fi)) {
		u64 data_offset;
		u64 data_len;
316

317 318 319 320 321 322 323 324
		data_offset = btrfs_file_extent_offset(eb, fi);
		data_len = btrfs_file_extent_num_bytes(eb, fi);

		if (extent_item_pos < data_offset ||
		    extent_item_pos >= data_offset + data_len)
			return 1;
		offset = extent_item_pos - data_offset;
	}
325 326 327 328 329 330 331

	e = kmalloc(sizeof(*e), GFP_NOFS);
	if (!e)
		return -ENOMEM;

	e->next = *eie;
	e->inum = key->objectid;
332
	e->offset = key->offset + offset;
333 334 335 336 337
	*eie = e;

	return 0;
}

338 339 340 341 342 343 344 345 346 347
static void free_inode_elem_list(struct extent_inode_elem *eie)
{
	struct extent_inode_elem *eie_next;

	for (; eie; eie = eie_next) {
		eie_next = eie->next;
		kfree(eie);
	}
}

348 349 350
static int find_extent_in_eb(const struct extent_buffer *eb,
			     u64 wanted_disk_byte, u64 extent_item_pos,
			     struct extent_inode_elem **eie)
351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386
{
	u64 disk_byte;
	struct btrfs_key key;
	struct btrfs_file_extent_item *fi;
	int slot;
	int nritems;
	int extent_type;
	int ret;

	/*
	 * from the shared data ref, we only have the leaf but we need
	 * the key. thus, we must look into all items and see that we
	 * find one (some) with a reference to our extent item.
	 */
	nritems = btrfs_header_nritems(eb);
	for (slot = 0; slot < nritems; ++slot) {
		btrfs_item_key_to_cpu(eb, &key, slot);
		if (key.type != BTRFS_EXTENT_DATA_KEY)
			continue;
		fi = btrfs_item_ptr(eb, slot, struct btrfs_file_extent_item);
		extent_type = btrfs_file_extent_type(eb, fi);
		if (extent_type == BTRFS_FILE_EXTENT_INLINE)
			continue;
		/* don't skip BTRFS_FILE_EXTENT_PREALLOC, we can handle that */
		disk_byte = btrfs_file_extent_disk_bytenr(eb, fi);
		if (disk_byte != wanted_disk_byte)
			continue;

		ret = check_extent_in_eb(&key, eb, fi, extent_item_pos, eie);
		if (ret < 0)
			return ret;
	}

	return 0;
}

387 388 389 390 391 392
/*
 * this structure records all encountered refs on the way up to the root
 */
struct __prelim_ref {
	struct list_head list;
	u64 root_id;
393
	struct btrfs_key key_for_search;
394 395
	int level;
	int count;
396
	struct extent_inode_elem *inode_list;
397 398 399 400
	u64 parent;
	u64 wanted_disk_byte;
};

401 402 403 404 405 406 407
static struct kmem_cache *btrfs_prelim_ref_cache;

int __init btrfs_prelim_ref_init(void)
{
	btrfs_prelim_ref_cache = kmem_cache_create("btrfs_prelim_ref",
					sizeof(struct __prelim_ref),
					0,
408
					SLAB_MEM_SPREAD,
409 410 411 412 413 414 415 416
					NULL);
	if (!btrfs_prelim_ref_cache)
		return -ENOMEM;
	return 0;
}

void btrfs_prelim_ref_exit(void)
{
417
	kmem_cache_destroy(btrfs_prelim_ref_cache);
418 419
}

420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458
/*
 * the rules for all callers of this function are:
 * - obtaining the parent is the goal
 * - if you add a key, you must know that it is a correct key
 * - if you cannot add the parent or a correct key, then we will look into the
 *   block later to set a correct key
 *
 * delayed refs
 * ============
 *        backref type | shared | indirect | shared | indirect
 * information         |   tree |     tree |   data |     data
 * --------------------+--------+----------+--------+----------
 *      parent logical |    y   |     -    |    -   |     -
 *      key to resolve |    -   |     y    |    y   |     y
 *  tree block logical |    -   |     -    |    -   |     -
 *  root for resolving |    y   |     y    |    y   |     y
 *
 * - column 1:       we've the parent -> done
 * - column 2, 3, 4: we use the key to find the parent
 *
 * on disk refs (inline or keyed)
 * ==============================
 *        backref type | shared | indirect | shared | indirect
 * information         |   tree |     tree |   data |     data
 * --------------------+--------+----------+--------+----------
 *      parent logical |    y   |     -    |    y   |     -
 *      key to resolve |    -   |     -    |    -   |     y
 *  tree block logical |    y   |     y    |    y   |     y
 *  root for resolving |    -   |     y    |    y   |     y
 *
 * - column 1, 3: we've the parent -> done
 * - column 2:    we take the first key from the block to find the parent
 *                (see __add_missing_keys)
 * - column 4:    we use the key to find the parent
 *
 * additional information that's available but not required to find the parent
 * block might help in merging entries to gain some speed.
 */

459
static int __add_prelim_ref(struct list_head *head, u64 root_id,
460
			    const struct btrfs_key *key, int level,
461 462
			    u64 parent, u64 wanted_disk_byte, int count,
			    gfp_t gfp_mask)
463 464 465
{
	struct __prelim_ref *ref;

466 467 468
	if (root_id == BTRFS_DATA_RELOC_TREE_OBJECTID)
		return 0;

469
	ref = kmem_cache_alloc(btrfs_prelim_ref_cache, gfp_mask);
470 471 472 473
	if (!ref)
		return -ENOMEM;

	ref->root_id = root_id;
474
	if (key) {
475
		ref->key_for_search = *key;
476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498
		/*
		 * We can often find data backrefs with an offset that is too
		 * large (>= LLONG_MAX, maximum allowed file offset) due to
		 * underflows when subtracting a file's offset with the data
		 * offset of its corresponding extent data item. This can
		 * happen for example in the clone ioctl.
		 * So if we detect such case we set the search key's offset to
		 * zero to make sure we will find the matching file extent item
		 * at add_all_parents(), otherwise we will miss it because the
		 * offset taken form the backref is much larger then the offset
		 * of the file extent item. This can make us scan a very large
		 * number of file extent items, but at least it will not make
		 * us miss any.
		 * This is an ugly workaround for a behaviour that should have
		 * never existed, but it does and a fix for the clone ioctl
		 * would touch a lot of places, cause backwards incompatibility
		 * and would not fix the problem for extents cloned with older
		 * kernels.
		 */
		if (ref->key_for_search.type == BTRFS_EXTENT_DATA_KEY &&
		    ref->key_for_search.offset >= LLONG_MAX)
			ref->key_for_search.offset = 0;
	} else {
499
		memset(&ref->key_for_search, 0, sizeof(ref->key_for_search));
500
	}
501

502
	ref->inode_list = NULL;
503 504 505 506 507 508 509 510 511 512
	ref->level = level;
	ref->count = count;
	ref->parent = parent;
	ref->wanted_disk_byte = wanted_disk_byte;
	list_add_tail(&ref->list, head);

	return 0;
}

static int add_all_parents(struct btrfs_root *root, struct btrfs_path *path,
513
			   struct ulist *parents, struct __prelim_ref *ref,
514 515
			   int level, u64 time_seq, const u64 *extent_item_pos,
			   u64 total_refs)
516
{
517 518 519 520
	int ret = 0;
	int slot;
	struct extent_buffer *eb;
	struct btrfs_key key;
521
	struct btrfs_key *key_for_search = &ref->key_for_search;
522
	struct btrfs_file_extent_item *fi;
523
	struct extent_inode_elem *eie = NULL, *old = NULL;
524
	u64 disk_byte;
525 526
	u64 wanted_disk_byte = ref->wanted_disk_byte;
	u64 count = 0;
527

528 529 530
	if (level != 0) {
		eb = path->nodes[level];
		ret = ulist_add(parents, eb->start, 0, GFP_NOFS);
531 532
		if (ret < 0)
			return ret;
533
		return 0;
534
	}
535 536

	/*
537 538 539
	 * We normally enter this function with the path already pointing to
	 * the first item to check. But sometimes, we may enter it with
	 * slot==nritems. In that case, go to the next leaf before we continue.
540
	 */
541
	if (path->slots[0] >= btrfs_header_nritems(path->nodes[0])) {
542
		if (time_seq == SEQ_LAST)
543 544 545 546
			ret = btrfs_next_leaf(root, path);
		else
			ret = btrfs_next_old_leaf(root, path, time_seq);
	}
547

548
	while (!ret && count < total_refs) {
549
		eb = path->nodes[0];
550 551 552 553 554 555 556 557 558 559 560 561 562
		slot = path->slots[0];

		btrfs_item_key_to_cpu(eb, &key, slot);

		if (key.objectid != key_for_search->objectid ||
		    key.type != BTRFS_EXTENT_DATA_KEY)
			break;

		fi = btrfs_item_ptr(eb, slot, struct btrfs_file_extent_item);
		disk_byte = btrfs_file_extent_disk_bytenr(eb, fi);

		if (disk_byte == wanted_disk_byte) {
			eie = NULL;
563
			old = NULL;
564
			count++;
565 566 567 568 569 570 571
			if (extent_item_pos) {
				ret = check_extent_in_eb(&key, eb, fi,
						*extent_item_pos,
						&eie);
				if (ret < 0)
					break;
			}
572 573
			if (ret > 0)
				goto next;
574 575
			ret = ulist_add_merge_ptr(parents, eb->start,
						  eie, (void **)&old, GFP_NOFS);
576 577 578 579 580 581
			if (ret < 0)
				break;
			if (!ret && extent_item_pos) {
				while (old->next)
					old = old->next;
				old->next = eie;
582
			}
583
			eie = NULL;
584
		}
585
next:
586
		if (time_seq == SEQ_LAST)
587 588 589
			ret = btrfs_next_item(root, path);
		else
			ret = btrfs_next_old_item(root, path, time_seq);
590 591
	}

