backref.c 52.0 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/vmalloc.h>
20 21 22
#include "ctree.h"
#include "disk-io.h"
#include "backref.h"
23 24 25
#include "ulist.h"
#include "transaction.h"
#include "delayed-ref.h"
26
#include "locking.h"
27

28 29 30
/* Just an arbitrary number so we can be sure this happened */
#define BACKREF_FOUND_SHARED 6

31 32 33 34 35 36 37 38 39 40 41
struct extent_inode_elem {
	u64 inum;
	u64 offset;
	struct extent_inode_elem *next;
};

static int check_extent_in_eb(struct btrfs_key *key, struct extent_buffer *eb,
				struct btrfs_file_extent_item *fi,
				u64 extent_item_pos,
				struct extent_inode_elem **eie)
{
42
	u64 offset = 0;
43 44
	struct extent_inode_elem *e;

45 46 47 48 49
	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;
50

51 52 53 54 55 56 57 58
		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;
	}
59 60 61 62 63 64 65

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

	e->next = *eie;
	e->inum = key->objectid;
66
	e->offset = key->offset + offset;
67 68 69 70 71
	*eie = e;

	return 0;
}

72 73 74 75 76 77 78 79 80 81
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);
	}
}

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
static int find_extent_in_eb(struct extent_buffer *eb, u64 wanted_disk_byte,
				u64 extent_item_pos,
				struct extent_inode_elem **eie)
{
	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;
}

121 122 123 124 125 126
/*
 * this structure records all encountered refs on the way up to the root
 */
struct __prelim_ref {
	struct list_head list;
	u64 root_id;
127
	struct btrfs_key key_for_search;
128 129
	int level;
	int count;
130
	struct extent_inode_elem *inode_list;
131 132 133 134
	u64 parent;
	u64 wanted_disk_byte;
};

135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150
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,
					SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD,
					NULL);
	if (!btrfs_prelim_ref_cache)
		return -ENOMEM;
	return 0;
}

void btrfs_prelim_ref_exit(void)
{
151
	kmem_cache_destroy(btrfs_prelim_ref_cache);
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
/*
 * 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.
 */

193
static int __add_prelim_ref(struct list_head *head, u64 root_id,
194
			    struct btrfs_key *key, int level,
195 196
			    u64 parent, u64 wanted_disk_byte, int count,
			    gfp_t gfp_mask)
197 198 199
{
	struct __prelim_ref *ref;

200 201 202
	if (root_id == BTRFS_DATA_RELOC_TREE_OBJECTID)
		return 0;

203
	ref = kmem_cache_alloc(btrfs_prelim_ref_cache, gfp_mask);
204 205 206 207
	if (!ref)
		return -ENOMEM;

	ref->root_id = root_id;
208
	if (key) {
209
		ref->key_for_search = *key;
210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232
		/*
		 * 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 {
233
		memset(&ref->key_for_search, 0, sizeof(ref->key_for_search));
234
	}
235

236
	ref->inode_list = NULL;
237 238 239 240 241 242 243 244 245 246
	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,
247
			   struct ulist *parents, struct __prelim_ref *ref,
248 249
			   int level, u64 time_seq, const u64 *extent_item_pos,
			   u64 total_refs)
250
{
251 252 253 254
	int ret = 0;
	int slot;
	struct extent_buffer *eb;
	struct btrfs_key key;
255
	struct btrfs_key *key_for_search = &ref->key_for_search;
256
	struct btrfs_file_extent_item *fi;
257
	struct extent_inode_elem *eie = NULL, *old = NULL;
258
	u64 disk_byte;
259 260
	u64 wanted_disk_byte = ref->wanted_disk_byte;
	u64 count = 0;
261

262 263 264
	if (level != 0) {
		eb = path->nodes[level];
		ret = ulist_add(parents, eb->start, 0, GFP_NOFS);
265 266
		if (ret < 0)
			return ret;
267
		return 0;
268
	}
269 270

	/*
271 272 273
	 * 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.
274
	 */
275 276 277 278 279 280
	if (path->slots[0] >= btrfs_header_nritems(path->nodes[0])) {
		if (time_seq == (u64)-1)
			ret = btrfs_next_leaf(root, path);
		else
			ret = btrfs_next_old_leaf(root, path, time_seq);
	}
281

282
	while (!ret && count < total_refs) {
283
		eb = path->nodes[0];
284 285 286 287 288 289 290 291 292 293 294 295 296
		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;
297
			old = NULL;
298
			count++;
299 300 301 302 303 304 305
			if (extent_item_pos) {
				ret = check_extent_in_eb(&key, eb, fi,
						*extent_item_pos,
						&eie);
				if (ret < 0)
					break;
			}
306 307
			if (ret > 0)
				goto next;
308 309
			ret = ulist_add_merge_ptr(parents, eb->start,
						  eie, (void **)&old, GFP_NOFS);
310 311 312 313 314 315
			if (ret < 0)
				break;
			if (!ret && extent_item_pos) {
				while (old->next)
					old = old->next;
				old->next = eie;
316
			}
317
			eie = NULL;
318
		}
319
next:
320 321 322 323
		if (time_seq == (u64)-1)
			ret = btrfs_next_item(root, path);
		else
			ret = btrfs_next_old_item(root, path, time_seq);
324 325
	}

326 327
	if (ret > 0)
		ret = 0;
328 329
	else if (ret < 0)
		free_inode_elem_list(eie);
330
	return ret;
331 332 333 334 335 336 337
}

