fs/btrfs/backref.c

   1 // SPDX-License-Identifier: GPL-2.0
   2 /*
   3  * Copyright (C) 2011 STRATO.  All rights reserved.
   4  */
   5
   6 #include <linux/mm.h>
   7 #include <linux/rbtree.h>
   8 #include <trace/events/btrfs.h>
   9 #include "ctree.h"
  10 #include "disk-io.h"
  11 #include "backref.h"
  12 #include "ulist.h"
  13 #include "transaction.h"
  14 #include "delayed-ref.h"
  15 #include "locking.h"
  16
  17 /* Just an arbitrary number so we can be sure this happened */
  18 #define BACKREF_FOUND_SHARED 6
  19
  20 struct extent_inode_elem {
  21         u64 inum;
  22         u64 offset;
  23         struct extent_inode_elem *next;
  24 };
  25
  26 static int check_extent_in_eb(const struct btrfs_key *key,
  27                               const struct extent_buffer *eb,
  28                               const struct btrfs_file_extent_item *fi,
  29                               u64 extent_item_pos,
  30                               struct extent_inode_elem **eie,
  31                               bool ignore_offset)
  32 {
  33         u64 offset = 0;
  34         struct extent_inode_elem *e;
  35
  36         if (!ignore_offset &&
  37             !btrfs_file_extent_compression(eb, fi) &&
  38             !btrfs_file_extent_encryption(eb, fi) &&
  39             !btrfs_file_extent_other_encoding(eb, fi)) {
  40                 u64 data_offset;
  41                 u64 data_len;
  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                 offset = extent_item_pos - data_offset;
  50         }
  51
  52         e = kmalloc(sizeof(*e), GFP_NOFS);
  53         if (!e)
  54                 return -ENOMEM;
  55
  56         e->next = *eie;
  57         e->inum = key->objectid;
  58         e->offset = key->offset + offset;
  59         *eie = e;
  60
  61         return 0;
  62 }
  63
  64 static void free_inode_elem_list(struct extent_inode_elem *eie)
  65 {
  66         struct extent_inode_elem *eie_next;
  67
  68         for (; eie; eie = eie_next) {
  69                 eie_next = eie->next;
  70                 kfree(eie);
  71         }
  72 }
  73
  74 static int find_extent_in_eb(const struct extent_buffer *eb,
  75                              u64 wanted_disk_byte, u64 extent_item_pos,
  76                              struct extent_inode_elem **eie,
  77                              bool ignore_offset)
  78 {
  79         u64 disk_byte;
  80         struct btrfs_key key;
  81         struct btrfs_file_extent_item *fi;
  82         int slot;
  83         int nritems;
  84         int extent_type;
  85         int ret;
  86
  87         /*
  88          * from the shared data ref, we only have the leaf but we need
  89          * the key. thus, we must look into all items and see that we
  90          * find one (some) with a reference to our extent item.
  91          */
  92         nritems = btrfs_header_nritems(eb);
  93         for (slot = 0; slot < nritems; ++slot) {
  94                 btrfs_item_key_to_cpu(eb, &key, slot);
  95                 if (key.type != BTRFS_EXTENT_DATA_KEY)
  96                         continue;
  97                 fi = btrfs_item_ptr(eb, slot, struct btrfs_file_extent_item);
  98                 extent_type = btrfs_file_extent_type(eb, fi);
  99                 if (extent_type == BTRFS_FILE_EXTENT_INLINE)
 100                         continue;
 101                 /* don't skip BTRFS_FILE_EXTENT_PREALLOC, we can handle that */
 102                 disk_byte = btrfs_file_extent_disk_bytenr(eb, fi);
 103                 if (disk_byte != wanted_disk_byte)
 104                         continue;
 105
 106                 ret = check_extent_in_eb(&key, eb, fi, extent_item_pos, eie, ignore_offset);
 107                 if (ret < 0)
 108                         return ret;
 109         }
 110
 111         return 0;
 112 }
 113
 114 struct preftree {
 115         struct rb_root_cached root;
 116         unsigned int count;
 117 };
 118
 119 #define PREFTREE_INIT   { .root = RB_ROOT_CACHED, .count = 0 }
 120
 121 struct preftrees {
 122         struct preftree direct;    /* BTRFS_SHARED_[DATA|BLOCK]_REF_KEY */
 123         struct preftree indirect;  /* BTRFS_[TREE_BLOCK|EXTENT_DATA]_REF_KEY */
 124         struct preftree indirect_missing_keys;
 125 };
 126
 127 /*
 128  * Checks for a shared extent during backref search.