592 593
	if (ret > 0)
		ret = 0;
594 595
	else if (ret < 0)
		free_inode_elem_list(eie);
596
	return ret;
597 598 599 600 601 602 603
}

/*
 * resolve an indirect backref in the form (root_id, key, level)
 * to a logical address
 */
static int __resolve_indirect_ref(struct btrfs_fs_info *fs_info,
604 605 606
				  struct btrfs_path *path, u64 time_seq,
				  struct __prelim_ref *ref,
				  struct ulist *parents,
607
				  const u64 *extent_item_pos, u64 total_refs)
608 609 610 611 612 613 614
{
	struct btrfs_root *root;
	struct btrfs_key root_key;
	struct extent_buffer *eb;
	int ret = 0;
	int root_level;
	int level = ref->level;
615
	int index;
616 617 618 619

	root_key.objectid = ref->root_id;
	root_key.type = BTRFS_ROOT_ITEM_KEY;
	root_key.offset = (u64)-1;
620 621 622

	index = srcu_read_lock(&fs_info->subvol_srcu);

623
	root = btrfs_get_fs_root(fs_info, &root_key, false);
624
	if (IS_ERR(root)) {
625
		srcu_read_unlock(&fs_info->subvol_srcu, index);
626 627 628 629
		ret = PTR_ERR(root);
		goto out;
	}

630
	if (btrfs_is_testing(fs_info)) {
J
Josef Bacik 已提交
631 632 633 634 635
		srcu_read_unlock(&fs_info->subvol_srcu, index);
		ret = -ENOENT;
		goto out;
	}

636 637
	if (path->search_commit_root)
		root_level = btrfs_header_level(root->commit_root);
638
	else if (time_seq == SEQ_LAST)
639
		root_level = btrfs_header_level(root->node);
640 641
	else
		root_level = btrfs_old_root_level(root, time_seq);
642

643 644
	if (root_level + 1 == level) {
		srcu_read_unlock(&fs_info->subvol_srcu, index);
645
		goto out;
646
	}
647 648

	path->lowest_level = level;
649
	if (time_seq == SEQ_LAST)
650 651 652 653 654
		ret = btrfs_search_slot(NULL, root, &ref->key_for_search, path,
					0, 0);
	else
		ret = btrfs_search_old_slot(root, &ref->key_for_search, path,
					    time_seq);
655 656 657 658

	/* root node has been locked, we can release @subvol_srcu safely here */
	srcu_read_unlock(&fs_info->subvol_srcu, index);

659 660
	btrfs_debug(fs_info,
		"search slot in root %llu (level %d, ref count %d) returned %d for key (%llu %u %llu)",
661 662 663
		 ref->root_id, level, ref->count, ret,
		 ref->key_for_search.objectid, ref->key_for_search.type,
		 ref->key_for_search.offset);
664 665 666 667
	if (ret < 0)
		goto out;

	eb = path->nodes[level];
668
	while (!eb) {
669
		if (WARN_ON(!level)) {
670 671 672 673 674
			ret = 1;
			goto out;
		}
		level--;
		eb = path->nodes[level];
675 676
	}

677
	ret = add_all_parents(root, path, parents, ref, level, time_seq,
678
			      extent_item_pos, total_refs);
679
out:
680 681
	path->lowest_level = 0;
	btrfs_release_path(path);
682 683 684
	return ret;
}

685 686 687 688 689 690 691 692
static struct extent_inode_elem *
unode_aux_to_inode_list(struct ulist_node *node)
{
	if (!node)
		return NULL;
	return (struct extent_inode_elem *)(uintptr_t)node->aux;
}

693 694 695 696
/*
 * resolve all indirect backrefs from the list
 */
static int __resolve_indirect_refs(struct btrfs_fs_info *fs_info,
697
				   struct btrfs_path *path, u64 time_seq,
698
				   struct list_head *head,
699 700
				   const u64 *extent_item_pos, u64 total_refs,
				   u64 root_objectid)
701 702 703 704 705 706 707 708
{
	int err;
	int ret = 0;
	struct __prelim_ref *ref;
	struct __prelim_ref *ref_safe;
	struct __prelim_ref *new_ref;
	struct ulist *parents;
	struct ulist_node *node;
J
Jan Schmidt 已提交
709
	struct ulist_iterator uiter;
710 711 712 713 714 715 716 717 718 719 720 721 722 723 724

	parents = ulist_alloc(GFP_NOFS);
	if (!parents)
		return -ENOMEM;

	/*
	 * _safe allows us to insert directly after the current item without
	 * iterating over the newly inserted items.
	 * we're also allowed to re-assign ref during iteration.
	 */
	list_for_each_entry_safe(ref, ref_safe, head, list) {
		if (ref->parent)	/* already direct */
			continue;
		if (ref->count == 0)
			continue;
725 726 727 728
		if (root_objectid && ref->root_id != root_objectid) {
			ret = BACKREF_FOUND_SHARED;
			goto out;
		}
729
		err = __resolve_indirect_ref(fs_info, path, time_seq, ref,
730 731
					     parents, extent_item_pos,
					     total_refs);
732 733 734 735 736
		/*
		 * we can only tolerate ENOENT,otherwise,we should catch error
		 * and return directly.
		 */
		if (err == -ENOENT) {
737
			continue;
738 739 740 741
		} else if (err) {
			ret = err;
			goto out;
		}
742 743

		/* we put the first parent into the ref at hand */
J
Jan Schmidt 已提交
744 745
		ULIST_ITER_INIT(&uiter);
		node = ulist_next(parents, &uiter);
746
		ref->parent = node ? node->val : 0;
747
		ref->inode_list = unode_aux_to_inode_list(node);
748 749

		/* additional parents require new refs being added here */
J
Jan Schmidt 已提交
750
		while ((node = ulist_next(parents, &uiter))) {
751 752
			new_ref = kmem_cache_alloc(btrfs_prelim_ref_cache,
						   GFP_NOFS);
753 754
			if (!new_ref) {
				ret = -ENOMEM;
755
				goto out;
756 757 758
			}
			memcpy(new_ref, ref, sizeof(*ref));
			new_ref->parent = node->val;
759
			new_ref->inode_list = unode_aux_to_inode_list(node);
760 761 762 763
			list_add(&new_ref->list, &ref->list);
		}
		ulist_reinit(parents);
	}
764
out:
765 766 767 768
	ulist_free(parents);
	return ret;
}

769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793
static inline int ref_for_same_block(struct __prelim_ref *ref1,
				     struct __prelim_ref *ref2)
{
	if (ref1->level != ref2->level)
		return 0;
	if (ref1->root_id != ref2->root_id)
		return 0;
	if (ref1->key_for_search.type != ref2->key_for_search.type)
		return 0;
	if (ref1->key_for_search.objectid != ref2->key_for_search.objectid)
		return 0;
	if (ref1->key_for_search.offset != ref2->key_for_search.offset)
		return 0;
	if (ref1->parent != ref2->parent)
		return 0;

	return 1;
}

/*
 * read tree blocks and add keys where required.
 */
static int __add_missing_keys(struct btrfs_fs_info *fs_info,
			      struct list_head *head)
{
794
	struct __prelim_ref *ref;
795 796
	struct extent_buffer *eb;

797
	list_for_each_entry(ref, head, list) {
798 799 800 801 802
		if (ref->parent)
			continue;
		if (ref->key_for_search.type)
			continue;
		BUG_ON(!ref->wanted_disk_byte);
803
		eb = read_tree_block(fs_info, ref->wanted_disk_byte, 0);
804 805 806
		if (IS_ERR(eb)) {
			return PTR_ERR(eb);
		} else if (!extent_buffer_uptodate(eb)) {
807 808 809
			free_extent_buffer(eb);
			return -EIO;
		}
810 811 812 813 814 815 816 817 818 819 820
		btrfs_tree_read_lock(eb);
		if (btrfs_header_level(eb) == 0)
			btrfs_item_key_to_cpu(eb, &ref->key_for_search, 0);
		else
			btrfs_node_key_to_cpu(eb, &ref->key_for_search, 0);
		btrfs_tree_read_unlock(eb);
		free_extent_buffer(eb);
	}
	return 0;
}

821
/*
822
 * merge backrefs and adjust counts accordingly
823
 *
824 825 826 827
 *    FIXME: For MERGE_IDENTICAL_KEYS, if we add more keys in __add_prelim_ref
 *           then we can merge more here. Additionally, we could even add a key
 *           range for the blocks we looked into to merge even more (-> replace
 *           unresolved refs by those having a parent).
828
 */
829
static void __merge_refs(struct list_head *head, enum merge_mode mode)
830
{
831
	struct __prelim_ref *pos1;
832

833 834
	list_for_each_entry(pos1, head, list) {
		struct __prelim_ref *pos2 = pos1, *tmp;
835

836
		list_for_each_entry_safe_continue(pos2, tmp, head, list) {
837
			struct __prelim_ref *ref1 = pos1, *ref2 = pos2;
838
			struct extent_inode_elem *eie;
839