/*
 * 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,
338 339 340
				  struct btrfs_path *path, u64 time_seq,
				  struct __prelim_ref *ref,
				  struct ulist *parents,
341
				  const u64 *extent_item_pos, u64 total_refs)
342 343 344 345 346 347 348
{
	struct btrfs_root *root;
	struct btrfs_key root_key;
	struct extent_buffer *eb;
	int ret = 0;
	int root_level;
	int level = ref->level;
349
	int index;
350 351 352 353

	root_key.objectid = ref->root_id;
	root_key.type = BTRFS_ROOT_ITEM_KEY;
	root_key.offset = (u64)-1;
354 355 356

	index = srcu_read_lock(&fs_info->subvol_srcu);

357
	root = btrfs_get_fs_root(fs_info, &root_key, false);
358
	if (IS_ERR(root)) {
359
		srcu_read_unlock(&fs_info->subvol_srcu, index);
360 361 362 363
		ret = PTR_ERR(root);
		goto out;
	}

J
Josef Bacik 已提交
364 365 366 367 368 369
	if (btrfs_test_is_dummy_root(root)) {
		srcu_read_unlock(&fs_info->subvol_srcu, index);
		ret = -ENOENT;
		goto out;
	}

370 371
	if (path->search_commit_root)
		root_level = btrfs_header_level(root->commit_root);
372 373
	else if (time_seq == (u64)-1)
		root_level = btrfs_header_level(root->node);
374 375
	else
		root_level = btrfs_old_root_level(root, time_seq);
376

377 378
	if (root_level + 1 == level) {
		srcu_read_unlock(&fs_info->subvol_srcu, index);
379
		goto out;
380
	}
381 382

	path->lowest_level = level;
383 384 385 386 387 388
	if (time_seq == (u64)-1)
		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);
389 390 391 392

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

393 394
	pr_debug("search slot in root %llu (level %d, ref count %d) returned "
		 "%d for key (%llu %u %llu)\n",
395 396 397
		 ref->root_id, level, ref->count, ret,
		 ref->key_for_search.objectid, ref->key_for_search.type,
		 ref->key_for_search.offset);
398 399 400 401
	if (ret < 0)
		goto out;

	eb = path->nodes[level];
402
	while (!eb) {
403
		if (WARN_ON(!level)) {
404 405 406 407 408
			ret = 1;
			goto out;
		}
		level--;
		eb = path->nodes[level];
409 410
	}

411
	ret = add_all_parents(root, path, parents, ref, level, time_seq,
412
			      extent_item_pos, total_refs);
413
out:
414 415
	path->lowest_level = 0;
	btrfs_release_path(path);
416 417 418 419 420 421 422
	return ret;
}

/*
 * resolve all indirect backrefs from the list
 */
static int __resolve_indirect_refs(struct btrfs_fs_info *fs_info,
423
				   struct btrfs_path *path, u64 time_seq,
424
				   struct list_head *head,
425 426
				   const u64 *extent_item_pos, u64 total_refs,
				   u64 root_objectid)
427 428 429 430 431 432 433 434
{
	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
435
	struct ulist_iterator uiter;
436 437 438 439 440 441 442 443 444 445 446 447 448 449 450

	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;
451 452 453 454
		if (root_objectid && ref->root_id != root_objectid) {
			ret = BACKREF_FOUND_SHARED;
			goto out;
		}
455
		err = __resolve_indirect_ref(fs_info, path, time_seq, ref,
456 457
					     parents, extent_item_pos,
					     total_refs);
458 459 460 461 462
		/*
		 * we can only tolerate ENOENT,otherwise,we should catch error
		 * and return directly.
		 */
		if (err == -ENOENT) {
463
			continue;
464 465 466 467
		} else if (err) {
			ret = err;
			goto out;
		}
468 469

		/* we put the first parent into the ref at hand */
J
Jan Schmidt 已提交
470 471
		ULIST_ITER_INIT(&uiter);
		node = ulist_next(parents, &uiter);
472
		ref->parent = node ? node->val : 0;
473
		ref->inode_list = node ?
474
			(struct extent_inode_elem *)(uintptr_t)node->aux : NULL;
475 476

		/* additional parents require new refs being added here */
J
Jan Schmidt 已提交
477
		while ((node = ulist_next(parents, &uiter))) {
478 479
			new_ref = kmem_cache_alloc(btrfs_prelim_ref_cache,
						   GFP_NOFS);
480 481
			if (!new_ref) {
				ret = -ENOMEM;
482
				goto out;
483 484 485
			}
			memcpy(new_ref, ref, sizeof(*ref));
			new_ref->parent = node->val;
486 487
			new_ref->inode_list = (struct extent_inode_elem *)
							(uintptr_t)node->aux;
488 489 490 491
			list_add(&new_ref->list, &ref->list);
		}
		ulist_reinit(parents);
	}
492
out:
493 494 495 496
	ulist_free(parents);
	return ret;
}

497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521
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)
{
522
	struct __prelim_ref *ref;
523 524
	struct extent_buffer *eb;

525
	list_for_each_entry(ref, head, list) {
526 527 528 529 530 531
		if (ref->parent)
			continue;
		if (ref->key_for_search.type)
			continue;
		BUG_ON(!ref->wanted_disk_byte);
		eb = read_tree_block(fs_info->tree_root, ref->wanted_disk_byte,
532
				     0);
533 534 535
		if (IS_ERR(eb)) {
			return PTR_ERR(eb);
		} else if (!extent_buffer_uptodate(eb)) {
536 537 538
			free_extent_buffer(eb);
			return -EIO;
		}
539 540 541 542 543 544 545 546 547 548 549
		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;
}