 129  *
 130  * The share_count tracks prelim_refs (direct and indirect) having a
 131  * ref->count >0:
 132  *  - incremented when a ref->count transitions to >0
 133  *  - decremented when a ref->count transitions to <1
 134  */
 135 struct share_check {
 136         u64 root_objectid;
 137         u64 inum;
 138         int share_count;
 139 };
 140
 141 static inline int extent_is_shared(struct share_check *sc)
 142 {
 143         return (sc && sc->share_count > 1) ? BACKREF_FOUND_SHARED : 0;
 144 }
 145
 146 static struct kmem_cache *btrfs_prelim_ref_cache;
 147
 148 int __init btrfs_prelim_ref_init(void)
 149 {
 150         btrfs_prelim_ref_cache = kmem_cache_create("btrfs_prelim_ref",
 151                                         sizeof(struct prelim_ref),
 152                                         0,
 153                                         SLAB_MEM_SPREAD,
 154                                         NULL);
 155         if (!btrfs_prelim_ref_cache)
 156                 return -ENOMEM;
 157         return 0;
 158 }
 159
 160 void __cold btrfs_prelim_ref_exit(void)
 161 {
 162         kmem_cache_destroy(btrfs_prelim_ref_cache);
 163 }
 164
 165 static void free_pref(struct prelim_ref *ref)
 166 {
 167         kmem_cache_free(btrfs_prelim_ref_cache, ref);
 168 }
 169
 170 /*
 171  * Return 0 when both refs are for the same block (and can be merged).
 172  * A -1 return indicates ref1 is a 'lower' block than ref2, while 1
 173  * indicates a 'higher' block.
 174  */
 175 static int prelim_ref_compare(struct prelim_ref *ref1,
 176                               struct prelim_ref *ref2)
 177 {
 178         if (ref1->level < ref2->level)
 179                 return -1;
 180         if (ref1->level > ref2->level)
 181                 return 1;
 182         if (ref1->root_id < ref2->root_id)
 183                 return -1;
 184         if (ref1->root_id > ref2->root_id)
 185                 return 1;
 186         if (ref1->key_for_search.type < ref2->key_for_search.type)
 187                 return -1;
 188         if (ref1->key_for_search.type > ref2->key_for_search.type)
 189                 return 1;
 190         if (ref1->key_for_search.objectid < ref2->key_for_search.objectid)
 191                 return -1;
 192         if (ref1->key_for_search.objectid > ref2->key_for_search.objectid)
 193                 return 1;
 194         if (ref1->key_for_search.offset < ref2->key_for_search.offset)
 195                 return -1;
 196         if (ref1->key_for_search.offset > ref2->key_for_search.offset)
 197                 return 1;
 198         if (ref1->parent < ref2->parent)
 199                 return -1;
 200         if (ref1->parent > ref2->parent)
 201                 return 1;
 202
 203         return 0;
 204 }
 205
 206 static void update_share_count(struct share_check *sc, int oldcount,
 207                                int newcount)
 208 {
 209         if ((!sc) || (oldcount == 0 && newcount < 1))
 210                 return;
 211
 212         if (oldcount > 0 && newcount < 1)
 213                 sc->share_count--;
 214         else if (oldcount < 1 && newcount > 0)
 215                 sc->share_count++;
 216 }
 217
 218 /*
 219  * Add @newref to the @root rbtree, merging identical refs.
 220  *
 221  * Callers should assume that newref has been freed after calling.
 222  */
 223 static void prelim_ref_insert(const struct btrfs_fs_info *fs_info,
 224                               struct preftree *preftree,
 225                               struct prelim_ref *newref,
 226                               struct share_check *sc)
 227 {
 228         struct rb_root_cached *root;
 229         struct rb_node **p;
 230         struct rb_node *parent = NULL;
 231         struct prelim_ref *ref;
 232         int result;
 233         bool leftmost = true;
 234
 235         root = &preftree->root;
 236         p = &root->rb_root.rb_node;
 237
 238         while (*p) {
 239                 parent = *p;
 240                 ref = rb_entry(parent, struct prelim_ref, rbnode);
 241                 result = prelim_ref_compare(ref, newref);
 242                 if (result < 0) {
 243                         p = &(*p)->rb_left;
 244                 } else if (result > 0) {
 245                         p = &(*p)->rb_right;
 246                         leftmost = false;
 247                 } else {
 248                         /* Identical refs, merge them and free @newref */
 249                         struct extent_inode_elem *eie = ref->inode_list;
 250
 251                         while (eie && eie->next)
 252                                 eie = eie->next;
 253
 254                         if (!eie)
 255                                 ref->inode_list = newref->inode_list;
 256                         else
 257                                 eie->next = newref->inode_list;
 258                         trace_btrfs_prelim_ref_merge(fs_info, ref, newref,
 259                                                      preftree->count);
 260                         /*
 261                          * A delayed ref can have newref->count < 0.
 262                          * The ref->count is updated to follow any
 263                          * BTRFS_[ADD|DROP]_DELAYED_REF actions.