840 841
			if (!ref_for_same_block(ref1, ref2))
				continue;
842
			if (mode == MERGE_IDENTICAL_KEYS) {
843 844
				if (!ref1->parent && ref2->parent)
					swap(ref1, ref2);
845 846 847 848
			} else {
				if (ref1->parent != ref2->parent)
					continue;
			}
849 850 851 852 853 854 855 856 857 858

			eie = ref1->inode_list;
			while (eie && eie->next)
				eie = eie->next;
			if (eie)
				eie->next = ref2->inode_list;
			else
				ref1->inode_list = ref2->inode_list;
			ref1->count += ref2->count;

859
			list_del(&ref2->list);
860
			kmem_cache_free(btrfs_prelim_ref_cache, ref2);
861
			cond_resched();
862 863 864 865 866 867 868 869 870 871
		}

	}
}

/*
 * add all currently queued delayed refs from this head whose seq nr is
 * smaller or equal that seq to the list
 */
static int __add_delayed_refs(struct btrfs_delayed_ref_head *head, u64 seq,
872 873
			      struct list_head *prefs, u64 *total_refs,
			      u64 inum)
874
{
875
	struct btrfs_delayed_ref_node *node;
876
	struct btrfs_delayed_extent_op *extent_op = head->extent_op;
877 878
	struct btrfs_key key;
	struct btrfs_key op_key = {0};
879
	int sgn;
880
	int ret = 0;
881 882

	if (extent_op && extent_op->update_key)
883
		btrfs_disk_key_to_cpu(&op_key, &extent_op->key);
884

885
	spin_lock(&head->lock);
886
	list_for_each_entry(node, &head->ref_list, list) {
887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903
		if (node->seq > seq)
			continue;

		switch (node->action) {
		case BTRFS_ADD_DELAYED_EXTENT:
		case BTRFS_UPDATE_DELAYED_HEAD:
			WARN_ON(1);
			continue;
		case BTRFS_ADD_DELAYED_REF:
			sgn = 1;
			break;
		case BTRFS_DROP_DELAYED_REF:
			sgn = -1;
			break;
		default:
			BUG_ON(1);
		}
904
		*total_refs += (node->ref_mod * sgn);
905 906 907 908 909
		switch (node->type) {
		case BTRFS_TREE_BLOCK_REF_KEY: {
			struct btrfs_delayed_tree_ref *ref;

			ref = btrfs_delayed_node_to_tree_ref(node);
910
			ret = __add_prelim_ref(prefs, ref->root, &op_key,
911
					       ref->level + 1, 0, node->bytenr,
912
					       node->ref_mod * sgn, GFP_ATOMIC);
913 914 915 916 917 918
			break;
		}
		case BTRFS_SHARED_BLOCK_REF_KEY: {
			struct btrfs_delayed_tree_ref *ref;

			ref = btrfs_delayed_node_to_tree_ref(node);
L
Liu Bo 已提交
919
			ret = __add_prelim_ref(prefs, 0, NULL,
920 921
					       ref->level + 1, ref->parent,
					       node->bytenr,
922
					       node->ref_mod * sgn, GFP_ATOMIC);
923 924 925 926 927 928 929 930 931
			break;
		}
		case BTRFS_EXTENT_DATA_REF_KEY: {
			struct btrfs_delayed_data_ref *ref;
			ref = btrfs_delayed_node_to_data_ref(node);

			key.objectid = ref->objectid;
			key.type = BTRFS_EXTENT_DATA_KEY;
			key.offset = ref->offset;
932 933 934 935 936 937 938 939 940 941

			/*
			 * Found a inum that doesn't match our known inum, we
			 * know it's shared.
			 */
			if (inum && ref->objectid != inum) {
				ret = BACKREF_FOUND_SHARED;
				break;
			}

942 943
			ret = __add_prelim_ref(prefs, ref->root, &key, 0, 0,
					       node->bytenr,
944
					       node->ref_mod * sgn, GFP_ATOMIC);
945 946 947 948 949 950
			break;
		}
		case BTRFS_SHARED_DATA_REF_KEY: {
			struct btrfs_delayed_data_ref *ref;

			ref = btrfs_delayed_node_to_data_ref(node);
L
Liu Bo 已提交
951
			ret = __add_prelim_ref(prefs, 0, NULL, 0,
952
					       ref->parent, node->bytenr,
953
					       node->ref_mod * sgn, GFP_ATOMIC);
954 955 956 957 958
			break;
		}
		default:
			WARN_ON(1);
		}
959
		if (ret)
960
			break;
961
	}
962 963
	spin_unlock(&head->lock);
	return ret;
964 965 966 967 968
}

/*
 * add all inline backrefs for bytenr to the list
 */
969
static int __add_inline_refs(struct btrfs_path *path, u64 bytenr,
970
			     int *info_level, struct list_head *prefs,
971
			     struct ref_root *ref_tree,
972
			     u64 *total_refs, u64 inum)
973
{
974
	int ret = 0;
975 976 977
	int slot;
	struct extent_buffer *leaf;
	struct btrfs_key key;
978
	struct btrfs_key found_key;
979 980 981 982 983 984 985 986 987 988
	unsigned long ptr;
	unsigned long end;
	struct btrfs_extent_item *ei;
	u64 flags;
	u64 item_size;

	/*
	 * enumerate all inline refs
	 */
	leaf = path->nodes[0];
989
	slot = path->slots[0];
990 991 992 993 994 995

	item_size = btrfs_item_size_nr(leaf, slot);
	BUG_ON(item_size < sizeof(*ei));

	ei = btrfs_item_ptr(leaf, slot, struct btrfs_extent_item);
	flags = btrfs_extent_flags(leaf, ei);
996
	*total_refs += btrfs_extent_refs(leaf, ei);
997
	btrfs_item_key_to_cpu(leaf, &found_key, slot);
998 999 1000 1001

	ptr = (unsigned long)(ei + 1);
	end = (unsigned long)ei + item_size;

1002 1003
	if (found_key.type == BTRFS_EXTENT_ITEM_KEY &&
	    flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
1004 1005 1006 1007 1008 1009
		struct btrfs_tree_block_info *info;

		info = (struct btrfs_tree_block_info *)ptr;
		*info_level = btrfs_tree_block_level(leaf, info);
		ptr += sizeof(struct btrfs_tree_block_info);
		BUG_ON(ptr > end);
1010 1011
	} else if (found_key.type == BTRFS_METADATA_ITEM_KEY) {
		*info_level = found_key.offset;
1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026
	} else {
		BUG_ON(!(flags & BTRFS_EXTENT_FLAG_DATA));
	}

	while (ptr < end) {
		struct btrfs_extent_inline_ref *iref;
		u64 offset;
		int type;

		iref = (struct btrfs_extent_inline_ref *)ptr;
		type = btrfs_extent_inline_ref_type(leaf, iref);
		offset = btrfs_extent_inline_ref_offset(leaf, iref);

		switch (type) {
		case BTRFS_SHARED_BLOCK_REF_KEY:
1027
			ret = __add_prelim_ref(prefs, 0, NULL,
1028
						*info_level + 1, offset,
1029
						bytenr, 1, GFP_NOFS);
1030 1031 1032 1033 1034 1035 1036 1037
			break;
		case BTRFS_SHARED_DATA_REF_KEY: {
			struct btrfs_shared_data_ref *sdref;
			int count;

			sdref = (struct btrfs_shared_data_ref *)(iref + 1);
			count = btrfs_shared_data_ref_count(leaf, sdref);
			ret = __add_prelim_ref(prefs, 0, NULL, 0, offset,
1038
					       bytenr, count, GFP_NOFS);
1039 1040 1041 1042 1043 1044 1045
			if (ref_tree) {
				if (!ret)
					ret = ref_tree_add(ref_tree, 0, 0, 0,
							   bytenr, count);
				if (!ret && ref_tree->unique_refs > 1)
					ret = BACKREF_FOUND_SHARED;
			}
1046 1047 1048
			break;
		}
		case BTRFS_TREE_BLOCK_REF_KEY:
1049 1050
			ret = __add_prelim_ref(prefs, offset, NULL,
					       *info_level + 1, 0,
1051
					       bytenr, 1, GFP_NOFS);
1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063
			break;
		case BTRFS_EXTENT_DATA_REF_KEY: {
			struct btrfs_extent_data_ref *dref;
			int count;
			u64 root;

			dref = (struct btrfs_extent_data_ref *)(&iref->offset);
			count = btrfs_extent_data_ref_count(leaf, dref);
			key.objectid = btrfs_extent_data_ref_objectid(leaf,
								      dref);
			key.type = BTRFS_EXTENT_DATA_KEY;
			key.offset = btrfs_extent_data_ref_offset(leaf, dref);
1064 1065 1066 1067 1068 1069

			if (inum && key.objectid != inum) {
				ret = BACKREF_FOUND_SHARED;
				break;
			}