550
/*
551
 * merge backrefs and adjust counts accordingly
552 553
 *
 * mode = 1: merge identical keys, if key is set
554 555 556 557
 *    FIXME: if we add more keys in __add_prelim_ref, 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).
558 559
 * mode = 2: merge identical parents
 */
560
static void __merge_refs(struct list_head *head, int mode)
561
{
562
	struct __prelim_ref *pos1;
563

564 565
	list_for_each_entry(pos1, head, list) {
		struct __prelim_ref *pos2 = pos1, *tmp;
566

567
		list_for_each_entry_safe_continue(pos2, tmp, head, list) {
568
			struct __prelim_ref *ref1 = pos1, *ref2 = pos2;
569
			struct extent_inode_elem *eie;
570

571 572
			if (!ref_for_same_block(ref1, ref2))
				continue;
573
			if (mode == 1) {
574 575
				if (!ref1->parent && ref2->parent)
					swap(ref1, ref2);
576 577 578 579
			} else {
				if (ref1->parent != ref2->parent)
					continue;
			}
580 581 582 583 584 585 586 587 588 589

			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;

590
			list_del(&ref2->list);
591
			kmem_cache_free(btrfs_prelim_ref_cache, ref2);
592 593 594 595 596 597 598 599 600 601
		}

	}
}

/*
 * 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,
602 603
			      struct list_head *prefs, u64 *total_refs,
			      u64 inum)
604
{
605
	struct btrfs_delayed_ref_node *node;
606
	struct btrfs_delayed_extent_op *extent_op = head->extent_op;
607 608
	struct btrfs_key key;
	struct btrfs_key op_key = {0};
609
	int sgn;
610
	int ret = 0;
611 612

	if (extent_op && extent_op->update_key)
613
		btrfs_disk_key_to_cpu(&op_key, &extent_op->key);
614

615
	spin_lock(&head->lock);
616
	list_for_each_entry(node, &head->ref_list, list) {
617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633
		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);
		}
634
		*total_refs += (node->ref_mod * sgn);
635 636 637 638 639
		switch (node->type) {
		case BTRFS_TREE_BLOCK_REF_KEY: {
			struct btrfs_delayed_tree_ref *ref;

			ref = btrfs_delayed_node_to_tree_ref(node);
640
			ret = __add_prelim_ref(prefs, ref->root, &op_key,
641
					       ref->level + 1, 0, node->bytenr,
642
					       node->ref_mod * sgn, GFP_ATOMIC);
643 644 645 646 647 648
			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
649
			ret = __add_prelim_ref(prefs, 0, NULL,
650 651
					       ref->level + 1, ref->parent,
					       node->bytenr,
652
					       node->ref_mod * sgn, GFP_ATOMIC);
653 654 655 656 657 658 659 660 661
			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;
662 663 664 665 666 667 668 669 670 671

			/*
			 * 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;
			}

672 673
			ret = __add_prelim_ref(prefs, ref->root, &key, 0, 0,
					       node->bytenr,
674
					       node->ref_mod * sgn, GFP_ATOMIC);
675 676 677 678 679 680
			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
681
			ret = __add_prelim_ref(prefs, 0, NULL, 0,
682
					       ref->parent, node->bytenr,
683
					       node->ref_mod * sgn, GFP_ATOMIC);
684 685 686 687 688
			break;
		}
		default:
			WARN_ON(1);
		}
689
		if (ret)
690
			break;
691
	}
692 693
	spin_unlock(&head->lock);
	return ret;
694 695 696 697 698 699 700
}

/*
 * add all inline backrefs for bytenr to the list
 */
static int __add_inline_refs(struct btrfs_fs_info *fs_info,
			     struct btrfs_path *path, u64 bytenr,
701
			     int *info_level, struct list_head *prefs,
702
			     u64 *total_refs, u64 inum)
703
{
704
	int ret = 0;
705 706 707
	int slot;
	struct extent_buffer *leaf;
	struct btrfs_key key;
708
	struct btrfs_key found_key;
709 710 711 712 713 714 715 716 717 718
	unsigned long ptr;
	unsigned long end;
	struct btrfs_extent_item *ei;
	u64 flags;
	u64 item_size;

	/*
	 * enumerate all inline refs
	 */
	leaf = path->nodes[0];
719
	slot = path->slots[0];
720 721 722 723 724 725

	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);
726
	*total_refs += btrfs_extent_refs(leaf, ei);
727
	btrfs_item_key_to_cpu(leaf, &found_key, slot);
728 729 730 731

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

732 733
	if (found_key.type == BTRFS_EXTENT_ITEM_KEY &&
	    flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
734 735 736 737 738 739
		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);
740 741
	} else if (found_key.type == BTRFS_METADATA_ITEM_KEY) {
		*info_level = found_key.offset;
742 743 744 745 746 747 748 749 750 751 752 753 754 755 756
	} 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:
757
			ret = __add_prelim_ref(prefs, 0, NULL,
758
						*info_level + 1, offset,
759
						bytenr, 1, GFP_NOFS);
760 761 762 763 764 765 766 767
			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,
768
					       bytenr, count, GFP_NOFS);
769 770 771
			break;
		}
		case BTRFS_TREE_BLOCK_REF_KEY:
772 773
			ret = __add_prelim_ref(prefs, offset, NULL,
					       *info_level + 1, 0,
774
					       bytenr, 1, GFP_NOFS);
775 776 777 778 779 780 781 782 783 784 785 786
			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);
787 788 789 790 791 792

