1 /* 2 * Copyright (C) 2011 STRATO. All rights reserved. 3 * 4 * This program is free software; you can redistribute it and/or 5 * modify it under the terms of the GNU General Public 6 * License v2 as published by the Free Software Foundation. 7 * 8 * This program is distributed in the hope that it will be useful, 9 * but WITHOUT ANY WARRANTY; without even the implied warranty of 10 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU 11 * General Public License for more details. 12 * 13 * You should have received a copy of the GNU General Public 14 * License along with this program; if not, write to the 15 * Free Software Foundation, Inc., 59 Temple Place - Suite 330, 16 * Boston, MA 021110-1307, USA. 17 */ 18 19 #include <linux/vmalloc.h> 20 #include "ctree.h" 21 #include "disk-io.h" 22 #include "backref.h" 23 #include "ulist.h" 24 #include "transaction.h" 25 #include "delayed-ref.h" 26 #include "locking.h" 27 28 struct extent_inode_elem { 29 u64 inum; 30 u64 offset; 31 struct extent_inode_elem *next; 32 }; 33 34 static int check_extent_in_eb(struct btrfs_key *key, struct extent_buffer *eb, 35 struct btrfs_file_extent_item *fi, 36 u64 extent_item_pos, 37 struct extent_inode_elem **eie) 38 { 39 u64 data_offset; 40 u64 data_len; 41 struct extent_inode_elem *e; 42 43 data_offset = btrfs_file_extent_offset(eb, fi); 44 data_len = btrfs_file_extent_num_bytes(eb, fi); 45 46 if (extent_item_pos < data_offset || 47 extent_item_pos >= data_offset + data_len) 48 return 1; 49 50 e = kmalloc(sizeof(*e), GFP_NOFS); 51 if (!e) 52 return -ENOMEM; 53 54 e->next = *eie; 55 e->inum = key->objectid; 56 e->offset = key->offset + (extent_item_pos - data_offset); 57 *eie = e; 58 59 return 0; 60 } 61 62 static int find_extent_in_eb(struct extent_buffer *eb, u64 wanted_disk_byte, 63 u64 extent_item_pos, 64 struct extent_inode_elem **eie) 65 { 66 u64 disk_byte; 67 struct btrfs_key key; 68 struct btrfs_file_extent_item *fi; 69 int slot; 70 int nritems; 71 int extent_type; 72 int ret; 73 74 /* 75 * from the shared data ref, we only have the leaf but we need 76 * the key. thus, we must look into all items and see that we 77 * find one (some) with a reference to our extent item. 78 */ 79 nritems = btrfs_header_nritems(eb); 80 for (slot = 0; slot < nritems; ++slot) { 81 btrfs_item_key_to_cpu(eb, &key, slot); 82 if (key.type != BTRFS_EXTENT_DATA_KEY) 83 continue; 84 fi = btrfs_item_ptr(eb, slot, struct btrfs_file_extent_item); 85 extent_type = btrfs_file_extent_type(eb, fi); 86 if (extent_type == BTRFS_FILE_EXTENT_INLINE) 87 continue; 88 /* don't skip BTRFS_FILE_EXTENT_PREALLOC, we can handle that */ 89 disk_byte = btrfs_file_extent_disk_bytenr(eb, fi); 90 if (disk_byte != wanted_disk_byte) 91 continue; 92 93 ret = check_extent_in_eb(&key, eb, fi, extent_item_pos, eie); 94 if (ret < 0) 95 return ret; 96 } 97 98 return 0; 99 } 100 101 /* 102 * this structure records all encountered refs on the way up to the root 103 */ 104 struct __prelim_ref { 105 struct list_head list; 106 u64 root_id; 107 struct btrfs_key key_for_search; 108 int level; 109 int count; 110 struct extent_inode_elem *inode_list; 111 u64 parent; 112 u64 wanted_disk_byte; 113 }; 114 115 /* 116 * the rules for all callers of this function are: 117 * - obtaining the parent is the goal 118 * - if you add a key, you must know that it is a correct key 119 * - if you cannot add the parent or a correct key, then we will look into the 120 * block later to set a correct key 121 * 122 * delayed refs 123 * ============ 124 * backref type | shared | indirect | shared | indirect 125 * information | tree | tree | data | data 126 * --------------------+--------+----------+--------+---------- 127 * parent logical | y | - | - | - 128 * key to resolve | - | y | y | y 129 * tree block logical | - | - | - | - 130 * root for resolving | y | y | y | y 131 * 132 * - column 1: we've the parent -> done 133 * - column 2, 3, 4: we use the key to find the parent 134 * 135 * on disk refs (inline or keyed) 136 * ============================== 137 * backref type | shared | indirect | shared | indirect 138 * information | tree | tree | data | data 139 * --------------------+--------+----------+--------+---------- 140 * parent logical | y | - | y | - 141 * key to resolve | - | - | - | y 142 * tree block logical | y | y | y | y 143 * root for resolving | - | y | y | y 144 * 145 * - column 1, 3: we've the parent -> done 146 * - column 2: we take the first key from the block to find the parent 147 * (see __add_missing_keys) 148 * - column 4: we use the key to find the parent 149 * 150 * additional information that's available but not required to find the parent 151 * block might help in merging entries to gain some speed. 152 */ 153 154 static int __add_prelim_ref(struct list_head *head, u64 root_id, 155 struct btrfs_key *key, int level, 156 u64 parent, u64 wanted_disk_byte, int count) 157 { 158 struct __prelim_ref *ref; 159 160 /* in case we're adding delayed refs, we're holding the refs spinlock */ 161 ref = kmalloc(sizeof(*ref), GFP_ATOMIC); 162 if (!ref) 163 return -ENOMEM; 164 165 ref->root_id = root_id; 166 if (key) 167 ref->key_for_search = *key; 168 else 169 memset(&ref->key_for_search, 0, sizeof(ref->key_for_search)); 170 171 ref->inode_list = NULL; 172 ref->level = level; 173 ref->count = count; 174 ref->parent = parent; 175 ref->wanted_disk_byte = wanted_disk_byte; 176 list_add_tail(&ref->list, head); 177 178 return 0; 179 } 180 181 static int add_all_parents(struct btrfs_root *root, struct btrfs_path *path, 182 struct ulist *parents, int level, 183 struct btrfs_key *key_for_search, u64 time_seq, 184 u64 wanted_disk_byte, 185 const u64 *extent_item_pos) 186 { 187 int ret = 0; 188 int slot; 189 struct extent_buffer *eb; 190 struct btrfs_key key; 191 struct btrfs_file_extent_item *fi; 192 struct extent_inode_elem *eie = NULL; 193 u64 disk_byte; 194 195 if (level != 0) { 196 eb = path->nodes[level]; 197 ret = ulist_add(parents, eb->start, 0, GFP_NOFS); 198 if (ret < 0) 199 return ret; 200 return 0; 201 } 202 203 /* 204 * We normally enter this function with the path already pointing to 205 * the first item to check. But sometimes, we may enter it with 206 * slot==nritems. In that case, go to the next leaf before we continue. 