1070
			root = btrfs_extent_data_ref_root(leaf, dref);
1071
			ret = __add_prelim_ref(prefs, root, &key, 0, 0,
1072
					       bytenr, count, GFP_NOFS);
1073 1074 1075 1076 1077 1078 1079 1080 1081
			if (ref_tree) {
				if (!ret)
					ret = ref_tree_add(ref_tree, root,
							   key.objectid,
							   key.offset, 0,
							   count);
				if (!ret && ref_tree->unique_refs > 1)
					ret = BACKREF_FOUND_SHARED;
			}
1082 1083 1084 1085 1086
			break;
		}
		default:
			WARN_ON(1);
		}
1087 1088
		if (ret)
			return ret;
1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099
		ptr += btrfs_extent_inline_ref_size(type);
	}

	return 0;
}

/*
 * add all non-inline backrefs for bytenr to the list
 */
static int __add_keyed_refs(struct btrfs_fs_info *fs_info,
			    struct btrfs_path *path, u64 bytenr,
1100 1101
			    int info_level, struct list_head *prefs,
			    struct ref_root *ref_tree, u64 inum)
1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130
{
	struct btrfs_root *extent_root = fs_info->extent_root;
	int ret;
	int slot;
	struct extent_buffer *leaf;
	struct btrfs_key key;

	while (1) {
		ret = btrfs_next_item(extent_root, path);
		if (ret < 0)
			break;
		if (ret) {
			ret = 0;
			break;
		}

		slot = path->slots[0];
		leaf = path->nodes[0];
		btrfs_item_key_to_cpu(leaf, &key, slot);

		if (key.objectid != bytenr)
			break;
		if (key.type < BTRFS_TREE_BLOCK_REF_KEY)
			continue;
		if (key.type > BTRFS_SHARED_DATA_REF_KEY)
			break;

		switch (key.type) {
		case BTRFS_SHARED_BLOCK_REF_KEY:
1131
			ret = __add_prelim_ref(prefs, 0, NULL,
1132
						info_level + 1, key.offset,
1133
						bytenr, 1, GFP_NOFS);
1134 1135 1136 1137 1138 1139 1140 1141 1142
			break;
		case BTRFS_SHARED_DATA_REF_KEY: {
			struct btrfs_shared_data_ref *sdref;
			int count;

			sdref = btrfs_item_ptr(leaf, slot,
					      struct btrfs_shared_data_ref);
			count = btrfs_shared_data_ref_count(leaf, sdref);
			ret = __add_prelim_ref(prefs, 0, NULL, 0, key.offset,
1143
						bytenr, count, GFP_NOFS);
1144 1145 1146 1147 1148 1149 1150
			if (ref_tree) {
				if (!ret)
					ret = ref_tree_add(ref_tree, 0, 0, 0,
							   bytenr, count);
				if (!ret && ref_tree->unique_refs > 1)
					ret = BACKREF_FOUND_SHARED;
			}
1151 1152 1153
			break;
		}
		case BTRFS_TREE_BLOCK_REF_KEY:
1154 1155
			ret = __add_prelim_ref(prefs, key.offset, NULL,
					       info_level + 1, 0,
1156
					       bytenr, 1, GFP_NOFS);
1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169
			break;
		case BTRFS_EXTENT_DATA_REF_KEY: {
			struct btrfs_extent_data_ref *dref;
			int count;
			u64 root;

			dref = btrfs_item_ptr(leaf, slot,
					      struct btrfs_extent_data_ref);
			count = btrfs_extent_data_ref_count(leaf, dref);
			key.objectid = btrfs_extent_data_ref_objectid(leaf,
								      dref);
			key.type = BTRFS_EXTENT_DATA_KEY;
			key.offset = btrfs_extent_data_ref_offset(leaf, dref);
1170 1171 1172 1173 1174 1175

			if (inum && key.objectid != inum) {
				ret = BACKREF_FOUND_SHARED;
				break;
			}

1176 1177
			root = btrfs_extent_data_ref_root(leaf, dref);
			ret = __add_prelim_ref(prefs, root, &key, 0, 0,
1178
					       bytenr, count, GFP_NOFS);
1179 1180 1181 1182 1183 1184 1185 1186 1187
			if (ref_tree) {
				if (!ret)
					ret = ref_tree_add(ref_tree, root,
							   key.objectid,
							   key.offset, 0,
							   count);
				if (!ret && ref_tree->unique_refs > 1)
					ret = BACKREF_FOUND_SHARED;
			}
1188 1189 1190 1191 1192
			break;
		}
		default:
			WARN_ON(1);
		}
1193 1194 1195
		if (ret)
			return ret;

1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206
	}

	return ret;
}

/*
 * this adds all existing backrefs (inline backrefs, backrefs and delayed
 * refs) for the given bytenr to the refs list, merges duplicates and resolves
 * indirect refs to their parent bytenr.
 * When roots are found, they're added to the roots list
 *
1207 1208
 * NOTE: This can return values > 0
 *
1209
 * If time_seq is set to SEQ_LAST, it will not search delayed_refs, and behave
1210 1211 1212 1213
 * much like trans == NULL case, the difference only lies in it will not
 * commit root.
 * The special case is for qgroup to search roots in commit_transaction().
 *
1214 1215 1216
 * If check_shared is set to 1, any extent has more than one ref item, will
 * be returned BACKREF_FOUND_SHARED immediately.
 *
1217 1218 1219 1220
 * FIXME some caching might speed things up
 */
static int find_parent_nodes(struct btrfs_trans_handle *trans,
			     struct btrfs_fs_info *fs_info, u64 bytenr,
1221
			     u64 time_seq, struct ulist *refs,
1222
			     struct ulist *roots, const u64 *extent_item_pos,
1223
			     u64 root_objectid, u64 inum, int check_shared)
1224 1225 1226 1227
{
	struct btrfs_key key;
	struct btrfs_path *path;
	struct btrfs_delayed_ref_root *delayed_refs = NULL;
1228
	struct btrfs_delayed_ref_head *head;
1229 1230 1231 1232 1233
	int info_level = 0;
	int ret;
	struct list_head prefs_delayed;
	struct list_head prefs;
	struct __prelim_ref *ref;
1234
	struct extent_inode_elem *eie = NULL;
1235
	struct ref_root *ref_tree = NULL;
1236
	u64 total_refs = 0;
1237 1238 1239 1240 1241 1242

	INIT_LIST_HEAD(&prefs);
	INIT_LIST_HEAD(&prefs_delayed);

	key.objectid = bytenr;
	key.offset = (u64)-1;
1243 1244 1245 1246
	if (btrfs_fs_incompat(fs_info, SKINNY_METADATA))
		key.type = BTRFS_METADATA_ITEM_KEY;
	else
		key.type = BTRFS_EXTENT_ITEM_KEY;
1247 1248 1249 1250

	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;
1251
	if (!trans) {
1252
		path->search_commit_root = 1;
1253 1254
		path->skip_locking = 1;
	}
1255

1256
	if (time_seq == SEQ_LAST)
1257 1258
		path->skip_locking = 1;

1259 1260 1261 1262 1263 1264
	/*
	 * grab both a lock on the path and a lock on the delayed ref head.
	 * We need both to get a consistent picture of how the refs look
	 * at a specified point in time
	 */
again:
1265 1266
	head = NULL;

1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278
	if (check_shared) {
		if (!ref_tree) {
			ref_tree = ref_root_alloc();
			if (!ref_tree) {
				ret = -ENOMEM;
				goto out;
			}
		} else {
			ref_root_fini(ref_tree);
		}
	}

1279 1280 1281 1282 1283
	ret = btrfs_search_slot(trans, fs_info->extent_root, &key, path, 0, 0);
	if (ret < 0)
		goto out;
	BUG_ON(ret == 0);

1284
#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
1285
	if (trans && likely(trans->type != __TRANS_DUMMY) &&
1286
	    time_seq != SEQ_LAST) {
1287
#else
1288
	if (trans && time_seq != SEQ_LAST) {
1289
#endif
1290 1291 1292 1293 1294 1295
		/*
		 * look if there are updates for this ref queued and lock the
		 * head
		 */
		delayed_refs = &trans->transaction->delayed_refs;
		spin_lock(&delayed_refs->lock);
1296
		head = btrfs_find_delayed_ref_head(delayed_refs, bytenr);
1297 1298
		if (head) {
			if (!mutex_trylock(&head->mutex)) {
1299
				refcount_inc(&head->node.refs);
1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312
				spin_unlock(&delayed_refs->lock);

				btrfs_release_path(path);

				/*
				 * Mutex was contended, block until it's
				 * released and try again
				 */
				mutex_lock(&head->mutex);
				mutex_unlock(&head->mutex);
				btrfs_put_delayed_ref(&head->node);
				goto again;
			}
1313
			spin_unlock(&delayed_refs->lock);
1314
			ret = __add_delayed_refs(head, time_seq,
1315 1316
						 &prefs_delayed, &total_refs,
						 inum);
1317
			mutex_unlock(&head->mutex);
1318
			if (ret)
1319
				goto out;
1320 1321
		} else {
			spin_unlock(&delayed_refs->lock);
1322
		}
1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352

		if (check_shared && !list_empty(&prefs_delayed)) {
			/*
			 * Add all delay_ref to the ref_tree and check if there
			 * are multiple ref items added.
			 */
			list_for_each_entry(ref, &prefs_delayed, list) {
				if (ref->key_for_search.type) {
					ret = ref_tree_add(ref_tree,
						ref->root_id,
						ref->key_for_search.objectid,
						ref->key_for_search.offset,
						0, ref->count);
					if (ret)
						goto out;
				} else {
					ret = ref_tree_add(ref_tree, 0, 0, 0,
						     ref->parent, ref->count);
					if (ret)
						goto out;
				}

			}

			if (ref_tree->unique_refs > 1) {
				ret = BACKREF_FOUND_SHARED;
				goto out;
			}

		}
1353 1354 1355 1356 1357 1358
	}

	if (path->slots[0]) {
		struct extent_buffer *leaf;
		int slot;

1359
		path->slots[0]--;
1360
		leaf = path->nodes[0];
1361
		slot = path->slots[0];
1362 1363
		btrfs_item_key_to_cpu(leaf, &key, slot);
		if (key.objectid == bytenr &&
1364 1365
		    (key.type == BTRFS_EXTENT_ITEM_KEY ||
		     key.type == BTRFS_METADATA_ITEM_KEY)) {
1366
			ret = __add_inline_refs(path, bytenr,
1367
						&info_level, &prefs,
1368 1369
						ref_tree, &total_refs,
						inum);
1370 1371
			if (ret)
				goto out;
1372
			ret = __add_keyed_refs(fs_info, path, bytenr,
1373 1374
					       info_level, &prefs,
					       ref_tree, inum);
1375 1376 1377 1378 1379 1380 1381 1382
			if (ret)
				goto out;
		}
	}
	btrfs_release_path(path);

	list_splice_init(&prefs_delayed, &prefs);