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

793
			root = btrfs_extent_data_ref_root(leaf, dref);
794
			ret = __add_prelim_ref(prefs, root, &key, 0, 0,
795
					       bytenr, count, GFP_NOFS);
796 797 798 799 800
			break;
		}
		default:
			WARN_ON(1);
		}
801 802
		if (ret)
			return ret;
803 804 805 806 807 808 809 810 811 812 813
		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,
814
			    int info_level, struct list_head *prefs, u64 inum)
815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843
{
	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:
844
			ret = __add_prelim_ref(prefs, 0, NULL,
845
						info_level + 1, key.offset,
846
						bytenr, 1, GFP_NOFS);
847 848 849 850 851 852 853 854 855
			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,
856
						bytenr, count, GFP_NOFS);
857 858 859
			break;
		}
		case BTRFS_TREE_BLOCK_REF_KEY:
860 861
			ret = __add_prelim_ref(prefs, key.offset, NULL,
					       info_level + 1, 0,
862
					       bytenr, 1, GFP_NOFS);
863 864 865 866 867 868 869 870 871 872 873 874 875
			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);
876 877 878 879 880 881

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

882 883
			root = btrfs_extent_data_ref_root(leaf, dref);
			ret = __add_prelim_ref(prefs, root, &key, 0, 0,
884
					       bytenr, count, GFP_NOFS);
885 886 887 888 889
			break;
		}
		default:
			WARN_ON(1);
		}
890 891 892
		if (ret)
			return ret;

893 894 895 896 897 898 899 900 901 902 903
	}

	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
 *
904 905
 * NOTE: This can return values > 0
 *
906 907 908 909 910
 * If time_seq is set to (u64)-1, it will not search delayed_refs, and behave
 * 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().
 *
911 912 913 914
 * 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,
915
			     u64 time_seq, struct ulist *refs,
916 917
			     struct ulist *roots, const u64 *extent_item_pos,
			     u64 root_objectid, u64 inum)
918 919 920 921
{
	struct btrfs_key key;
	struct btrfs_path *path;
	struct btrfs_delayed_ref_root *delayed_refs = NULL;
922
	struct btrfs_delayed_ref_head *head;
923 924 925 926 927
	int info_level = 0;
	int ret;
	struct list_head prefs_delayed;
	struct list_head prefs;
	struct __prelim_ref *ref;
928
	struct extent_inode_elem *eie = NULL;
929
	u64 total_refs = 0;
930 931 932 933 934 935

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

	key.objectid = bytenr;
	key.offset = (u64)-1;
936 937 938 939
	if (btrfs_fs_incompat(fs_info, SKINNY_METADATA))
		key.type = BTRFS_METADATA_ITEM_KEY;
	else
		key.type = BTRFS_EXTENT_ITEM_KEY;
940 941 942 943

	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;
944
	if (!trans) {
945
		path->search_commit_root = 1;
946 947
		path->skip_locking = 1;
	}
948

949 950 951
	if (time_seq == (u64)-1)
		path->skip_locking = 1;

952 953 954 955 956 957
	/*
	 * 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:
958 959
	head = NULL;

960 961 962 963 964
	ret = btrfs_search_slot(trans, fs_info->extent_root, &key, path, 0, 0);
	if (ret < 0)
		goto out;
	BUG_ON(ret == 0);

965
#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
966 967
	if (trans && likely(trans->type != __TRANS_DUMMY) &&
	    time_seq != (u64)-1) {
968
#else
969
	if (trans && time_seq != (u64)-1) {
970
#endif
971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993
		/*
		 * look if there are updates for this ref queued and lock the
		 * head
		 */
		delayed_refs = &trans->transaction->delayed_refs;
		spin_lock(&delayed_refs->lock);
		head = btrfs_find_delayed_ref_head(trans, bytenr);
		if (head) {
			if (!mutex_trylock(&head->mutex)) {
				atomic_inc(&head->node.refs);
				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;
			}
994
			spin_unlock(&delayed_refs->lock);
995
			ret = __add_delayed_refs(head, time_seq,
996 997
						 &prefs_delayed, &total_refs,
						 inum);
998
			mutex_unlock(&head->mutex);
999
			if (ret)
1000
				goto out;
1001 1002
		} else {
			spin_unlock(&delayed_refs->lock);
1003
		}
1004 1005 1006 1007 1008 1009
	}

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

1010
		path->slots[0]--;
1011
		leaf = path->nodes[0];
1012
		slot = path->slots[0];
1013 1014
		btrfs_item_key_to_cpu(leaf, &key, slot);
		if (key.objectid == bytenr &&
1015 1016
		    (key.type == BTRFS_EXTENT_ITEM_KEY ||
		     key.type == BTRFS_METADATA_ITEM_KEY)) {
1017
			ret = __add_inline_refs(fs_info, path, bytenr,
1018
						&info_level, &prefs,
1019
						&total_refs, inum);
1020 1021
			if (ret)
				goto out;
1022
			ret = __add_keyed_refs(fs_info, path, bytenr,
1023
					       info_level, &prefs, inum);
1024 1025 1026 1027 1028 1029 1030 1031
			if (ret)
				goto out;
		}
	}
	btrfs_release_path(path);

	list_splice_init(&prefs_delayed, &prefs);