207 */ 208 if (path->slots[0] >= btrfs_header_nritems(path->nodes[0])) 209 ret = btrfs_next_old_leaf(root, path, time_seq); 210 211 while (!ret) { 212 eb = path->nodes[0]; 213 slot = path->slots[0]; 214 215 btrfs_item_key_to_cpu(eb, &key, slot); 216 217 if (key.objectid != key_for_search->objectid || 218 key.type != BTRFS_EXTENT_DATA_KEY) 219 break; 220 221 fi = btrfs_item_ptr(eb, slot, struct btrfs_file_extent_item); 222 disk_byte = btrfs_file_extent_disk_bytenr(eb, fi); 223 224 if (disk_byte == wanted_disk_byte) { 225 eie = NULL; 226 if (extent_item_pos) { 227 ret = check_extent_in_eb(&key, eb, fi, 228 *extent_item_pos, 229 &eie); 230 if (ret < 0) 231 break; 232 } 233 if (!ret) { 234 ret = ulist_add(parents, eb->start, 235 (uintptr_t)eie, GFP_NOFS); 236 if (ret < 0) 237 break; 238 if (!extent_item_pos) { 239 ret = btrfs_next_old_leaf(root, path, 240 time_seq); 241 continue; 242 } 243 } 244 } 245 ret = btrfs_next_old_item(root, path, time_seq); 246 } 247 248 if (ret > 0) 249 ret = 0; 250 return ret; 251 } 252 253 /* 254 * resolve an indirect backref in the form (root_id, key, level) 255 * to a logical address 256 */ 257 static int __resolve_indirect_ref(struct btrfs_fs_info *fs_info, 258 int search_commit_root, 259 u64 time_seq, 260 struct __prelim_ref *ref, 261 struct ulist *parents, 262 const u64 *extent_item_pos) 263 { 264 struct btrfs_path *path; 265 struct btrfs_root *root; 266 struct btrfs_key root_key; 267 struct extent_buffer *eb; 268 int ret = 0; 269 int root_level; 270 int level = ref->level; 271 272 path = btrfs_alloc_path(); 273 if (!path) 274 return -ENOMEM; 275 path->search_commit_root = !!search_commit_root; 276 277 root_key.objectid = ref->root_id; 278 root_key.type = BTRFS_ROOT_ITEM_KEY; 279 root_key.offset = (u64)-1; 280 root = btrfs_read_fs_root_no_name(fs_info, &root_key); 281 if (IS_ERR(root)) { 282 ret = PTR_ERR(root); 283 goto out; 284 } 285 286 root_level = btrfs_old_root_level(root, time_seq); 287 288 if (root_level + 1 == level) 289 goto out; 290 291 path->lowest_level = level; 292 ret = btrfs_search_old_slot(root, &ref->key_for_search, path, time_seq); 293 pr_debug("search slot in root %llu (level %d, ref count %d) returned " 294 "%d for key (%llu %u %llu)\n", 295 (unsigned long long)ref->root_id, level, ref->count, ret, 296 (unsigned long long)ref->key_for_search.objectid, 297 ref->key_for_search.type, 298 (unsigned long long)ref->key_for_search.offset); 299 if (ret < 0) 300 goto out; 301 302 eb = path->nodes[level]; 303 while (!eb) { 304 if (!level) { 305 WARN_ON(1); 306 ret = 1; 307 goto out; 308 } 309 level--; 310 eb = path->nodes[level]; 311 } 312 313 ret = add_all_parents(root, path, parents, level, &ref->key_for_search, 314 time_seq, ref->wanted_disk_byte, 315 extent_item_pos); 316 out: 317 btrfs_free_path(path); 318 return ret; 319 } 320 321 /* 322 * resolve all indirect backrefs from the list 323 */ 324 static int __resolve_indirect_refs(struct btrfs_fs_info *fs_info, 325 int search_commit_root, u64 time_seq, 326 struct list_head *head, 327 const u64 *extent_item_pos) 328 { 329 int err; 330 int ret = 0; 331 struct __prelim_ref *ref; 332 struct __prelim_ref *ref_safe; 333 struct __prelim_ref *new_ref; 334 struct ulist *parents; 335 struct ulist_node *node; 336 struct ulist_iterator uiter; 337 338 parents = ulist_alloc(GFP_NOFS); 339 if (!parents) 340 return -ENOMEM; 341 342 /* 343 * _safe allows us to insert directly after the current item without 344 * iterating over the newly inserted items. 345 * we're also allowed to re-assign ref during iteration. 346 */ 347 list_for_each_entry_safe(ref, ref_safe, head, list) { 348 if (ref->parent) /* already direct */ 349 continue; 350 if (ref->count == 0) 351 continue; 352 err = __resolve_indirect_ref(fs_info, search_commit_root, 353 time_seq, ref, parents, 354 extent_item_pos); 355 if (err) 356 continue; 357 358 /* we put the first parent into the ref at hand */ 359 ULIST_ITER_INIT(&uiter); 360 node = ulist_next(parents, &uiter); 361 ref->parent = node ? node->val : 0; 362 ref->inode_list = node ? 363 (struct extent_inode_elem *)(uintptr_t)node->aux : 0; 364 365 /* additional parents require new refs being added here */ 366 while ((node = ulist_next(parents, &uiter))) { 367 new_ref = kmalloc(sizeof(*new_ref), GFP_NOFS); 368 if (!new_ref) { 369 ret = -ENOMEM; 370 break; 371 } 372 memcpy(new_ref, ref, sizeof(*ref)); 373 new_ref->parent = node->val; 374 new_ref->inode_list = (struct extent_inode_elem *) 375 (uintptr_t)node->aux; 376 list_add(&new_ref->list, &ref->list); 377 } 378 ulist_reinit(parents); 379 } 380 381 ulist_free(parents); 382 return ret; 383 } 384 385 static inline int ref_for_same_block(struct __prelim_ref *ref1, 386 struct __prelim_ref *ref2) 387 { 388 if (ref1->level != ref2->level) 389 return 0; 390 if (ref1->root_id != ref2->root_id) 391 return 0; 392 if (ref1->key_for_search.type != ref2->key_for_search.type) 393 return 0; 394 if (ref1->key_for_search.objectid != ref2->key_for_search.objectid) 395 return 0; 396 if (ref1->key_for_search.offset != ref2->key_for_search.offset) 397 return 0; 398 if (ref1->parent != ref2->parent) 399 return 0; 400 401 return 1; 402 } 403 404 /* 405 * read tree blocks and add keys where required. 406 */ 407 static int __add_missing_keys(struct btrfs_fs_info *fs_info, 408 struct list_head *head) 409 { 410 struct list_head *pos; 411 struct extent_buffer *eb; 412 413 list_for_each(pos, head) { 414 struct __prelim_ref *ref; 415 ref = list_entry(pos, struct __prelim_ref, list); 416 417 if (ref->parent) 418 continue; 419 if (ref->key_for_search.type) 420 continue; 421 BUG_ON(!ref->wanted_disk_byte); 422 eb = read_tree_block(fs_info->tree_root, ref->wanted_disk_byte, 423 fs_info->tree_root->leafsize, 0); 424 BUG_ON(!eb); 425 btrfs_tree_read_lock(eb); 426 if (btrfs_header_level(eb) == 0) 427 btrfs_item_key_to_cpu(eb, &ref->key_for_search, 0); 428 else 429 btrfs_node_key_to_cpu(eb, &ref->key_for_search, 0); 430 btrfs_tree_read_unlock(eb); 431 free_extent_buffer(eb); 432 } 433 return 0; 434 } 435 436 /* 437 * merge two lists of backrefs and adjust counts accordingly 438 * 439 * mode = 1: merge identical keys, if key is set 440 * FIXME: if we add more keys in __add_prelim_ref, we can merge more here. 441 * additionally, we could even add a key range for the blocks we 442 * looked into to merge even more (-> replace unresolved refs by those 443 * having a parent). 