1383 1384 1385 1386
	ret = __add_missing_keys(fs_info, &prefs);
	if (ret)
		goto out;

1387
	__merge_refs(&prefs, MERGE_IDENTICAL_KEYS);
1388

1389
	ret = __resolve_indirect_refs(fs_info, path, time_seq, &prefs,
1390 1391
				      extent_item_pos, total_refs,
				      root_objectid);
1392 1393 1394
	if (ret)
		goto out;

1395
	__merge_refs(&prefs, MERGE_IDENTICAL_PARENTS);
1396 1397 1398

	while (!list_empty(&prefs)) {
		ref = list_first_entry(&prefs, struct __prelim_ref, list);
J
Julia Lawall 已提交
1399
		WARN_ON(ref->count < 0);
1400
		if (roots && ref->count && ref->root_id && ref->parent == 0) {
1401 1402 1403 1404 1405
			if (root_objectid && ref->root_id != root_objectid) {
				ret = BACKREF_FOUND_SHARED;
				goto out;
			}

1406 1407
			/* no parent == root of tree */
			ret = ulist_add(roots, ref->root_id, 0, GFP_NOFS);
1408 1409
			if (ret < 0)
				goto out;
1410 1411
		}
		if (ref->count && ref->parent) {
1412 1413
			if (extent_item_pos && !ref->inode_list &&
			    ref->level == 0) {
1414
				struct extent_buffer *eb;
1415

1416
				eb = read_tree_block(fs_info, ref->parent, 0);
1417 1418 1419 1420
				if (IS_ERR(eb)) {
					ret = PTR_ERR(eb);
					goto out;
				} else if (!extent_buffer_uptodate(eb)) {
1421
					free_extent_buffer(eb);
1422 1423
					ret = -EIO;
					goto out;
1424
				}
1425 1426
				btrfs_tree_read_lock(eb);
				btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK);
1427 1428
				ret = find_extent_in_eb(eb, bytenr,
							*extent_item_pos, &eie);
1429
				btrfs_tree_read_unlock_blocking(eb);
1430
				free_extent_buffer(eb);
1431 1432 1433
				if (ret < 0)
					goto out;
				ref->inode_list = eie;
1434
			}
1435 1436 1437
			ret = ulist_add_merge_ptr(refs, ref->parent,
						  ref->inode_list,
						  (void **)&eie, GFP_NOFS);
1438 1439
			if (ret < 0)
				goto out;
1440 1441 1442 1443 1444 1445 1446 1447 1448 1449
			if (!ret && extent_item_pos) {
				/*
				 * we've recorded that parent, so we must extend
				 * its inode list here
				 */
				BUG_ON(!eie);
				while (eie->next)
					eie = eie->next;
				eie->next = ref->inode_list;
			}
1450
			eie = NULL;
1451
		}
1452
		list_del(&ref->list);
1453
		kmem_cache_free(btrfs_prelim_ref_cache, ref);
1454 1455 1456 1457
	}

out:
	btrfs_free_path(path);
1458
	ref_root_free(ref_tree);
1459 1460 1461
	while (!list_empty(&prefs)) {
		ref = list_first_entry(&prefs, struct __prelim_ref, list);
		list_del(&ref->list);
1462
		kmem_cache_free(btrfs_prelim_ref_cache, ref);
1463 1464 1465 1466 1467
	}
	while (!list_empty(&prefs_delayed)) {
		ref = list_first_entry(&prefs_delayed, struct __prelim_ref,
				       list);
		list_del(&ref->list);
1468
		kmem_cache_free(btrfs_prelim_ref_cache, ref);
1469
	}
1470 1471
	if (ret < 0)
		free_inode_elem_list(eie);
1472 1473 1474
	return ret;
}

1475 1476 1477 1478 1479 1480 1481 1482 1483 1484
static void free_leaf_list(struct ulist *blocks)
{
	struct ulist_node *node = NULL;
	struct extent_inode_elem *eie;
	struct ulist_iterator uiter;

	ULIST_ITER_INIT(&uiter);
	while ((node = ulist_next(blocks, &uiter))) {
		if (!node->aux)
			continue;
1485
		eie = unode_aux_to_inode_list(node);
1486
		free_inode_elem_list(eie);
1487 1488 1489 1490 1491 1492
		node->aux = 0;
	}

	ulist_free(blocks);
}

1493 1494 1495 1496 1497 1498 1499 1500 1501 1502
/*
 * Finds all leafs with a reference to the specified combination of bytenr and
 * offset. key_list_head will point to a list of corresponding keys (caller must
 * free each list element). The leafs will be stored in the leafs ulist, which
 * must be freed with ulist_free.
 *
 * returns 0 on success, <0 on error
 */
static int btrfs_find_all_leafs(struct btrfs_trans_handle *trans,
				struct btrfs_fs_info *fs_info, u64 bytenr,
1503
				u64 time_seq, struct ulist **leafs,
1504
				const u64 *extent_item_pos)
1505 1506 1507 1508
{
	int ret;

	*leafs = ulist_alloc(GFP_NOFS);
1509
	if (!*leafs)
1510 1511
		return -ENOMEM;

1512 1513
	ret = find_parent_nodes(trans, fs_info, bytenr, time_seq,
				*leafs, NULL, extent_item_pos, 0, 0, 0);
1514
	if (ret < 0 && ret != -ENOENT) {
1515
		free_leaf_list(*leafs);
1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534
		return ret;
	}

	return 0;
}

/*
 * walk all backrefs for a given extent to find all roots that reference this
 * extent. Walking a backref means finding all extents that reference this
 * extent and in turn walk the backrefs of those, too. Naturally this is a
 * recursive process, but here it is implemented in an iterative fashion: We
 * find all referencing extents for the extent in question and put them on a
 * list. In turn, we find all referencing extents for those, further appending
 * to the list. The way we iterate the list allows adding more elements after
 * the current while iterating. The process stops when we reach the end of the
 * list. Found roots are added to the roots list.
 *
 * returns 0 on success, < 0 on error.
 */
1535 1536 1537
static int __btrfs_find_all_roots(struct btrfs_trans_handle *trans,
				  struct btrfs_fs_info *fs_info, u64 bytenr,
				  u64 time_seq, struct ulist **roots)
1538 1539 1540
{
	struct ulist *tmp;
	struct ulist_node *node = NULL;
J
Jan Schmidt 已提交
1541
	struct ulist_iterator uiter;
1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552
	int ret;

	tmp = ulist_alloc(GFP_NOFS);
	if (!tmp)
		return -ENOMEM;
	*roots = ulist_alloc(GFP_NOFS);
	if (!*roots) {
		ulist_free(tmp);
		return -ENOMEM;
	}

J
Jan Schmidt 已提交
1553
	ULIST_ITER_INIT(&uiter);
1554
	while (1) {
1555 1556
		ret = find_parent_nodes(trans, fs_info, bytenr, time_seq,
					tmp, *roots, NULL, 0, 0, 0);
1557 1558 1559 1560 1561
		if (ret < 0 && ret != -ENOENT) {
			ulist_free(tmp);
			ulist_free(*roots);
			return ret;
		}
J
Jan Schmidt 已提交
1562
		node = ulist_next(tmp, &uiter);
1563 1564 1565
		if (!node)
			break;
		bytenr = node->val;
1566
		cond_resched();
1567 1568 1569 1570 1571 1572
	}

	ulist_free(tmp);
	return 0;
}

1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586
int btrfs_find_all_roots(struct btrfs_trans_handle *trans,
			 struct btrfs_fs_info *fs_info, u64 bytenr,
			 u64 time_seq, struct ulist **roots)
{
	int ret;

	if (!trans)
		down_read(&fs_info->commit_root_sem);
	ret = __btrfs_find_all_roots(trans, fs_info, bytenr, time_seq, roots);
	if (!trans)
		up_read(&fs_info->commit_root_sem);
	return ret;
}

1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599
/**
 * btrfs_check_shared - tell us whether an extent is shared
 *
 * @trans: optional trans handle
 *
 * btrfs_check_shared uses the backref walking code but will short
 * circuit as soon as it finds a root or inode that doesn't match the
 * one passed in. This provides a significant performance benefit for
 * callers (such as fiemap) which want to know whether the extent is
 * shared but do not need a ref count.
 *
 * Return: 0 if extent is not shared, 1 if it is shared, < 0 on error.
 */
1600 1601 1602 1603 1604 1605 1606 1607
int btrfs_check_shared(struct btrfs_trans_handle *trans,
		       struct btrfs_fs_info *fs_info, u64 root_objectid,
		       u64 inum, u64 bytenr)
{
	struct ulist *tmp = NULL;
	struct ulist *roots = NULL;
	struct ulist_iterator uiter;
	struct ulist_node *node;
1608
	struct seq_list elem = SEQ_LIST_INIT(elem);
1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625
	int ret = 0;

	tmp = ulist_alloc(GFP_NOFS);
	roots = ulist_alloc(GFP_NOFS);
	if (!tmp || !roots) {
		ulist_free(tmp);
		ulist_free(roots);
		return -ENOMEM;
	}

	if (trans)
		btrfs_get_tree_mod_seq(fs_info, &elem);
	else
		down_read(&fs_info->commit_root_sem);
	ULIST_ITER_INIT(&uiter);
	while (1) {
		ret = find_parent_nodes(trans, fs_info, bytenr, elem.seq, tmp,
1626
					roots, NULL, root_objectid, inum, 1);
1627
		if (ret == BACKREF_FOUND_SHARED) {
1628
			/* this is the only condition under which we return 1 */
1629 1630 1631 1632 1633
			ret = 1;
			break;
		}
		if (ret < 0 && ret != -ENOENT)
			break;
1634
		ret = 0;
1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649
		node = ulist_next(tmp, &uiter);
		if (!node)
			break;
		bytenr = node->val;
		cond_resched();
	}
	if (trans)
		btrfs_put_tree_mod_seq(fs_info, &elem);
	else
		up_read(&fs_info->commit_root_sem);
	ulist_free(tmp);
	ulist_free(roots);
	return ret;
}