1032 1033 1034 1035
	ret = __add_missing_keys(fs_info, &prefs);
	if (ret)
		goto out;

1036
	__merge_refs(&prefs, 1);
1037

1038
	ret = __resolve_indirect_refs(fs_info, path, time_seq, &prefs,
1039 1040
				      extent_item_pos, total_refs,
				      root_objectid);
1041 1042 1043
	if (ret)
		goto out;

1044
	__merge_refs(&prefs, 2);
1045 1046 1047

	while (!list_empty(&prefs)) {
		ref = list_first_entry(&prefs, struct __prelim_ref, list);
J
Julia Lawall 已提交
1048
		WARN_ON(ref->count < 0);
1049
		if (roots && ref->count && ref->root_id && ref->parent == 0) {
1050 1051 1052 1053 1054
			if (root_objectid && ref->root_id != root_objectid) {
				ret = BACKREF_FOUND_SHARED;
				goto out;
			}

1055 1056
			/* no parent == root of tree */
			ret = ulist_add(roots, ref->root_id, 0, GFP_NOFS);
1057 1058
			if (ret < 0)
				goto out;
1059 1060
		}
		if (ref->count && ref->parent) {
1061 1062
			if (extent_item_pos && !ref->inode_list &&
			    ref->level == 0) {
1063
				struct extent_buffer *eb;
1064

1065
				eb = read_tree_block(fs_info->extent_root,
1066
							   ref->parent, 0);
1067 1068 1069 1070
				if (IS_ERR(eb)) {
					ret = PTR_ERR(eb);
					goto out;
				} else if (!extent_buffer_uptodate(eb)) {
1071
					free_extent_buffer(eb);
1072 1073
					ret = -EIO;
					goto out;
1074
				}
1075 1076
				btrfs_tree_read_lock(eb);
				btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK);
1077 1078
				ret = find_extent_in_eb(eb, bytenr,
							*extent_item_pos, &eie);
1079
				btrfs_tree_read_unlock_blocking(eb);
1080
				free_extent_buffer(eb);
1081 1082 1083
				if (ret < 0)
					goto out;
				ref->inode_list = eie;
1084
			}
1085 1086 1087
			ret = ulist_add_merge_ptr(refs, ref->parent,
						  ref->inode_list,
						  (void **)&eie, GFP_NOFS);
1088 1089
			if (ret < 0)
				goto out;
1090 1091 1092 1093 1094 1095 1096 1097 1098 1099
			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;
			}
1100
			eie = NULL;
1101
		}
1102
		list_del(&ref->list);
1103
		kmem_cache_free(btrfs_prelim_ref_cache, ref);
1104 1105 1106 1107 1108 1109 1110
	}

out:
	btrfs_free_path(path);
	while (!list_empty(&prefs)) {
		ref = list_first_entry(&prefs, struct __prelim_ref, list);
		list_del(&ref->list);
1111
		kmem_cache_free(btrfs_prelim_ref_cache, ref);
1112 1113 1114 1115 1116
	}
	while (!list_empty(&prefs_delayed)) {
		ref = list_first_entry(&prefs_delayed, struct __prelim_ref,
				       list);
		list_del(&ref->list);
1117
		kmem_cache_free(btrfs_prelim_ref_cache, ref);
1118
	}
1119 1120
	if (ret < 0)
		free_inode_elem_list(eie);
1121 1122 1123
	return ret;
}

1124 1125 1126 1127 1128 1129 1130 1131 1132 1133
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;
1134
		eie = (struct extent_inode_elem *)(uintptr_t)node->aux;
1135
		free_inode_elem_list(eie);
1136 1137 1138 1139 1140 1141
		node->aux = 0;
	}

	ulist_free(blocks);
}

1142 1143 1144 1145 1146 1147 1148 1149 1150 1151
/*
 * 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,
1152
				u64 time_seq, struct ulist **leafs,
1153
				const u64 *extent_item_pos)
1154 1155 1156 1157
{
	int ret;

	*leafs = ulist_alloc(GFP_NOFS);
1158
	if (!*leafs)
1159 1160
		return -ENOMEM;

1161
	ret = find_parent_nodes(trans, fs_info, bytenr,
1162
				time_seq, *leafs, NULL, extent_item_pos, 0, 0);
1163
	if (ret < 0 && ret != -ENOENT) {
1164
		free_leaf_list(*leafs);
1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183
		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.
 */
1184 1185 1186
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)
1187 1188 1189
{
	struct ulist *tmp;
	struct ulist_node *node = NULL;
J
Jan Schmidt 已提交
1190
	struct ulist_iterator uiter;
1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201
	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
1202
	ULIST_ITER_INIT(&uiter);
1203
	while (1) {
1204
		ret = find_parent_nodes(trans, fs_info, bytenr,
1205
					time_seq, tmp, *roots, NULL, 0, 0);
1206 1207 1208 1209 1210
		if (ret < 0 && ret != -ENOENT) {
			ulist_free(tmp);
			ulist_free(*roots);
			return ret;
		}
J
Jan Schmidt 已提交
1211
		node = ulist_next(tmp, &uiter);
1212 1213 1214
		if (!node)
			break;
		bytenr = node->val;
1215
		cond_resched();
1216 1217 1218 1219 1220 1221
	}

	ulist_free(tmp);
	return 0;
}

1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235
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;
}