444 * mode = 2: merge identical parents 445 */ 446 static int __merge_refs(struct list_head *head, int mode) 447 { 448 struct list_head *pos1; 449 450 list_for_each(pos1, head) { 451 struct list_head *n2; 452 struct list_head *pos2; 453 struct __prelim_ref *ref1; 454 455 ref1 = list_entry(pos1, struct __prelim_ref, list); 456 457 for (pos2 = pos1->next, n2 = pos2->next; pos2 != head; 458 pos2 = n2, n2 = pos2->next) { 459 struct __prelim_ref *ref2; 460 struct __prelim_ref *xchg; 461 struct extent_inode_elem *eie; 462 463 ref2 = list_entry(pos2, struct __prelim_ref, list); 464 465 if (mode == 1) { 466 if (!ref_for_same_block(ref1, ref2)) 467 continue; 468 if (!ref1->parent && ref2->parent) { 469 xchg = ref1; 470 ref1 = ref2; 471 ref2 = xchg; 472 } 473 } else { 474 if (ref1->parent != ref2->parent) 475 continue; 476 } 477 478 eie = ref1->inode_list; 479 while (eie && eie->next) 480 eie = eie->next; 481 if (eie) 482 eie->next = ref2->inode_list; 483 else 484 ref1->inode_list = ref2->inode_list; 485 ref1->count += ref2->count; 486 487 list_del(&ref2->list); 488 kfree(ref2); 489 } 490 491 } 492 return 0; 493 } 494 495 /* 496 * add all currently queued delayed refs from this head whose seq nr is 497 * smaller or equal that seq to the list 498 */ 499 static int __add_delayed_refs(struct btrfs_delayed_ref_head *head, u64 seq, 500 struct list_head *prefs) 501 { 502 struct btrfs_delayed_extent_op *extent_op = head->extent_op; 503 struct rb_node *n = &head->node.rb_node; 504 struct btrfs_key key; 505 struct btrfs_key op_key = {0}; 506 int sgn; 507 int ret = 0; 508 509 if (extent_op && extent_op->update_key) 510 btrfs_disk_key_to_cpu(&op_key, &extent_op->key); 511 512 while ((n = rb_prev(n))) { 513 struct btrfs_delayed_ref_node *node; 514 node = rb_entry(n, struct btrfs_delayed_ref_node, 515 rb_node); 516 if (node->bytenr != head->node.bytenr) 517 break; 518 WARN_ON(node->is_head); 519 520 if (node->seq > seq) 521 continue; 522 523 switch (node->action) { 524 case BTRFS_ADD_DELAYED_EXTENT: 525 case BTRFS_UPDATE_DELAYED_HEAD: 526 WARN_ON(1); 527 continue; 528 case BTRFS_ADD_DELAYED_REF: 529 sgn = 1; 530 break; 531 case BTRFS_DROP_DELAYED_REF: 532 sgn = -1; 533 break; 534 default: 535 BUG_ON(1); 536 } 537 switch (node->type) { 538 case BTRFS_TREE_BLOCK_REF_KEY: { 539 struct btrfs_delayed_tree_ref *ref; 540 541 ref = btrfs_delayed_node_to_tree_ref(node); 542 ret = __add_prelim_ref(prefs, ref->root, &op_key, 543 ref->level + 1, 0, node->bytenr, 544 node->ref_mod * sgn); 545 break; 546 } 547 case BTRFS_SHARED_BLOCK_REF_KEY: { 548 struct btrfs_delayed_tree_ref *ref; 549 550 ref = btrfs_delayed_node_to_tree_ref(node); 551 ret = __add_prelim_ref(prefs, ref->root, NULL, 552 ref->level + 1, ref->parent, 553 node->bytenr, 554 node->ref_mod * sgn); 555 break; 556 } 557 case BTRFS_EXTENT_DATA_REF_KEY: { 558 struct btrfs_delayed_data_ref *ref; 559 ref = btrfs_delayed_node_to_data_ref(node); 560 561 key.objectid = ref->objectid; 562 key.type = BTRFS_EXTENT_DATA_KEY; 563 key.offset = ref->offset; 564 ret = __add_prelim_ref(prefs, ref->root, &key, 0, 0, 565 node->bytenr, 566 node->ref_mod * sgn); 567 break; 568 } 569 case BTRFS_SHARED_DATA_REF_KEY: { 570 struct btrfs_delayed_data_ref *ref; 571 572 ref = btrfs_delayed_node_to_data_ref(node); 573 574 key.objectid = ref->objectid; 575 key.type = BTRFS_EXTENT_DATA_KEY; 576 key.offset = ref->offset; 577 ret = __add_prelim_ref(prefs, ref->root, &key, 0, 578 ref->parent, node->bytenr, 579 node->ref_mod * sgn); 580 break; 581 } 582 default: 583 WARN_ON(1); 584 } 585 BUG_ON(ret); 586 } 587 588 return 0; 589 } 590 591 /* 592 * add all inline backrefs for bytenr to the list 593 */ 594 static int __add_inline_refs(struct btrfs_fs_info *fs_info, 595 struct btrfs_path *path, u64 bytenr, 596 int *info_level, struct list_head *prefs) 597 { 598 int ret = 0; 599 int slot; 600 struct extent_buffer *leaf; 601 struct btrfs_key key; 602 unsigned long ptr; 603 unsigned long end; 604 struct btrfs_extent_item *ei; 605 u64 flags; 606 u64 item_size; 607 608 /* 609 * enumerate all inline refs 610 */ 611 leaf = path->nodes[0]; 612 slot = path->slots[0]; 613 614 item_size = btrfs_item_size_nr(leaf, slot); 615 BUG_ON(item_size < sizeof(*ei)); 616 617 ei = btrfs_item_ptr(leaf, slot, struct btrfs_extent_item); 618 flags = btrfs_extent_flags(leaf, ei); 619 620 ptr = (unsigned long)(ei + 1); 621 end = (unsigned long)ei + item_size; 622 623 if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) { 624 struct btrfs_tree_block_info *info; 625 626 info = (struct btrfs_tree_block_info *)ptr; 627 *info_level = btrfs_tree_block_level(leaf, info); 628 ptr += sizeof(struct btrfs_tree_block_info); 629 BUG_ON(ptr > end); 630 } else { 631 BUG_ON(!(flags & BTRFS_EXTENT_FLAG_DATA)); 632 } 633 634 while (ptr < end) { 635 struct btrfs_extent_inline_ref *iref; 636 u64 offset; 637 int type; 638 639 iref = (struct btrfs_extent_inline_ref *)ptr; 640 type = btrfs_extent_inline_ref_type(leaf, iref); 641 offset = btrfs_extent_inline_ref_offset(leaf, iref); 642 643 switch (type) { 644 case BTRFS_SHARED_BLOCK_REF_KEY: 645 ret = __add_prelim_ref(prefs, 0, NULL, 646 *info_level + 1, offset, 647 bytenr, 1); 648 break; 649 case BTRFS_SHARED_DATA_REF_KEY: { 650 struct btrfs_shared_data_ref *sdref; 651 int count; 652 653 sdref = (struct btrfs_shared_data_ref *)(iref + 1); 654 count = btrfs_shared_data_ref_count(leaf, sdref); 655 ret = __add_prelim_ref(prefs, 0, NULL, 0, offset, 656 bytenr, count); 657 break; 658 } 659 case BTRFS_TREE_BLOCK_REF_KEY: 660 ret = __add_prelim_ref(prefs, offset, NULL, 661 *info_level + 1, 0, 662 bytenr, 1); 663 break; 664 case BTRFS_EXTENT_DATA_REF_KEY: { 665 struct btrfs_extent_data_ref *dref; 666 int count; 667 u64 root; 668 669 dref = (struct btrfs_extent_data_ref *)(&iref->offset); 670 count = btrfs_extent_data_ref_count(leaf, dref); 671 key.objectid = btrfs_extent_data_ref_objectid(leaf, 672 dref); 673 key.type = BTRFS_EXTENT_DATA_KEY; 674 key.offset = btrfs_extent_data_ref_offset(leaf, dref); 675 root = btrfs_extent_data_ref_root(leaf, dref); 676 ret = __add_prelim_ref(prefs, root, &key, 0, 0, 677 bytenr, count); 678 break; 679 } 680 default: 681 WARN_ON(1); 682 } 683 BUG_ON(ret); 684 ptr += btrfs_extent_inline_ref_size(type); 685 } 686 687 return 0; 688 } 689 690 /* 691 * add all non-inline backrefs for bytenr to the list 692 */ 693 static int __add_keyed_refs(struct btrfs_fs_info *fs_info, 694 struct btrfs_path *path, u64 bytenr, 695 int info_level, struct list_head *prefs) 696 { 697 struct btrfs_root *extent_root = fs_info->extent_root; 698 int ret; 699 int slot; 700 struct extent_buffer *leaf; 701 struct btrfs_key key; 702 703 while (1) { 704 ret = btrfs_next_item(extent_root, path); 705 if (ret < 0) 706 break; 707 if (ret) { 708 ret = 0; 709 break; 710 } 711 712 slot = path->slots[0]; 713 leaf = path->nodes[0]; 714 btrfs_item_key_to_cpu(leaf, &key, slot); 715 716 if (key.objectid != bytenr) 717 break; 718 if (key.type < BTRFS_TREE_BLOCK_REF_KEY) 719 continue; 720 if (key.type > BTRFS_SHARED_DATA_REF_KEY) 721 break; 722 723 switch (key.type) { 724 case BTRFS_SHARED_BLOCK_REF_KEY: 725 ret = __add_prelim_ref(prefs, 0, NULL, 726 info_level + 1, key.offset, 727 bytenr, 1); 728 break; 729 case BTRFS_SHARED_DATA_REF_KEY: { 730 struct btrfs_shared_data_ref *sdref; 731 int count; 732 733 sdref = btrfs_item_ptr(leaf, slot, 734 struct btrfs_shared_data_ref); 735 count = btrfs_shared_data_ref_count(leaf, sdref); 736 ret = __add_prelim_ref(prefs, 0, NULL, 0, key.offset, 737 bytenr, count); 738 break; 739 } 740 case BTRFS_TREE_BLOCK_REF_KEY: 741 ret = __add_prelim_ref(prefs, key.offset, NULL, 742 info_level + 1, 0, 743 bytenr, 1); 744 break; 745 case BTRFS_EXTENT_DATA_REF_KEY: { 746 struct btrfs_extent_data_ref *dref; 747 int count; 748 u64 root; 749 750 dref = btrfs_item_ptr(leaf, slot, 751 struct btrfs_extent_data_ref); 752 count = btrfs_extent_data_ref_count(leaf, dref); 753 key.objectid = btrfs_extent_data_ref_objectid(leaf, 754 dref); 755 key.type = BTRFS_EXTENT_DATA_KEY; 756 key.offset = btrfs_extent_data_ref_offset(leaf, dref); 757 root = btrfs_extent_data_ref_root(leaf, dref); 758 ret = __add_prelim_ref(prefs, root, &key, 0, 0, 759 bytenr, count); 760 break; 761 } 762 default: 763 WARN_ON(1); 764 } 765 BUG_ON(ret); 766 } 767 768 return ret; 769 } 770 771 /* 772 * this adds all existing backrefs (inline backrefs, backrefs and delayed 773 * refs) for the given bytenr to the refs list, merges duplicates and resolves 774 * indirect refs to their parent bytenr. 775 * When roots are found, they're added to the roots list 776 * 777 * FIXME some caching might speed things up 778 */ 779 static int find_parent_nodes(struct btrfs_trans_handle *trans, 780 struct btrfs_fs_info *fs_info, u64 bytenr, 781 u64 time_seq, struct ulist *refs, 782 struct ulist *roots, const u64 *extent_item_pos) 783 { 784 struct btrfs_key key; 785 struct btrfs_path *path; 786 struct btrfs_delayed_ref_root *delayed_refs = NULL; 787 struct btrfs_delayed_ref_head *head; 788 int info_level = 0; 789 int ret; 790 int search_commit_root = (trans == BTRFS_BACKREF_SEARCH_COMMIT_ROOT); 791 struct list_head prefs_delayed; 792 struct list_head prefs; 793 struct __prelim_ref *ref; 794 795 INIT_LIST_HEAD(&prefs); 796 INIT_LIST_HEAD(&prefs_delayed); 797 798 key.objectid = bytenr; 799 key.type = BTRFS_EXTENT_ITEM_KEY; 800 key.offset = (u64)-1; 801 802 path = btrfs_alloc_path(); 803 if (!path) 804 return -ENOMEM; 805 path->search_commit_root = !!search_commit_root; 806 807 /* 808 * grab both a lock on the path and a lock on the delayed ref head. 809 * We need both to get a consistent picture of how the refs look 810 * at a specified point in time 811 */ 812 again: 813 head = NULL; 814 815 ret = btrfs_search_slot(trans, fs_info->extent_root, &key, path, 0, 0); 816 if (ret < 0) 817 goto out; 818 BUG_ON(ret == 0); 819 820 if (trans != BTRFS_BACKREF_SEARCH_COMMIT_ROOT) { 821 /* 822 * look if there are updates for this ref queued and lock the 823 * head 824 */ 825 delayed_refs = &trans->transaction->delayed_refs; 826 spin_lock(&delayed_refs->lock); 827 head = btrfs_find_delayed_ref_head(trans, bytenr); 828 if (head) { 829 if (!mutex_trylock(&head->mutex)) { 830 atomic_inc(&head->node.refs); 831 spin_unlock(&delayed_refs->lock); 832 833 btrfs_release_path(path); 834 835 /* 836 * Mutex was contended, block until it's 837 * released and try again 838 */ 839 mutex_lock(&head->mutex); 840 mutex_unlock(&head->mutex); 841 btrfs_put_delayed_ref(&head->node); 842 goto again; 843 } 844 ret = __add_delayed_refs(head, time_seq, 845 &prefs_delayed); 846 mutex_unlock(&head->mutex); 847 if (ret) { 848 spin_unlock(&delayed_refs->lock); 849 goto out; 850 } 851 } 852 spin_unlock(&delayed_refs->lock); 853 } 854 855 if (path->slots[0]) { 856 struct extent_buffer *leaf; 857 int slot; 858 859 path->slots[0]--; 860 leaf = path->nodes[0]; 861 slot = path->slots[0]; 862 btrfs_item_key_to_cpu(leaf, &key, slot); 863 if (key.objectid == bytenr && 864 key.type == BTRFS_EXTENT_ITEM_KEY) { 865 ret = __add_inline_refs(fs_info, path, bytenr, 866 &info_level, &prefs); 867 if (ret) 868 goto out; 869 ret = __add_keyed_refs(fs_info, path, bytenr, 870 info_level, &prefs); 871 if (ret) 872 goto out; 873 } 874 } 875 btrfs_release_path(path); 876 877 list_splice_init(&prefs_delayed, &prefs); 878 879 ret = __add_missing_keys(fs_info, &prefs); 880 if (ret) 881 goto out; 882 883 ret = __merge_refs(&prefs, 1); 884 if (ret) 885 goto out; 886 887 ret = __resolve_indirect_refs(fs_info, search_commit_root, time_seq, 888 &prefs, extent_item_pos); 889 if (ret) 890 goto out; 891 892 ret = __merge_refs(&prefs, 2); 893 if (ret) 894 goto out; 895 896 while (!list_empty(&prefs)) { 897 ref = list_first_entry(&prefs, struct __prelim_ref, list); 898 list_del(&ref->list); 899 WARN_ON(ref->count < 0); 900 if (ref->count && ref->root_id && ref->parent == 0) { 901 /* no parent == root of tree */ 902 ret = ulist_add(roots, ref->root_id, 0, GFP_NOFS); 903 BUG_ON(ret < 0); 904 } 905 if (ref->count && ref->parent) { 906 struct extent_inode_elem *eie = NULL; 907 if (extent_item_pos && !ref->inode_list) { 908 u32 bsz; 909 struct extent_buffer *eb; 910 bsz = btrfs_level_size(fs_info->extent_root, 911 info_level); 912 eb = read_tree_block(fs_info->extent_root, 913 ref->parent, bsz, 0); 914 BUG_ON(!eb); 915 ret = find_extent_in_eb(eb, bytenr, 916 *extent_item_pos, &eie); 917 ref->inode_list = eie; 918 free_extent_buffer(eb); 919 } 920 ret = ulist_add_merge(refs, ref->parent, 921 (uintptr_t)ref->inode_list, 922 (u64 *)&eie, GFP_NOFS); 923 if (!ret && extent_item_pos) { 924 /* 925 * we've recorded that parent, so we must extend 926 * its inode list here 927 */ 928 BUG_ON(!eie); 929 while (eie->next) 930 eie = eie->next; 931 eie->next = ref->inode_list; 932 } 933 BUG_ON(ret < 0); 934 } 935 kfree(ref); 936 } 937 938 out: 939 btrfs_free_path(path); 940 while (!list_empty(&prefs)) { 941 ref = list_first_entry(&prefs, struct __prelim_ref, list); 942 list_del(&ref->list); 943 kfree(ref); 944 } 945 while (!list_empty(&prefs_delayed)) { 946 ref = list_first_entry(&prefs_delayed, struct __prelim_ref, 947 list); 948 list_del(&ref->list); 949 kfree(ref); 950 } 951 952 