M
Mark Fasheh 已提交
1650 1651 1652 1653 1654 1655 1656 1657 1658
int btrfs_find_one_extref(struct btrfs_root *root, u64 inode_objectid,
			  u64 start_off, struct btrfs_path *path,
			  struct btrfs_inode_extref **ret_extref,
			  u64 *found_off)
{
	int ret, slot;
	struct btrfs_key key;
	struct btrfs_key found_key;
	struct btrfs_inode_extref *extref;
1659
	const struct extent_buffer *leaf;
M
Mark Fasheh 已提交
1660 1661 1662
	unsigned long ptr;

	key.objectid = inode_objectid;
1663
	key.type = BTRFS_INODE_EXTREF_KEY;
M
Mark Fasheh 已提交
1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702
	key.offset = start_off;

	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
	if (ret < 0)
		return ret;

	while (1) {
		leaf = path->nodes[0];
		slot = path->slots[0];
		if (slot >= btrfs_header_nritems(leaf)) {
			/*
			 * If the item at offset is not found,
			 * btrfs_search_slot will point us to the slot
			 * where it should be inserted. In our case
			 * that will be the slot directly before the
			 * next INODE_REF_KEY_V2 item. In the case
			 * that we're pointing to the last slot in a
			 * leaf, we must move one leaf over.
			 */
			ret = btrfs_next_leaf(root, path);
			if (ret) {
				if (ret >= 1)
					ret = -ENOENT;
				break;
			}
			continue;
		}

		btrfs_item_key_to_cpu(leaf, &found_key, slot);

		/*
		 * Check that we're still looking at an extended ref key for
		 * this particular objectid. If we have different
		 * objectid or type then there are no more to be found
		 * in the tree and we can exit.
		 */
		ret = -ENOENT;
		if (found_key.objectid != inode_objectid)
			break;
1703
		if (found_key.type != BTRFS_INODE_EXTREF_KEY)
M
Mark Fasheh 已提交
1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717
			break;

		ret = 0;
		ptr = btrfs_item_ptr_offset(leaf, path->slots[0]);
		extref = (struct btrfs_inode_extref *)ptr;
		*ret_extref = extref;
		if (found_off)
			*found_off = found_key.offset;
		break;
	}

	return ret;
}

1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731
/*
 * this iterates to turn a name (from iref/extref) into a full filesystem path.
 * Elements of the path are separated by '/' and the path is guaranteed to be
 * 0-terminated. the path is only given within the current file system.
 * Therefore, it never starts with a '/'. the caller is responsible to provide
 * "size" bytes in "dest". the dest buffer will be filled backwards. finally,
 * the start point of the resulting string is returned. this pointer is within
 * dest, normally.
 * in case the path buffer would overflow, the pointer is decremented further
 * as if output was written to the buffer, though no more output is actually
 * generated. that way, the caller can determine how much space would be
 * required for the path to fit into the buffer. in that case, the returned
 * value will be smaller than dest. callers must check this!
 */
1732 1733 1734 1735
char *btrfs_ref_to_path(struct btrfs_root *fs_root, struct btrfs_path *path,
			u32 name_len, unsigned long name_off,
			struct extent_buffer *eb_in, u64 parent,
			char *dest, u32 size)
1736 1737 1738 1739
{
	int slot;
	u64 next_inum;
	int ret;
1740
	s64 bytes_left = ((s64)size) - 1;
1741 1742
	struct extent_buffer *eb = eb_in;
	struct btrfs_key found_key;
1743
	int leave_spinning = path->leave_spinning;
M
Mark Fasheh 已提交
1744
	struct btrfs_inode_ref *iref;
1745 1746 1747 1748

	if (bytes_left >= 0)
		dest[bytes_left] = '\0';

1749
	path->leave_spinning = 1;
1750
	while (1) {
M
Mark Fasheh 已提交
1751
		bytes_left -= name_len;
1752 1753
		if (bytes_left >= 0)
			read_extent_buffer(eb, dest + bytes_left,
M
Mark Fasheh 已提交
1754
					   name_off, name_len);
1755
		if (eb != eb_in) {
1756 1757
			if (!path->skip_locking)
				btrfs_tree_read_unlock_blocking(eb);
1758
			free_extent_buffer(eb);
1759
		}
1760 1761
		ret = btrfs_find_item(fs_root, path, parent, 0,
				BTRFS_INODE_REF_KEY, &found_key);
1762 1763
		if (ret > 0)
			ret = -ENOENT;
1764 1765
		if (ret)
			break;
M
Mark Fasheh 已提交
1766

1767 1768 1769 1770 1771 1772 1773 1774 1775
		next_inum = found_key.offset;

		/* regular exit ahead */
		if (parent == next_inum)
			break;

		slot = path->slots[0];
		eb = path->nodes[0];
		/* make sure we can use eb after releasing the path */
1776
		if (eb != eb_in) {
1777 1778 1779 1780
			if (!path->skip_locking)
				btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK);
			path->nodes[0] = NULL;
			path->locks[0] = 0;
1781
		}
1782 1783
		btrfs_release_path(path);
		iref = btrfs_item_ptr(eb, slot, struct btrfs_inode_ref);
M
Mark Fasheh 已提交
1784 1785 1786 1787

		name_len = btrfs_inode_ref_name_len(eb, iref);
		name_off = (unsigned long)(iref + 1);

1788 1789 1790 1791 1792 1793 1794
		parent = next_inum;
		--bytes_left;
		if (bytes_left >= 0)
			dest[bytes_left] = '/';
	}

	btrfs_release_path(path);
1795
	path->leave_spinning = leave_spinning;
1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808

	if (ret)
		return ERR_PTR(ret);

	return dest + bytes_left;
}

/*
 * this makes the path point to (logical EXTENT_ITEM *)
 * returns BTRFS_EXTENT_FLAG_DATA for data, BTRFS_EXTENT_FLAG_TREE_BLOCK for
 * tree blocks and <0 on error.
 */
int extent_from_logical(struct btrfs_fs_info *fs_info, u64 logical,
1809 1810
			struct btrfs_path *path, struct btrfs_key *found_key,
			u64 *flags_ret)
1811 1812 1813
{
	int ret;
	u64 flags;
1814
	u64 size = 0;
1815
	u32 item_size;
1816
	const struct extent_buffer *eb;
1817 1818 1819
	struct btrfs_extent_item *ei;
	struct btrfs_key key;

1820 1821 1822 1823
	if (btrfs_fs_incompat(fs_info, SKINNY_METADATA))
		key.type = BTRFS_METADATA_ITEM_KEY;
	else
		key.type = BTRFS_EXTENT_ITEM_KEY;
1824 1825 1826 1827 1828 1829 1830
	key.objectid = logical;
	key.offset = (u64)-1;

	ret = btrfs_search_slot(NULL, fs_info->extent_root, &key, path, 0, 0);
	if (ret < 0)
		return ret;

1831 1832 1833 1834 1835
	ret = btrfs_previous_extent_item(fs_info->extent_root, path, 0);
	if (ret) {
		if (ret > 0)
			ret = -ENOENT;
		return ret;
1836
	}
1837
	btrfs_item_key_to_cpu(path->nodes[0], found_key, path->slots[0]);
1838
	if (found_key->type == BTRFS_METADATA_ITEM_KEY)
1839
		size = fs_info->nodesize;
1840 1841 1842
	else if (found_key->type == BTRFS_EXTENT_ITEM_KEY)
		size = found_key->offset;

1843
	if (found_key->objectid > logical ||
1844
	    found_key->objectid + size <= logical) {
1845 1846
		btrfs_debug(fs_info,
			"logical %llu is not within any extent", logical);
1847
		return -ENOENT;
J
Jan Schmidt 已提交
1848
	}
1849 1850 1851 1852 1853 1854 1855 1856

	eb = path->nodes[0];
	item_size = btrfs_item_size_nr(eb, path->slots[0]);
	BUG_ON(item_size < sizeof(*ei));

	ei = btrfs_item_ptr(eb, path->slots[0], struct btrfs_extent_item);
	flags = btrfs_extent_flags(eb, ei);

1857 1858
	btrfs_debug(fs_info,
		"logical %llu is at position %llu within the extent (%llu EXTENT_ITEM %llu) flags %#llx size %u",
1859 1860
		 logical, logical - found_key->objectid, found_key->objectid,
		 found_key->offset, flags, item_size);
1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871