1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248
/**
 * 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.
 */
1249 1250 1251 1252 1253 1254 1255 1256
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;
1257
	struct seq_list elem = SEQ_LIST_INIT(elem);
1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276
	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,
					roots, NULL, root_objectid, inum);
		if (ret == BACKREF_FOUND_SHARED) {
1277
			/* this is the only condition under which we return 1 */
1278 1279 1280 1281 1282
			ret = 1;
			break;
		}
		if (ret < 0 && ret != -ENOENT)
			break;
1283
		ret = 0;
1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298
		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
1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311
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;
	struct extent_buffer *leaf;
	unsigned long ptr;

	key.objectid = inode_objectid;
1312
	key.type = BTRFS_INODE_EXTREF_KEY;
M
Mark Fasheh 已提交
1313 1314 1315 1316 1317 1318 1319 1320 1321 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
	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;
1352
		if (found_key.type != BTRFS_INODE_EXTREF_KEY)
M
Mark Fasheh 已提交
1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366
			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;
}

1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380
/*
 * 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!
 */
1381 1382 1383 1384
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)
1385 1386 1387 1388
{
	int slot;
	u64 next_inum;
	int ret;
1389
	s64 bytes_left = ((s64)size) - 1;
1390 1391
	struct extent_buffer *eb = eb_in;
	struct btrfs_key found_key;
1392
	int leave_spinning = path->leave_spinning;
M
Mark Fasheh 已提交
1393
	struct btrfs_inode_ref *iref;
1394 1395 1396 1397

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

1398
	path->leave_spinning = 1;
1399
	while (1) {
M
Mark Fasheh 已提交
1400
		bytes_left -= name_len;
1401 1402
		if (bytes_left >= 0)
			read_extent_buffer(eb, dest + bytes_left,
M
Mark Fasheh 已提交
1403
					   name_off, name_len);
1404
		if (eb != eb_in) {
1405 1406
			if (!path->skip_locking)
				btrfs_tree_read_unlock_blocking(eb);
1407
			free_extent_buffer(eb);
1408
		}
1409 1410
		ret = btrfs_find_item(fs_root, path, parent, 0,
				BTRFS_INODE_REF_KEY, &found_key);
1411 1412
		if (ret > 0)
			ret = -ENOENT;
1413 1414
		if (ret)
			break;
M
Mark Fasheh 已提交
1415

1416 1417 1418 1419 1420 1421 1422 1423 1424
		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 */
1425
		if (eb != eb_in) {
1426 1427 1428 1429
			if (!path->skip_locking)
				btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK);
			path->nodes[0] = NULL;
			path->locks[0] = 0;
1430
		}
1431 1432
		btrfs_release_path(path);
		iref = btrfs_item_ptr(eb, slot, struct btrfs_inode_ref);
M
Mark Fasheh 已提交
1433 1434 1435 1436

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

1437 1438 1439 1440 1441 1442 1443
		parent = next_inum;
		--bytes_left;
		if (bytes_left >= 0)
			dest[bytes_left] = '/';
	}

	btrfs_release_path(path);
1444
	path->leave_spinning = leave_spinning;
1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457

	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,
1458 1459
			struct btrfs_path *path, struct btrfs_key *found_key,
			u64 *flags_ret)
1460 1461 1462
{
	int ret;
	u64 flags;
1463
	u64 size = 0;
1464 1465 1466 1467 1468
	u32 item_size;
	struct extent_buffer *eb;
	struct btrfs_extent_item *ei;
	struct btrfs_key key;

1469 1470 1471 1472
	if (btrfs_fs_incompat(fs_info, SKINNY_METADATA))
		key.type = BTRFS_METADATA_ITEM_KEY;
	else
		key.type = BTRFS_EXTENT_ITEM_KEY;
1473 1474 1475 1476 1477 1478 1479
	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;

1480 1481 1482 1483 1484
	ret = btrfs_previous_extent_item(fs_info->extent_root, path, 0);
	if (ret) {
		if (ret > 0)
			ret = -ENOENT;
		return ret;
1485
	}
1486
	btrfs_item_key_to_cpu(path->nodes[0], found_key, path->slots[0]);
1487
	if (found_key->type == BTRFS_METADATA_ITEM_KEY)
1488
		size = fs_info->extent_root->nodesize;
1489 1490 1491
	else if (found_key->type == BTRFS_EXTENT_ITEM_KEY)
		size = found_key->offset;

1492
	if (found_key->objectid > logical ||
1493
	    found_key->objectid + size <= logical) {
1494
		pr_debug("logical %llu is not within any extent\n", logical);
1495
		return -ENOENT;
J
Jan Schmidt 已提交
1496
	}
1497 1498 1499 1500 1501 1502 1503 1504

	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);

J
Jan Schmidt 已提交
1505 1506
	pr_debug("logical %llu is at position %llu within the extent (%llu "
		 "EXTENT_ITEM %llu) flags %#llx size %u\n",
1507 1508
		 logical, logical - found_key->objectid, found_key->objectid,
		 found_key->offset, flags, item_size);
1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519

	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;
	}
1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532