return ret; 953 } 954 955 static void free_leaf_list(struct ulist *blocks) 956 { 957 struct ulist_node *node = NULL; 958 struct extent_inode_elem *eie; 959 struct extent_inode_elem *eie_next; 960 struct ulist_iterator uiter; 961 962 ULIST_ITER_INIT(&uiter); 963 while ((node = ulist_next(blocks, &uiter))) { 964 if (!node->aux) 965 continue; 966 eie = (struct extent_inode_elem *)(uintptr_t)node->aux; 967 for (; eie; eie = eie_next) { 968 eie_next = eie->next; 969 kfree(eie); 970 } 971 node->aux = 0; 972 } 973 974 ulist_free(blocks); 975 } 976 977 /* 978 * Finds all leafs with a reference to the specified combination of bytenr and 979 * offset. key_list_head will point to a list of corresponding keys (caller must 980 * free each list element). The leafs will be stored in the leafs ulist, which 981 * must be freed with ulist_free. 982 * 983 * returns 0 on success, <0 on error 984 */ 985 static int btrfs_find_all_leafs(struct btrfs_trans_handle *trans, 986 struct btrfs_fs_info *fs_info, u64 bytenr, 987 u64 time_seq, struct ulist **leafs, 988 const u64 *extent_item_pos) 989 { 990 struct ulist *tmp; 991 int ret; 992 993 tmp = ulist_alloc(GFP_NOFS); 994 if (!tmp) 995 return -ENOMEM; 996 *leafs = ulist_alloc(GFP_NOFS); 997 if (!*leafs) { 998 ulist_free(tmp); 999 return -ENOMEM; 1000 } 1001 1002 ret = find_parent_nodes(trans, fs_info, bytenr, 1003 time_seq, *leafs, tmp, extent_item_pos); 1004 ulist_free(tmp); 1005 1006 if (ret < 0 && ret != -ENOENT) { 1007 free_leaf_list(*leafs); 1008 return ret; 1009 } 1010 1011 return 0; 1012 } 1013 1014 /* 1015 * walk all backrefs for a given extent to find all roots that reference this 1016 * extent. Walking a backref means finding all extents that reference this 1017 * extent and in turn walk the backrefs of those, too. Naturally this is a 1018 * recursive process, but here it is implemented in an iterative fashion: We 1019 * find all referencing extents for the extent in question and put them on a 1020 * list. In turn, we find all referencing extents for those, further appending 1021 * to the list. The way we iterate the list allows adding more elements after 1022 * the current while iterating. The process stops when we reach the end of the 1023 * list. Found roots are added to the roots list. 1024 * 1025 * returns 0 on success, < 0 on error. 1026 */ 1027 int btrfs_find_all_roots(struct btrfs_trans_handle *trans, 1028 struct btrfs_fs_info *fs_info, u64 bytenr, 1029 u64 time_seq, struct ulist **roots) 1030 { 1031 struct ulist *tmp; 1032 struct ulist_node *node = NULL; 1033 struct ulist_iterator uiter; 1034 int ret; 1035 1036 tmp = ulist_alloc(GFP_NOFS); 1037 if (!tmp) 1038 return -ENOMEM; 1039 *roots = ulist_alloc(GFP_NOFS); 1040 if (!*roots) { 1041 ulist_free(tmp); 1042 return -ENOMEM; 1043 } 1044 1045 ULIST_ITER_INIT(&uiter); 1046 while (1) { 1047 ret = find_parent_nodes(trans, fs_info, bytenr, 1048 time_seq, tmp, *roots, NULL); 1049 if (ret < 0 && ret != -ENOENT) { 1050 ulist_free(tmp); 1051 ulist_free(*roots); 1052 return ret; 1053 } 1054 node = ulist_next(tmp, &uiter); 1055 if (!node) 1056 break; 1057 bytenr = node->val; 1058 } 1059 1060 ulist_free(tmp); 1061 return 0; 1062 } 1063 1064 1065 static int __inode_info(u64 inum, u64 ioff, u8 key_type, 1066 struct btrfs_root *fs_root, struct btrfs_path *path, 1067 struct btrfs_key *found_key) 1068 { 1069 int ret; 1070 struct btrfs_key key; 1071 struct extent_buffer *eb; 1072 1073 key.type = key_type; 1074 key.objectid = inum; 1075 key.offset = ioff; 1076 1077 ret = btrfs_search_slot(NULL, fs_root, &key, path, 0, 0); 1078 if (ret < 0) 1079 return ret; 1080 1081 eb = path->nodes[0]; 1082 if (ret && path->slots[0] >= btrfs_header_nritems(eb)) { 1083 ret = btrfs_next_leaf(fs_root, path); 1084 if (ret) 1085 return ret; 1086 eb = path->nodes[0]; 1087 } 1088 1089 btrfs_item_key_to_cpu(eb, found_key, path->slots[0]); 1090 if (found_key->type != key.type || found_key->objectid != key.objectid) 1091 return 1; 1092 1093 return 0; 1094 } 1095 1096 /* 1097 * this makes the path point to (inum INODE_ITEM ioff) 1098 */ 1099 int inode_item_info(u64 inum, u64 ioff, struct btrfs_root *fs_root, 1100 struct btrfs_path *path) 1101 { 1102 struct btrfs_key key; 1103 return __inode_info(inum, ioff, BTRFS_INODE_ITEM_KEY, fs_root, path, 1104 &key); 1105 } 1106 1107 static int inode_ref_info(u64 inum, u64 ioff, struct btrfs_root *fs_root, 1108 struct btrfs_path *path, 1109 struct btrfs_key *found_key) 1110 { 1111 return __inode_info(inum, ioff, BTRFS_INODE_REF_KEY, fs_root, path, 1112 found_key); 1113 } 1114 1115 int btrfs_find_one_extref(struct btrfs_root *root, u64 inode_objectid, 1116 u64 start_off, struct btrfs_path *path, 1117 struct btrfs_inode_extref **ret_extref, 1118 u64 *found_off) 1119 { 1120 int ret, slot; 1121 struct btrfs_key key; 1122 struct btrfs_key found_key; 1123 struct btrfs_inode_extref *extref; 1124 struct extent_buffer *leaf; 1125 unsigned long ptr; 1126 1127 key.objectid = inode_objectid; 1128 btrfs_set_key_type(&key, BTRFS_INODE_EXTREF_KEY); 1129 key.offset = start_off; 1130 1131 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); 1132 if (ret < 0) 1133 return ret; 1134 1135 while (1) { 1136 leaf = path->nodes[0]; 1137 slot = path->slots[0]; 1138 if (slot >= btrfs_header_nritems(leaf)) { 1139 /* 1140 * If the item at offset is not found, 1141 * btrfs_search_slot will point us to the slot 1142 * where it should be inserted. In our case 1143 * that will be the slot directly before the 1144 * next INODE_REF_KEY_V2 item. In the case 1145 * that we're pointing to the last slot in a 1146 * leaf, we must move one leaf over. 1147 */ 1148 ret = btrfs_next_leaf(root, path); 1149 if (ret) { 1150 if (ret >= 1) 1151 ret = -ENOENT; 1152 break; 1153 } 1154 continue; 1155 } 1156 1157 btrfs_item_key_to_cpu(leaf, &found_key, slot); 1158 1159 /* 1160 * Check that we're still looking at an extended ref key for 1161 * this particular objectid. If we have different 1162 * objectid or type then there are no more to be found 1163 * in the tree and we can exit. 1164 */ 1165 ret = -ENOENT; 1166 if (found_key.objectid != inode_objectid) 1167 break; 1168 if (btrfs_key_type(&found_key) != BTRFS_INODE_EXTREF_KEY) 1169 break; 1170 1171 ret = 0; 1172 ptr = btrfs_item_ptr_offset(leaf, path->slots[0]); 1173 extref = (struct btrfs_inode_extref *)ptr; 1174 *ret_extref = extref; 1175 if (found_off) 1176 *found_off = found_key.offset; 1177 break; 1178 } 1179 1180 return ret; 1181 } 1182 1183 