	WARN_ON(!flags_ret);
	if (flags_ret) {
		if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK)
			*flags_ret = BTRFS_EXTENT_FLAG_TREE_BLOCK;
		else if (flags & BTRFS_EXTENT_FLAG_DATA)
			*flags_ret = BTRFS_EXTENT_FLAG_DATA;
		else
			BUG_ON(1);
		return 0;
	}
1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883

	return -EIO;
}

/*
 * helper function to iterate extent inline refs. ptr must point to a 0 value
 * for the first call and may be modified. it is used to track state.
 * if more refs exist, 0 is returned and the next call to
 * __get_extent_inline_ref must pass the modified ptr parameter to get the
 * next ref. after the last ref was processed, 1 is returned.
 * returns <0 on error
 */
1884 1885 1886 1887 1888
static int __get_extent_inline_ref(unsigned long *ptr,
				   const struct extent_buffer *eb,
				   const struct btrfs_key *key,
				   const struct btrfs_extent_item *ei,
				   u32 item_size,
1889 1890
				   struct btrfs_extent_inline_ref **out_eiref,
				   int *out_type)
1891 1892 1893 1894 1895 1896 1897 1898 1899
{
	unsigned long end;
	u64 flags;
	struct btrfs_tree_block_info *info;

	if (!*ptr) {
		/* first call */
		flags = btrfs_extent_flags(eb, ei);
		if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
1900 1901 1902 1903 1904 1905 1906 1907 1908 1909
			if (key->type == BTRFS_METADATA_ITEM_KEY) {
				/* a skinny metadata extent */
				*out_eiref =
				     (struct btrfs_extent_inline_ref *)(ei + 1);
			} else {
				WARN_ON(key->type != BTRFS_EXTENT_ITEM_KEY);
				info = (struct btrfs_tree_block_info *)(ei + 1);
				*out_eiref =
				   (struct btrfs_extent_inline_ref *)(info + 1);
			}
1910 1911 1912 1913
		} else {
			*out_eiref = (struct btrfs_extent_inline_ref *)(ei + 1);
		}
		*ptr = (unsigned long)*out_eiref;
1914
		if ((unsigned long)(*ptr) >= (unsigned long)ei + item_size)
1915 1916 1917 1918
			return -ENOENT;
	}

	end = (unsigned long)ei + item_size;
1919
	*out_eiref = (struct btrfs_extent_inline_ref *)(*ptr);
1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937
	*out_type = btrfs_extent_inline_ref_type(eb, *out_eiref);

	*ptr += btrfs_extent_inline_ref_size(*out_type);
	WARN_ON(*ptr > end);
	if (*ptr == end)
		return 1; /* last */

	return 0;
}

/*
 * reads the tree block backref for an extent. tree level and root are returned
 * through out_level and out_root. ptr must point to a 0 value for the first
 * call and may be modified (see __get_extent_inline_ref comment).
 * returns 0 if data was provided, 1 if there was no more data to provide or
 * <0 on error.
 */
int tree_backref_for_extent(unsigned long *ptr, struct extent_buffer *eb,
1938 1939
			    struct btrfs_key *key, struct btrfs_extent_item *ei,
			    u32 item_size, u64 *out_root, u8 *out_level)
1940 1941 1942 1943 1944 1945 1946 1947 1948
{
	int ret;
	int type;
	struct btrfs_extent_inline_ref *eiref;

	if (*ptr == (unsigned long)-1)
		return 1;

	while (1) {
1949 1950
		ret = __get_extent_inline_ref(ptr, eb, key, ei, item_size,
					      &eiref, &type);
1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963
		if (ret < 0)
			return ret;

		if (type == BTRFS_TREE_BLOCK_REF_KEY ||
		    type == BTRFS_SHARED_BLOCK_REF_KEY)
			break;

		if (ret == 1)
			return 1;
	}

	/* we can treat both ref types equally here */
	*out_root = btrfs_extent_inline_ref_offset(eb, eiref);
1964 1965 1966 1967 1968 1969 1970 1971 1972 1973

	if (key->type == BTRFS_EXTENT_ITEM_KEY) {
		struct btrfs_tree_block_info *info;

		info = (struct btrfs_tree_block_info *)(ei + 1);
		*out_level = btrfs_tree_block_level(eb, info);
	} else {
		ASSERT(key->type == BTRFS_METADATA_ITEM_KEY);
		*out_level = (u8)key->offset;
	}
1974 1975 1976 1977 1978 1979 1980

	if (ret == 1)
		*ptr = (unsigned long)-1;

	return 0;
}

1981 1982 1983 1984
static int iterate_leaf_refs(struct btrfs_fs_info *fs_info,
			     struct extent_inode_elem *inode_list,
			     u64 root, u64 extent_item_objectid,
			     iterate_extent_inodes_t *iterate, void *ctx)
1985
{
1986
	struct extent_inode_elem *eie;
J
Jan Schmidt 已提交
1987 1988
	int ret = 0;

1989
	for (eie = inode_list; eie; eie = eie->next) {
1990 1991 1992 1993
		btrfs_debug(fs_info,
			    "ref for %llu resolved, key (%llu EXTEND_DATA %llu), root %llu",
			    extent_item_objectid, eie->inum,
			    eie->offset, root);
1994
		ret = iterate(eie->inum, eie->offset, root, ctx);
J
Jan Schmidt 已提交
1995
		if (ret) {
1996 1997 1998
			btrfs_debug(fs_info,
				    "stopping iteration for %llu due to ret=%d",
				    extent_item_objectid, ret);
J
Jan Schmidt 已提交
1999 2000
			break;
		}
2001 2002 2003 2004 2005 2006 2007
	}

	return ret;
}

/*
 * calls iterate() for every inode that references the extent identified by
J
Jan Schmidt 已提交
2008
 * the given parameters.
2009 2010 2011
 * when the iterator function returns a non-zero value, iteration stops.
 */
int iterate_extent_inodes(struct btrfs_fs_info *fs_info,
J
Jan Schmidt 已提交
2012
				u64 extent_item_objectid, u64 extent_item_pos,
2013
				int search_commit_root,
2014 2015 2016
				iterate_extent_inodes_t *iterate, void *ctx)
{
	int ret;
2017
	struct btrfs_trans_handle *trans = NULL;
2018 2019
	struct ulist *refs = NULL;
	struct ulist *roots = NULL;
J
Jan Schmidt 已提交
2020 2021
	struct ulist_node *ref_node = NULL;
	struct ulist_node *root_node = NULL;
2022
	struct seq_list tree_mod_seq_elem = SEQ_LIST_INIT(tree_mod_seq_elem);
J
Jan Schmidt 已提交
2023 2024
	struct ulist_iterator ref_uiter;
	struct ulist_iterator root_uiter;
2025

2026
	btrfs_debug(fs_info, "resolving all inodes for extent %llu",
J
Jan Schmidt 已提交
2027
			extent_item_objectid);
2028

2029
	if (!search_commit_root) {
2030 2031 2032
		trans = btrfs_join_transaction(fs_info->extent_root);
		if (IS_ERR(trans))
			return PTR_ERR(trans);
2033
		btrfs_get_tree_mod_seq(fs_info, &tree_mod_seq_elem);
2034 2035
	} else {
		down_read(&fs_info->commit_root_sem);
2036
	}
2037

J
Jan Schmidt 已提交
2038
	ret = btrfs_find_all_leafs(trans, fs_info, extent_item_objectid,
2039
				   tree_mod_seq_elem.seq, &refs,
2040
				   &extent_item_pos);
J
Jan Schmidt 已提交
2041 2042
	if (ret)
		goto out;
2043

J
Jan Schmidt 已提交
2044 2045
	ULIST_ITER_INIT(&ref_uiter);
	while (!ret && (ref_node = ulist_next(refs, &ref_uiter))) {
2046 2047
		ret = __btrfs_find_all_roots(trans, fs_info, ref_node->val,
					     tree_mod_seq_elem.seq, &roots);
J
Jan Schmidt 已提交
2048 2049
		if (ret)
			break;
J
Jan Schmidt 已提交
2050 2051
		ULIST_ITER_INIT(&root_uiter);
		while (!ret && (root_node = ulist_next(roots, &root_uiter))) {
2052 2053 2054 2055 2056 2057
			btrfs_debug(fs_info,
				    "root %llu references leaf %llu, data list %#llx",
				    root_node->val, ref_node->val,
				    ref_node->aux);
			ret = iterate_leaf_refs(fs_info,
						(struct extent_inode_elem *)
2058 2059 2060 2061
						(uintptr_t)ref_node->aux,
						root_node->val,
						extent_item_objectid,
						iterate, ctx);
J
Jan Schmidt 已提交
2062
		}
2063
		ulist_free(roots);
2064 2065
	}

2066
	free_leaf_list(refs);
J
Jan Schmidt 已提交
2067
out:
2068
	if (!search_commit_root) {
2069
		btrfs_put_tree_mod_seq(fs_info, &tree_mod_seq_elem);
2070
		btrfs_end_transaction(trans);
2071 2072
	} else {
		up_read(&fs_info->commit_root_sem);
2073 2074
	}

2075 2076 2077 2078 2079 2080 2081 2082
	return ret;
}

int iterate_inodes_from_logical(u64 logical, struct btrfs_fs_info *fs_info,
				struct btrfs_path *path,
				iterate_extent_inodes_t *iterate, void *ctx)
{
	int ret;
J
Jan Schmidt 已提交
2083
	u64 extent_item_pos;
2084
	u64 flags = 0;
2085
	struct btrfs_key found_key;
2086
	int search_commit_root = path->search_commit_root;
2087

2088
	ret = extent_from_logical(fs_info, logical, path, &found_key, &flags);
J
Jan Schmidt 已提交
2089
	btrfs_release_path(path);
2090 2091
	if (ret < 0)
		return ret;
2092
	if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK)
2093
		return -EINVAL;
2094