	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
 */
static int __get_extent_inline_ref(unsigned long *ptr, struct extent_buffer *eb,
1533 1534 1535 1536
				   struct btrfs_key *key,
				   struct btrfs_extent_item *ei, u32 item_size,
				   struct btrfs_extent_inline_ref **out_eiref,
				   int *out_type)
1537 1538 1539 1540 1541 1542 1543 1544 1545
{
	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) {
1546 1547 1548 1549 1550 1551 1552 1553 1554 1555
			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);
			}
1556 1557 1558 1559
		} else {
			*out_eiref = (struct btrfs_extent_inline_ref *)(ei + 1);
		}
		*ptr = (unsigned long)*out_eiref;
1560
		if ((unsigned long)(*ptr) >= (unsigned long)ei + item_size)
1561 1562 1563 1564
			return -ENOENT;
	}

	end = (unsigned long)ei + item_size;
1565
	*out_eiref = (struct btrfs_extent_inline_ref *)(*ptr);
1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583
	*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,
1584 1585
			    struct btrfs_key *key, struct btrfs_extent_item *ei,
			    u32 item_size, u64 *out_root, u8 *out_level)
1586 1587 1588 1589 1590 1591 1592 1593 1594
{
	int ret;
	int type;
	struct btrfs_extent_inline_ref *eiref;

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

	while (1) {
1595 1596
		ret = __get_extent_inline_ref(ptr, eb, key, ei, item_size,
					      &eiref, &type);
1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609
		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);
1610 1611 1612 1613 1614 1615 1616 1617 1618 1619

	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;
	}
1620 1621 1622 1623 1624 1625 1626

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

	return 0;
}

1627 1628
static int iterate_leaf_refs(struct extent_inode_elem *inode_list,
				u64 root, u64 extent_item_objectid,
J
Jan Schmidt 已提交
1629
				iterate_extent_inodes_t *iterate, void *ctx)
1630
{
1631
	struct extent_inode_elem *eie;
J
Jan Schmidt 已提交
1632 1633
	int ret = 0;

1634
	for (eie = inode_list; eie; eie = eie->next) {
J
Jan Schmidt 已提交
1635
		pr_debug("ref for %llu resolved, key (%llu EXTEND_DATA %llu), "
1636 1637 1638
			 "root %llu\n", extent_item_objectid,
			 eie->inum, eie->offset, root);
		ret = iterate(eie->inum, eie->offset, root, ctx);
J
Jan Schmidt 已提交
1639
		if (ret) {
1640 1641
			pr_debug("stopping iteration for %llu due to ret=%d\n",
				 extent_item_objectid, ret);
J
Jan Schmidt 已提交
1642 1643
			break;
		}
1644 1645 1646 1647 1648 1649 1650
	}

	return ret;
}

/*
 * calls iterate() for every inode that references the extent identified by
J
Jan Schmidt 已提交
1651
 * the given parameters.
1652 1653 1654
 * 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
1655
				u64 extent_item_objectid, u64 extent_item_pos,
1656
				int search_commit_root,
1657 1658 1659
				iterate_extent_inodes_t *iterate, void *ctx)
{
	int ret;
1660
	struct btrfs_trans_handle *trans = NULL;
1661 1662
	struct ulist *refs = NULL;
	struct ulist *roots = NULL;
J
Jan Schmidt 已提交
1663 1664
	struct ulist_node *ref_node = NULL;
	struct ulist_node *root_node = NULL;
1665
	struct seq_list tree_mod_seq_elem = SEQ_LIST_INIT(tree_mod_seq_elem);
J
Jan Schmidt 已提交
1666 1667
	struct ulist_iterator ref_uiter;
	struct ulist_iterator root_uiter;
1668

J
Jan Schmidt 已提交
1669 1670
	pr_debug("resolving all inodes for extent %llu\n",
			extent_item_objectid);
1671

1672
	if (!search_commit_root) {
1673 1674 1675
		trans = btrfs_join_transaction(fs_info->extent_root);
		if (IS_ERR(trans))
			return PTR_ERR(trans);
1676
		btrfs_get_tree_mod_seq(fs_info, &tree_mod_seq_elem);
1677 1678
	} else {
		down_read(&fs_info->commit_root_sem);
1679
	}
1680

J
Jan Schmidt 已提交
1681
	ret = btrfs_find_all_leafs(trans, fs_info, extent_item_objectid,
1682
				   tree_mod_seq_elem.seq, &refs,
1683
				   &extent_item_pos);
J
Jan Schmidt 已提交
1684 1685
	if (ret)
		goto out;
1686

J
Jan Schmidt 已提交
1687 1688
	ULIST_ITER_INIT(&ref_uiter);
	while (!ret && (ref_node = ulist_next(refs, &ref_uiter))) {
1689 1690
		ret = __btrfs_find_all_roots(trans, fs_info, ref_node->val,
					     tree_mod_seq_elem.seq, &roots);
J
Jan Schmidt 已提交
1691 1692
		if (ret)
			break;
J
Jan Schmidt 已提交
1693 1694
		ULIST_ITER_INIT(&root_uiter);
		while (!ret && (root_node = ulist_next(roots, &root_uiter))) {
1695
			pr_debug("root %llu references leaf %llu, data list "
1696
				 "%#llx\n", root_node->val, ref_node->val,
1697
				 ref_node->aux);
1698 1699 1700 1701 1702
			ret = iterate_leaf_refs((struct extent_inode_elem *)
						(uintptr_t)ref_node->aux,
						root_node->val,
						extent_item_objectid,
						iterate, ctx);
J
Jan Schmidt 已提交
1703
		}
1704
		ulist_free(roots);
1705 1706
	}