char *btrfs_ref_to_path(struct btrfs_root *fs_root, struct btrfs_path *path, 1184 u32 name_len, unsigned long name_off, 1185 struct extent_buffer *eb_in, u64 parent, 1186 char *dest, u32 size) 1187 { 1188 int slot; 1189 u64 next_inum; 1190 int ret; 1191 s64 bytes_left = ((s64)size) - 1; 1192 struct extent_buffer *eb = eb_in; 1193 struct btrfs_key found_key; 1194 int leave_spinning = path->leave_spinning; 1195 struct btrfs_inode_ref *iref; 1196 1197 if (bytes_left >= 0) 1198 dest[bytes_left] = '\0'; 1199 1200 path->leave_spinning = 1; 1201 while (1) { 1202 bytes_left -= name_len; 1203 if (bytes_left >= 0) 1204 read_extent_buffer(eb, dest + bytes_left, 1205 name_off, name_len); 1206 if (eb != eb_in) { 1207 btrfs_tree_read_unlock_blocking(eb); 1208 free_extent_buffer(eb); 1209 } 1210 ret = inode_ref_info(parent, 0, fs_root, path, &found_key); 1211 if (ret > 0) 1212 ret = -ENOENT; 1213 if (ret) 1214 break; 1215 1216 next_inum = found_key.offset; 1217 1218 /* regular exit ahead */ 1219 if (parent == next_inum) 1220 break; 1221 1222 slot = path->slots[0]; 1223 eb = path->nodes[0]; 1224 /* make sure we can use eb after releasing the path */ 1225 if (eb != eb_in) { 1226 atomic_inc(&eb->refs); 1227 btrfs_tree_read_lock(eb); 1228 btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK); 1229 } 1230 btrfs_release_path(path); 1231 iref = btrfs_item_ptr(eb, slot, struct btrfs_inode_ref); 1232 1233 name_len = btrfs_inode_ref_name_len(eb, iref); 1234 name_off = (unsigned long)(iref + 1); 1235 1236 parent = next_inum; 1237 --bytes_left; 1238 if (bytes_left >= 0) 1239 dest[bytes_left] = '/'; 1240 } 1241 1242 btrfs_release_path(path); 1243 path->leave_spinning = leave_spinning; 1244 1245 if (ret) 1246 return ERR_PTR(ret); 1247 1248 return dest + bytes_left; 1249 } 1250 1251 /* 1252 * this iterates to turn a btrfs_inode_ref into a full filesystem path. elements 1253 * of the path are separated by '/' and the path is guaranteed to be 1254 * 0-terminated. the path is only given within the current file system. 1255 * Therefore, it never starts with a '/'. the caller is responsible to provide 1256 * "size" bytes in "dest". the dest buffer will be filled backwards. finally, 1257 * the start point of the resulting string is returned. this pointer is within 1258 * dest, normally. 1259 * in case the path buffer would overflow, the pointer is decremented further 1260 * as if output was written to the buffer, though no more output is actually 1261 * generated. that way, the caller can determine how much space would be 1262 * required for the path to fit into the buffer. in that case, the returned 1263 * value will be smaller than dest. callers must check this! 1264 */ 1265 char *btrfs_iref_to_path(struct btrfs_root *fs_root, 1266 struct btrfs_path *path, 1267 struct btrfs_inode_ref *iref, 1268 struct extent_buffer *eb_in, u64 parent, 1269 char *dest, u32 size) 1270 { 1271 return btrfs_ref_to_path(fs_root, path, 1272 btrfs_inode_ref_name_len(eb_in, iref), 1273 (unsigned long)(iref + 1), 1274 eb_in, parent, dest, size); 1275 } 1276 1277 /* 1278 * this makes the path point to (logical EXTENT_ITEM *) 1279 * returns BTRFS_EXTENT_FLAG_DATA for data, BTRFS_EXTENT_FLAG_TREE_BLOCK for 1280 * tree blocks and <0 on error. 1281 */ 1282 int extent_from_logical(struct btrfs_fs_info *fs_info, u64 logical, 1283 struct btrfs_path *path, struct btrfs_key *found_key, 1284 u64 *flags_ret) 1285 { 1286 int ret; 1287 u64 flags; 1288 u32 item_size; 1289 struct extent_buffer *eb; 1290 struct btrfs_extent_item *ei; 1291 struct btrfs_key key; 1292 1293 key.type = BTRFS_EXTENT_ITEM_KEY; 1294 key.objectid = logical; 1295 key.offset = (u64)-1; 1296 1297 ret = btrfs_search_slot(NULL, fs_info->extent_root, &key, path, 0, 0); 1298 if (ret < 0) 1299 return ret; 1300 ret = btrfs_previous_item(fs_info->extent_root, path, 1301 0, BTRFS_EXTENT_ITEM_KEY); 1302 if (ret < 0) 1303 return ret; 1304 1305 btrfs_item_key_to_cpu(path->nodes[0], found_key, path->slots[0]); 1306 if (found_key->type != BTRFS_EXTENT_ITEM_KEY || 1307 found_key->objectid > logical || 1308 found_key->objectid + found_key->offset <= logical) { 1309 pr_debug("logical %llu is not within any extent\n", 1310 (unsigned long long)logical); 1311 return -ENOENT; 1312 } 1313 1314 eb = path->nodes[0]; 1315 item_size = btrfs_item_size_nr(eb, path->slots[0]); 1316 BUG_ON(item_size < sizeof(*ei)); 1317 1318 ei = btrfs_item_ptr(eb, path->slots[0], struct btrfs_extent_item); 1319 flags = btrfs_extent_flags(eb, ei); 1320 1321 pr_debug("logical %llu is at position %llu within the extent (%llu " 1322 "EXTENT_ITEM %llu) flags %#llx size %u\n", 1323 (unsigned long long)logical, 1324 (unsigned long long)(logical - found_key->objectid), 1325 (unsigned long long)found_key->objectid, 1326 (unsigned long long)found_key->offset, 1327 (unsigned long long)flags, item_size); 1328 1329 WARN_ON(!flags_ret); 1330 if (flags_ret) { 1331 if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) 1332 *flags_ret = BTRFS_EXTENT_FLAG_TREE_BLOCK; 1333 else if (flags & BTRFS_EXTENT_FLAG_DATA) 1334 *flags_ret = BTRFS_EXTENT_FLAG_DATA; 1335 else 1336 BUG_ON(1); 1337 return 0; 1338 } 1339 1340 return -EIO; 1341 } 1342 1343 /* 1344 * helper function to iterate extent inline refs. ptr must point to a 0 value 1345 * for the first call and may be modified. it is used to track state. 1346 * if more refs exist, 0 is returned and the next call to 1347 * __get_extent_inline_ref must pass the modified ptr parameter to get the 1348 * next ref. after the last ref was processed, 1 is returned. 1349 * returns <0 on error 1350 */ 1351 static int __get_extent_inline_ref(unsigned long *ptr, struct extent_buffer *eb, 1352 struct btrfs_extent_item *ei, u32 item_size, 1353 struct btrfs_extent_inline_ref **out_eiref, 1354 int *out_type) 1355 { 1356 unsigned long end; 1357 u64 flags; 1358 struct btrfs_tree_block_info *info; 1359 1360 if (!*ptr) { 1361 /* first call */ 1362 flags = btrfs_extent_flags(eb, ei); 1363 if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) { 1364 info = (struct btrfs_tree_block_info *)(ei + 1); 1365 *out_eiref = 1366 (struct btrfs_extent_inline_ref *)(info + 1); 1367 } else { 1368 *out_eiref = (struct btrfs_extent_inline_ref *)(ei + 1); 1369 } 1370 *ptr = (unsigned long)*out_eiref; 1371 if ((void *)*ptr >= (void *)ei + item_size) 1372 return -ENOENT; 1373 } 1374 1375 end = (unsigned long)ei + item_size; 1376 *out_eiref = (struct btrfs_extent_inline_ref *)*ptr; 1377 *out_type = btrfs_extent_inline_ref_type(eb, *out_eiref); 1378 1379 *ptr += btrfs_extent_inline_ref_size(*out_type); 1380 WARN_ON(*ptr > end); 1381 if (*ptr == end) 1382 return 1; /* last */ 1383 1384 return 0; 1385 } 1386 1387 /* 1388 * reads the tree block backref for an extent. tree level and root are returned 1389 * through out_level and out_root. ptr must point to a 0 value for the first 1390 * call and may be modified (see __get_extent_inline_ref comment). 