J
Jan Schmidt 已提交
2095
	extent_item_pos = logical - found_key.objectid;
2096 2097 2098
	ret = iterate_extent_inodes(fs_info, found_key.objectid,
					extent_item_pos, search_commit_root,
					iterate, ctx);
2099 2100 2101 2102

	return ret;
}

M
Mark Fasheh 已提交
2103 2104 2105 2106 2107 2108
typedef int (iterate_irefs_t)(u64 parent, u32 name_len, unsigned long name_off,
			      struct extent_buffer *eb, void *ctx);

static int iterate_inode_refs(u64 inum, struct btrfs_root *fs_root,
			      struct btrfs_path *path,
			      iterate_irefs_t *iterate, void *ctx)
2109
{
2110
	int ret = 0;
2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121
	int slot;
	u32 cur;
	u32 len;
	u32 name_len;
	u64 parent = 0;
	int found = 0;
	struct extent_buffer *eb;
	struct btrfs_item *item;
	struct btrfs_inode_ref *iref;
	struct btrfs_key found_key;

2122
	while (!ret) {
2123 2124 2125 2126
		ret = btrfs_find_item(fs_root, path, inum,
				parent ? parent + 1 : 0, BTRFS_INODE_REF_KEY,
				&found_key);

2127 2128 2129 2130 2131 2132 2133 2134 2135 2136
		if (ret < 0)
			break;
		if (ret) {
			ret = found ? 0 : -ENOENT;
			break;
		}
		++found;

		parent = found_key.offset;
		slot = path->slots[0];
2137 2138 2139 2140 2141 2142
		eb = btrfs_clone_extent_buffer(path->nodes[0]);
		if (!eb) {
			ret = -ENOMEM;
			break;
		}
		extent_buffer_get(eb);
2143 2144
		btrfs_tree_read_lock(eb);
		btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK);
2145 2146
		btrfs_release_path(path);

2147
		item = btrfs_item_nr(slot);
2148 2149 2150 2151 2152
		iref = btrfs_item_ptr(eb, slot, struct btrfs_inode_ref);

		for (cur = 0; cur < btrfs_item_size(eb, item); cur += len) {
			name_len = btrfs_inode_ref_name_len(eb, iref);
			/* path must be released before calling iterate()! */
2153 2154 2155
			btrfs_debug(fs_root->fs_info,
				"following ref at offset %u for inode %llu in tree %llu",
				cur, found_key.objectid, fs_root->objectid);
M
Mark Fasheh 已提交
2156 2157
			ret = iterate(parent, name_len,
				      (unsigned long)(iref + 1), eb, ctx);
2158
			if (ret)
2159 2160 2161 2162
				break;
			len = sizeof(*iref) + name_len;
			iref = (struct btrfs_inode_ref *)((char *)iref + len);
		}
2163
		btrfs_tree_read_unlock_blocking(eb);
2164 2165 2166 2167 2168 2169 2170 2171
		free_extent_buffer(eb);
	}

	btrfs_release_path(path);

	return ret;
}

M
Mark Fasheh 已提交
2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198
static int iterate_inode_extrefs(u64 inum, struct btrfs_root *fs_root,
				 struct btrfs_path *path,
				 iterate_irefs_t *iterate, void *ctx)
{
	int ret;
	int slot;
	u64 offset = 0;
	u64 parent;
	int found = 0;
	struct extent_buffer *eb;
	struct btrfs_inode_extref *extref;
	u32 item_size;
	u32 cur_offset;
	unsigned long ptr;

	while (1) {
		ret = btrfs_find_one_extref(fs_root, inum, offset, path, &extref,
					    &offset);
		if (ret < 0)
			break;
		if (ret) {
			ret = found ? 0 : -ENOENT;
			break;
		}
		++found;

		slot = path->slots[0];
2199 2200 2201 2202 2203 2204
		eb = btrfs_clone_extent_buffer(path->nodes[0]);
		if (!eb) {
			ret = -ENOMEM;
			break;
		}
		extent_buffer_get(eb);
M
Mark Fasheh 已提交
2205 2206 2207 2208 2209

		btrfs_tree_read_lock(eb);
		btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK);
		btrfs_release_path(path);

2210 2211
		item_size = btrfs_item_size_nr(eb, slot);
		ptr = btrfs_item_ptr_offset(eb, slot);
M
Mark Fasheh 已提交
2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224
		cur_offset = 0;

		while (cur_offset < item_size) {
			u32 name_len;

			extref = (struct btrfs_inode_extref *)(ptr + cur_offset);
			parent = btrfs_inode_extref_parent(eb, extref);
			name_len = btrfs_inode_extref_name_len(eb, extref);
			ret = iterate(parent, name_len,
				      (unsigned long)&extref->name, eb, ctx);
			if (ret)
				break;

2225
			cur_offset += btrfs_inode_extref_name_len(eb, extref);
M
Mark Fasheh 已提交
2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258
			cur_offset += sizeof(*extref);
		}
		btrfs_tree_read_unlock_blocking(eb);
		free_extent_buffer(eb);

		offset++;
	}

	btrfs_release_path(path);

	return ret;
}

static int iterate_irefs(u64 inum, struct btrfs_root *fs_root,
			 struct btrfs_path *path, iterate_irefs_t *iterate,
			 void *ctx)
{
	int ret;
	int found_refs = 0;

	ret = iterate_inode_refs(inum, fs_root, path, iterate, ctx);
	if (!ret)
		++found_refs;
	else if (ret != -ENOENT)
		return ret;

	ret = iterate_inode_extrefs(inum, fs_root, path, iterate, ctx);
	if (ret == -ENOENT && found_refs)
		return 0;

	return ret;
}

2259 2260 2261 2262
/*
 * returns 0 if the path could be dumped (probably truncated)
 * returns <0 in case of an error
 */
M
Mark Fasheh 已提交
2263 2264
static int inode_to_path(u64 inum, u32 name_len, unsigned long name_off,
			 struct extent_buffer *eb, void *ctx)
2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275
{
	struct inode_fs_paths *ipath = ctx;
	char *fspath;
	char *fspath_min;
	int i = ipath->fspath->elem_cnt;
	const int s_ptr = sizeof(char *);
	u32 bytes_left;

	bytes_left = ipath->fspath->bytes_left > s_ptr ?
					ipath->fspath->bytes_left - s_ptr : 0;

2276
	fspath_min = (char *)ipath->fspath->val + (i + 1) * s_ptr;
2277 2278
	fspath = btrfs_ref_to_path(ipath->fs_root, ipath->btrfs_path, name_len,
				   name_off, eb, inum, fspath_min, bytes_left);
2279 2280 2281 2282
	if (IS_ERR(fspath))
		return PTR_ERR(fspath);

	if (fspath > fspath_min) {
2283
		ipath->fspath->val[i] = (u64)(unsigned long)fspath;
2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297
		++ipath->fspath->elem_cnt;
		ipath->fspath->bytes_left = fspath - fspath_min;
	} else {
		++ipath->fspath->elem_missed;
		ipath->fspath->bytes_missing += fspath_min - fspath;
		ipath->fspath->bytes_left = 0;
	}

	return 0;
}

/*
 * this dumps all file system paths to the inode into the ipath struct, provided
 * is has been created large enough. each path is zero-terminated and accessed
2298
 * from ipath->fspath->val[i].
2299
 * when it returns, there are ipath->fspath->elem_cnt number of paths available
2300
 * in ipath->fspath->val[]. when the allocated space wasn't sufficient, the
2301
 * number of missed paths is recorded in ipath->fspath->elem_missed, otherwise,
2302 2303 2304 2305 2306 2307
 * it's zero. ipath->fspath->bytes_missing holds the number of bytes that would
 * have been needed to return all paths.
 */
int paths_from_inode(u64 inum, struct inode_fs_paths *ipath)
{
	return iterate_irefs(inum, ipath->fs_root, ipath->btrfs_path,
M
Mark Fasheh 已提交
2308
			     inode_to_path, ipath);
2309 2310 2311 2312 2313 2314 2315 2316
}

struct btrfs_data_container *init_data_container(u32 total_bytes)
{
	struct btrfs_data_container *data;
	size_t alloc_bytes;

	alloc_bytes = max_t(size_t, total_bytes, sizeof(*data));
2317
	data = kvmalloc(alloc_bytes, GFP_KERNEL);
2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350
	if (!data)
		return ERR_PTR(-ENOMEM);

	if (total_bytes >= sizeof(*data)) {
		data->bytes_left = total_bytes - sizeof(*data);
		data->bytes_missing = 0;
	} else {
		data->bytes_missing = sizeof(*data) - total_bytes;
		data->bytes_left = 0;
	}

	data->elem_cnt = 0;
	data->elem_missed = 0;

	return data;
}

/*
 * allocates space to return multiple file system paths for an inode.
 * total_bytes to allocate are passed, note that space usable for actual path
 * information will be total_bytes - sizeof(struct inode_fs_paths).
 * the returned pointer must be freed with free_ipath() in the end.
 */
struct inode_fs_paths *init_ipath(s32 total_bytes, struct btrfs_root *fs_root,
					struct btrfs_path *path)
{
	struct inode_fs_paths *ifp;
	struct btrfs_data_container *fspath;

	fspath = init_data_container(total_bytes);
	if (IS_ERR(fspath))
		return (void *)fspath;

2351
	ifp = kmalloc(sizeof(*ifp), GFP_KERNEL);
2352
	if (!ifp) {
2353
		kvfree(fspath);
2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365
		return ERR_PTR(-ENOMEM);
	}

	ifp->btrfs_path = path;
	ifp->fspath = fspath;
	ifp->fs_root = fs_root;

	return ifp;
}

void free_ipath(struct inode_fs_paths *ipath)
{
2366 2367
	if (!ipath)
		return;
2368
	kvfree(ipath->fspath);
2369 2370
	kfree(ipath);
}