1707
	free_leaf_list(refs);
J
Jan Schmidt 已提交
1708
out:
1709
	if (!search_commit_root) {
1710
		btrfs_put_tree_mod_seq(fs_info, &tree_mod_seq_elem);
1711
		btrfs_end_transaction(trans, fs_info->extent_root);
1712 1713
	} else {
		up_read(&fs_info->commit_root_sem);
1714 1715
	}

1716 1717 1718 1719 1720 1721 1722 1723
	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
1724
	u64 extent_item_pos;
1725
	u64 flags = 0;
1726
	struct btrfs_key found_key;
1727
	int search_commit_root = path->search_commit_root;
1728

1729
	ret = extent_from_logical(fs_info, logical, path, &found_key, &flags);
J
Jan Schmidt 已提交
1730
	btrfs_release_path(path);
1731 1732
	if (ret < 0)
		return ret;
1733
	if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK)
1734
		return -EINVAL;
1735

J
Jan Schmidt 已提交
1736
	extent_item_pos = logical - found_key.objectid;
1737 1738 1739
	ret = iterate_extent_inodes(fs_info, found_key.objectid,
					extent_item_pos, search_commit_root,
					iterate, ctx);
1740 1741 1742 1743

	return ret;
}

M
Mark Fasheh 已提交
1744 1745 1746 1747 1748 1749
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)
1750
{
1751
	int ret = 0;
1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762
	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;

1763
	while (!ret) {
1764 1765 1766 1767
		ret = btrfs_find_item(fs_root, path, inum,
				parent ? parent + 1 : 0, BTRFS_INODE_REF_KEY,
				&found_key);

1768 1769 1770 1771 1772 1773 1774 1775 1776 1777
		if (ret < 0)
			break;
		if (ret) {
			ret = found ? 0 : -ENOENT;
			break;
		}
		++found;

		parent = found_key.offset;
		slot = path->slots[0];
1778 1779 1780 1781 1782 1783
		eb = btrfs_clone_extent_buffer(path->nodes[0]);
		if (!eb) {
			ret = -ENOMEM;
			break;
		}
		extent_buffer_get(eb);
1784 1785
		btrfs_tree_read_lock(eb);
		btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK);
1786 1787
		btrfs_release_path(path);

1788
		item = btrfs_item_nr(slot);
1789 1790 1791 1792 1793
		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()! */
J
Jan Schmidt 已提交
1794
			pr_debug("following ref at offset %u for inode %llu in "
1795 1796
				 "tree %llu\n", cur, found_key.objectid,
				 fs_root->objectid);
M
Mark Fasheh 已提交
1797 1798
			ret = iterate(parent, name_len,
				      (unsigned long)(iref + 1), eb, ctx);
1799
			if (ret)
1800 1801 1802 1803
				break;
			len = sizeof(*iref) + name_len;
			iref = (struct btrfs_inode_ref *)((char *)iref + len);
		}
1804
		btrfs_tree_read_unlock_blocking(eb);
1805 1806 1807 1808 1809 1810 1811 1812
		free_extent_buffer(eb);
	}

	btrfs_release_path(path);

	return ret;
}

M
Mark Fasheh 已提交
1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839
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];
1840 1841 1842 1843 1844 1845
		eb = btrfs_clone_extent_buffer(path->nodes[0]);
		if (!eb) {
			ret = -ENOMEM;
			break;
		}
		extent_buffer_get(eb);
M
Mark Fasheh 已提交
1846 1847 1848 1849 1850

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

1851 1852
		item_size = btrfs_item_size_nr(eb, slot);
		ptr = btrfs_item_ptr_offset(eb, slot);
M
Mark Fasheh 已提交
1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865
		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;

1866
			cur_offset += btrfs_inode_extref_name_len(eb, extref);
M
Mark Fasheh 已提交
1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899
			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;
}

1900 1901 1902 1903
/*
 * returns 0 if the path could be dumped (probably truncated)
 * returns <0 in case of an error
 */
M
Mark Fasheh 已提交
1904 1905
static int inode_to_path(u64 inum, u32 name_len, unsigned long name_off,
			 struct extent_buffer *eb, void *ctx)
1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916
{
	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;

1917
	fspath_min = (char *)ipath->fspath->val + (i + 1) * s_ptr;
1918 1919
	fspath = btrfs_ref_to_path(ipath->fs_root, ipath->btrfs_path, name_len,
				   name_off, eb, inum, fspath_min, bytes_left);
1920 1921 1922 1923
	if (IS_ERR(fspath))
		return PTR_ERR(fspath);

	if (fspath > fspath_min) {
1924
		ipath->fspath->val[i] = (u64)(unsigned long)fspath;
1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938
		++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
1939
 * from ipath->fspath->val[i].
1940
 * when it returns, there are ipath->fspath->elem_cnt number of paths available
1941
 * in ipath->fspath->val[]. when the allocated space wasn't sufficient, the
1942
 * number of missed paths is recorded in ipath->fspath->elem_missed, otherwise,
1943 1944 1945 1946 1947 1948
 * 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
1949
			     inode_to_path, ipath);
1950 1951 1952 1953 1954 1955 1956 1957
}

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));
1958
	data = vmalloc(alloc_bytes);
1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993
	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;

	ifp = kmalloc(sizeof(*ifp), GFP_NOFS);
	if (!ifp) {
1994
		vfree(fspath);
1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006
		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)
{
2007 2008
	if (!ipath)
		return;
2009
	vfree(ipath->fspath);
2010 2011
	kfree(ipath);
}