1391 * returns 0 if data was provided, 1 if there was no more data to provide or 1392 * <0 on error. 1393 */ 1394 int tree_backref_for_extent(unsigned long *ptr, struct extent_buffer *eb, 1395 struct btrfs_extent_item *ei, u32 item_size, 1396 u64 *out_root, u8 *out_level) 1397 { 1398 int ret; 1399 int type; 1400 struct btrfs_tree_block_info *info; 1401 struct btrfs_extent_inline_ref *eiref; 1402 1403 if (*ptr == (unsigned long)-1) 1404 return 1; 1405 1406 while (1) { 1407 ret = __get_extent_inline_ref(ptr, eb, ei, item_size, 1408 &eiref, &type); 1409 if (ret < 0) 1410 return ret; 1411 1412 if (type == BTRFS_TREE_BLOCK_REF_KEY || 1413 type == BTRFS_SHARED_BLOCK_REF_KEY) 1414 break; 1415 1416 if (ret == 1) 1417 return 1; 1418 } 1419 1420 /* we can treat both ref types equally here */ 1421 info = (struct btrfs_tree_block_info *)(ei + 1); 1422 *out_root = btrfs_extent_inline_ref_offset(eb, eiref); 1423 *out_level = btrfs_tree_block_level(eb, info); 1424 1425 if (ret == 1) 1426 *ptr = (unsigned long)-1; 1427 1428 return 0; 1429 } 1430 1431 static int iterate_leaf_refs(struct extent_inode_elem *inode_list, 1432 u64 root, u64 extent_item_objectid, 1433 iterate_extent_inodes_t *iterate, void *ctx) 1434 { 1435 struct extent_inode_elem *eie; 1436 int ret = 0; 1437 1438 for (eie = inode_list; eie; eie = eie->next) { 1439 pr_debug("ref for %llu resolved, key (%llu EXTEND_DATA %llu), " 1440 "root %llu\n", extent_item_objectid, 1441 eie->inum, eie->offset, root); 1442 ret = iterate(eie->inum, eie->offset, root, ctx); 1443 if (ret) { 1444 pr_debug("stopping iteration for %llu due to ret=%d\n", 1445 extent_item_objectid, ret); 1446 break; 1447 } 1448 } 1449 1450 return ret; 1451 } 1452 1453 /* 1454 * calls iterate() for every inode that references the extent identified by 1455 * the given parameters. 1456 * when the iterator function returns a non-zero value, iteration stops. 1457 */ 1458 int iterate_extent_inodes(struct btrfs_fs_info *fs_info, 1459 u64 extent_item_objectid, u64 extent_item_pos, 1460 int search_commit_root, 1461 iterate_extent_inodes_t *iterate, void *ctx) 1462 { 1463 int ret; 1464 struct list_head data_refs = LIST_HEAD_INIT(data_refs); 1465 struct list_head shared_refs = LIST_HEAD_INIT(shared_refs); 1466 struct btrfs_trans_handle *trans; 1467 struct ulist *refs = NULL; 1468 struct ulist *roots = NULL; 1469 struct ulist_node *ref_node = NULL; 1470 struct ulist_node *root_node = NULL; 1471 struct seq_list tree_mod_seq_elem = {}; 1472 struct ulist_iterator ref_uiter; 1473 struct ulist_iterator root_uiter; 1474 1475 pr_debug("resolving all inodes for extent %llu\n", 1476 extent_item_objectid); 1477 1478 if (search_commit_root) { 1479 trans = BTRFS_BACKREF_SEARCH_COMMIT_ROOT; 1480 } else { 1481 trans = btrfs_join_transaction(fs_info->extent_root); 1482 if (IS_ERR(trans)) 1483 return PTR_ERR(trans); 1484 btrfs_get_tree_mod_seq(fs_info, &tree_mod_seq_elem); 1485 } 1486 1487 ret = btrfs_find_all_leafs(trans, fs_info, extent_item_objectid, 1488 tree_mod_seq_elem.seq, &refs, 1489 &extent_item_pos); 1490 if (ret) 1491 goto out; 1492 1493 ULIST_ITER_INIT(&ref_uiter); 1494 while (!ret && (ref_node = ulist_next(refs, &ref_uiter))) { 1495 ret = btrfs_find_all_roots(trans, fs_info, ref_node->val, 1496 tree_mod_seq_elem.seq, &roots); 1497 if (ret) 1498 break; 1499 ULIST_ITER_INIT(&root_uiter); 1500 while (!ret && (root_node = ulist_next(roots, &root_uiter))) { 1501 pr_debug("root %llu references leaf %llu, data list " 1502 "%#llx\n", root_node->val, ref_node->val, 1503 (long long)ref_node->aux); 1504 ret = iterate_leaf_refs((struct extent_inode_elem *) 1505 (uintptr_t)ref_node->aux, 1506 root_node->val, 1507 extent_item_objectid, 1508 iterate, ctx); 1509 } 1510 ulist_free(roots); 1511 roots = NULL; 1512 } 1513 1514 free_leaf_list(refs); 1515 ulist_free(roots); 1516 out: 1517 if (!search_commit_root) { 1518 btrfs_put_tree_mod_seq(fs_info, &tree_mod_seq_elem); 1519 btrfs_end_transaction(trans, fs_info->extent_root); 1520 } 1521 1522 return ret; 1523 } 1524 1525 int iterate_inodes_from_logical(u64 logical, struct btrfs_fs_info *fs_info, 1526 struct btrfs_path *path, 1527 iterate_extent_inodes_t *iterate, void *ctx) 1528 { 1529 int ret; 1530 u64 extent_item_pos; 1531 u64 flags = 0; 1532 struct btrfs_key found_key; 1533 int search_commit_root = path->search_commit_root; 1534 1535 ret = extent_from_logical(fs_info, logical, path, &found_key, &flags); 1536 btrfs_release_path(path); 1537 if (ret < 0) 1538 return ret; 1539 if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) 1540 return -EINVAL; 1541 1542 extent_item_pos = logical - found_key.objectid; 1543 ret = iterate_extent_inodes(fs_info, found_key.objectid, 1544 extent_item_pos, search_commit_root, 1545 iterate, ctx); 1546 1547 return ret; 1548 } 1549 1550 typedef int (iterate_irefs_t)(u64 parent, u32 name_len, unsigned long name_off, 1551 struct extent_buffer *eb, void *ctx); 1552 1553 static int iterate_inode_refs(u64 inum, struct btrfs_root *fs_root, 1554 struct btrfs_path *path, 1555 iterate_irefs_t *iterate, void *ctx) 1556 { 1557 int ret = 0; 1558 int slot; 1559 u32 cur; 1560 u32 len; 1561 u32 name_len; 1562 u64 parent = 0; 1563 int found = 0; 1564 struct extent_buffer *eb; 1565 struct btrfs_item *item; 1566 struct btrfs_inode_ref *iref; 1567 struct btrfs_key found_key; 1568 1569 while (!ret) { 1570 path->leave_spinning = 1; 1571 ret = inode_ref_info(inum, parent ? parent+1 : 0, fs_root, path, 1572 &found_key); 1573 if (ret < 0) 1574 break; 1575 if (ret) { 1576 ret = found ? 0 : -ENOENT; 1577 break; 1578 } 1579 ++found; 1580 1581 parent = found_key.offset; 1582 slot = path->slots[0]; 1583 eb = path->nodes[0]; 1584 /* make sure we can use eb after releasing the path */ 1585 atomic_inc(&eb->refs); 1586 btrfs_tree_read_lock(eb); 1587 btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK); 1588 btrfs_release_path(path); 1589 1590 item = btrfs_item_nr(eb, slot); 1591 iref = btrfs_item_ptr(eb, slot, struct btrfs_inode_ref); 1592 1593 for (cur = 0; cur < btrfs_item_size(eb, item); cur += len) { 1594 name_len = btrfs_inode_ref_name_len(eb, iref); 1595 /* path must be released before calling iterate()! */ 1596 pr_debug("following ref at offset %u for inode %llu in " 1597 "tree %llu\n", cur, 1598 (unsigned long long)found_key.objectid, 1599 (unsigned long long)fs_root->objectid); 1600 ret = iterate(parent, name_len, 1601 (unsigned long)(iref + 1), eb, ctx); 1602 if (ret) 1603 break; 1604 len = sizeof(*iref) + name_len; 1605 iref = (struct btrfs_inode_ref *)((char *)iref + len); 1606 } 1607 btrfs_tree_read_unlock_blocking(eb); 1608 free_extent_buffer(eb); 1609 } 1610 1611 btrfs_release_path(path); 1612 1613 return ret; 1614 } 1615 1616 static int iterate_inode_extrefs(u64 inum, struct btrfs_root *fs_root, 1617 struct btrfs_path *path, 1618 iterate_irefs_t *iterate, void *ctx) 1619 { 1620 int ret; 1621 int slot; 1622 u64 offset = 0; 1623 u64 parent; 1624 int found = 0; 1625 struct extent_buffer *eb; 1626 struct btrfs_inode_extref *extref; 1627 struct extent_buffer *leaf; 1628 u32 item_size; 1629 u32 cur_offset; 1630 unsigned long ptr; 1631 1632 while (1) { 1633 ret = btrfs_find_one_extref(fs_root, inum, offset, path, &extref, 1634 &offset); 1635 if (ret < 0) 1636 break; 1637 if (ret) { 1638 ret = found ? 0 : -ENOENT; 1639 break; 1640 } 1641 ++found; 1642 1643 slot = path->slots[0]; 1644 eb = path->nodes[0]; 1645 /* make sure we can use eb after releasing the path */ 1646 atomic_inc(&eb->refs); 1647 1648 btrfs_tree_read_lock(eb); 1649 btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK); 1650 btrfs_release_path(path); 1651 1652 leaf = path->nodes[0]; 1653 item_size = btrfs_item_size_nr(leaf, path->slots[0]); 1654 ptr = btrfs_item_ptr_offset(leaf, path->slots[0]); 1655 cur_offset = 0; 1656 1657 while (cur_offset < item_size) { 1658 u32 name_len; 1659 1660 extref = (struct btrfs_inode_extref *)(ptr + cur_offset); 1661 parent = btrfs_inode_extref_parent(eb, extref); 1662 name_len = btrfs_inode_extref_name_len(eb, extref); 1663 ret = iterate(parent, name_len, 1664 (unsigned long)&extref->name, eb, ctx); 1665 if (ret) 1666 break; 1667 1668 cur_offset += btrfs_inode_extref_name_len(leaf, extref); 1669 cur_offset += sizeof(*extref); 1670 } 1671 btrfs_tree_read_unlock_blocking(eb); 1672 free_extent_buffer(eb); 1673 1674 offset++; 1675 } 1676 1677 btrfs_release_path(path); 1678 1679 return ret; 1680 } 1681 1682 static int iterate_irefs(u64 inum, struct btrfs_root *fs_root, 1683 struct btrfs_path *path, iterate_irefs_t *iterate, 1684 void *ctx) 1685 { 1686 int ret; 1687 int found_refs = 0; 1688 1689 ret = iterate_inode_refs(inum, fs_root, path, iterate, ctx); 1690 if (!ret) 1691 ++found_refs; 1692 else if (ret != -ENOENT) 1693 return ret; 1694 1695 ret = iterate_inode_extrefs(inum, fs_root, path, iterate, ctx); 1696 if (ret == -ENOENT && found_refs) 1697 return 0; 1698 1699 return ret; 1700 } 1701 1702 /* 1703 * returns 0 if the path could be dumped (probably truncated) 1704 * returns <0 in case of an error 1705 */ 1706 static int inode_to_path(u64 inum, u32 name_len, unsigned long name_off, 1707 struct extent_buffer *eb, void *ctx) 1708 { 1709 struct inode_fs_paths *ipath = ctx; 1710 char *fspath; 1711 char *fspath_min; 1712 int i = ipath->fspath->elem_cnt; 1713 const int s_ptr = sizeof(char *); 1714 u32 bytes_left; 1715 1716 bytes_left = ipath->fspath->bytes_left > s_ptr ? 1717 ipath->fspath->bytes_left - s_ptr : 0; 1718 1719 fspath_min = (char *)ipath->fspath->val + (i + 1) * s_ptr; 1720 fspath = btrfs_ref_to_path(ipath->fs_root, ipath->btrfs_path, name_len, 1721 name_off, eb, inum, fspath_min, bytes_left); 1722 if (IS_ERR(fspath)) 1723 return PTR_ERR(fspath); 1724 1725 if (fspath > fspath_min) { 1726 ipath->fspath->val[i] = (u64)(unsigned long)fspath; 1727 ++ipath->fspath->elem_cnt; 1728 ipath->fspath->bytes_left = fspath - fspath_min; 1729 } else { 1730 ++ipath->fspath->elem_missed; 1731 ipath->fspath->bytes_missing += fspath_min - fspath; 1732 ipath->fspath->bytes_left = 0; 1733 } 1734 1735 return 0; 1736 } 1737 1738 /* 1739 * this dumps all file system paths to the inode into the ipath struct, provided 1740 * is has been created large enough. each path is zero-terminated and accessed 1741 * from ipath->fspath->val[i]. 1742 * when it returns, there are ipath->fspath->elem_cnt number of paths available 1743 * in ipath->fspath->val[]. when the allocated space wasn't sufficient, the 1744 * number of missed paths in recored in ipath->fspath->elem_missed, otherwise, 1745 * it's zero. ipath->fspath->bytes_missing holds the number of bytes that would 1746 * have been needed to return all paths. 1747 */ 1748 int paths_from_inode(u64 inum, struct inode_fs_paths *ipath) 1749 { 1750 return iterate_irefs(inum, ipath->fs_root, ipath->btrfs_path, 1751 inode_to_path, ipath); 1752 } 1753 1754 struct btrfs_data_container *init_data_container(u32 total_bytes) 1755 { 1756 struct btrfs_data_container *data; 1757 size_t alloc_bytes; 1758 1759 alloc_bytes = max_t(size_t, total_bytes, sizeof(*data)); 1760 data = vmalloc(alloc_bytes); 1761 if (!data) 1762 return ERR_PTR(-ENOMEM); 1763 1764 if (total_bytes >= sizeof(*data)) { 1765 data->bytes_left = total_bytes - sizeof(*data); 1766 data->bytes_missing = 0; 1767 } else { 1768 data->bytes_missing = sizeof(*data) - total_bytes; 1769 data->bytes_left = 0; 1770 } 1771 1772 data->elem_cnt = 0; 1773 data->elem_missed = 0; 1774 1775 return data; 1776 } 1777 1778 /* 1779 * allocates space to return multiple file system paths for an inode. 1780 * total_bytes to allocate are passed, note that space usable for actual path 1781 * information will be total_bytes - sizeof(struct inode_fs_paths). 1782 * the returned pointer must be freed with free_ipath() in the end. 1783 */ 1784 struct inode_fs_paths *init_ipath(s32 total_bytes, struct btrfs_root *fs_root, 1785 struct btrfs_path *path) 1786 { 1787 struct inode_fs_paths *ifp; 1788 struct btrfs_data_container *fspath; 1789 1790 fspath = init_data_container(total_bytes); 1791 if (IS_ERR(fspath)) 1792 return (void *)fspath; 1793 1794 ifp = kmalloc(sizeof(*ifp), GFP_NOFS); 1795 if (!ifp) { 1796 kfree(fspath); 1797 return ERR_PTR(-ENOMEM); 1798 } 1799 1800 ifp->btrfs_path = path; 1801 ifp->fspath = fspath; 1802 ifp->fs_root = fs_root; 1803 1804 return ifp; 1805 } 1806 1807 void free_ipath(struct inode_fs_paths *ipath) 1808 { 1809 if (!ipath) 1810 return; 1811 vfree(ipath->fspath); 1812 kfree(ipath); 1813 } 1814