 264                          */
 265                         update_share_count(sc, ref->count,
 266                                            ref->count + newref->count);
 267                         ref->count += newref->count;
 268                         free_pref(newref);
 269                         return;
 270                 }
 271         }
 272
 273         update_share_count(sc, 0, newref->count);
 274         preftree->count++;
 275         trace_btrfs_prelim_ref_insert(fs_info, newref, NULL, preftree->count);
 276         rb_link_node(&newref->rbnode, parent, p);
 277         rb_insert_color_cached(&newref->rbnode, root, leftmost);
 278 }
 279
 280 /*
 281  * Release the entire tree.  We don't care about internal consistency so
 282  * just free everything and then reset the tree root.
 283  */
 284 static void prelim_release(struct preftree *preftree)
 285 {
 286         struct prelim_ref *ref, *next_ref;
 287
 288         rbtree_postorder_for_each_entry_safe(ref, next_ref,
 289                                              &preftree->root.rb_root, rbnode)
 290                 free_pref(ref);
 291
 292         preftree->root = RB_ROOT_CACHED;
 293         preftree->count = 0;
 294 }
 295
 296 /*
 297  * the rules for all callers of this function are:
 298  * - obtaining the parent is the goal
 299  * - if you add a key, you must know that it is a correct key
 300  * - if you cannot add the parent or a correct key, then we will look into the
 301  *   block later to set a correct key
 302  *
 303  * delayed refs
 304  * ============
 305  *        backref type | shared | indirect | shared | indirect
 306  * information         |   tree |     tree |   data |     data
 307  * --------------------+--------+----------+--------+----------
 308  *      parent logical |    y   |     -    |    -   |     -
 309  *      key to resolve |    -   |     y    |    y   |     y
 310  *  tree block logical |    -   |     -    |    -   |     -
 311  *  root for resolving |    y   |     y    |    y   |     y
 312  *
 313  * - column 1:       we've the parent -> done
 314  * - column 2, 3, 4: we use the key to find the parent
 315  *
 316  * on disk refs (inline or keyed)
 317  * ==============================
 318  *        backref type | shared | indirect | shared | indirect
 319  * information         |   tree |     tree |   data |     data
 320  * --------------------+--------+----------+--------+----------
 321  *      parent logical |    y   |     -    |    y   |     -
 322  *      key to resolve |    -   |     -    |    -   |     y
 323  *  tree block logical |    y   |     y    |    y   |     y
 324  *  root for resolving |    -   |     y    |    y   |     y
 325  *
 326  * - column 1, 3: we've the parent -> done
 327  * - column 2:    we take the first key from the block to find the parent
 328  *                (see add_missing_keys)
 329  * - column 4:    we use the key to find the parent
 330  *
 331  * additional information that's available but not required to find the parent
 332  * block might help in merging entries to gain some speed.
 333  */
 334 static int add_prelim_ref(const struct btrfs_fs_info *fs_info,
 335                           struct preftree *preftree, u64 root_id,
 336                           const struct btrfs_key *key, int level, u64 parent,
 337                           u64 wanted_disk_byte, int count,
 338                           struct share_check *sc, gfp_t gfp_mask)
 339 {
 340         struct prelim_ref *ref;
 341
 342         if (root_id == BTRFS_DATA_RELOC_TREE_OBJECTID)
 343                 return 0;
 344
 345         ref = kmem_cache_alloc(btrfs_prelim_ref_cache, gfp_mask);
 346         if (!ref)
 347                 return -ENOMEM;
 348
 349         ref->root_id = root_id;
 350         if (key)
 351                 ref->key_for_search = *key;
 352         else
 353                 memset(&ref->key_for_search, 0, sizeof(ref->key_for_search));
 354
 355         ref->inode_list = NULL;
 356         ref->level = level;
 357         ref->count = count;
 358         ref->parent = parent;
 359         ref->wanted_disk_byte = wanted_disk_byte;
 360         prelim_ref_insert(fs_info, preftree, ref, sc);
 361         return extent_is_shared(sc);
 362 }
 363
 364 /* direct refs use root == 0, key == NULL */
 365 static int add_direct_ref(const struct btrfs_fs_info *fs_info,
 366                           struct preftrees *preftrees, int level, u64 parent,
 367                           u64 wanted_disk_byte, int count,
 368                           struct share_check *sc, gfp_t gfp_mask)
 369 {
 370         return add_prelim_ref(fs_info, &preftrees->direct, 0, NULL, level,
 371                               parent, wanted_disk_byte, count, sc, gfp_mask);
 372 }
 373
 374 /* indirect refs use parent == 0 */
 375 static int add_indirect_ref(const struct btrfs_fs_info *fs_info,
 376                             struct preftrees *preftrees, u64 root_id,
 377                             const struct btrfs_key *key, int level,
 378                             u64 wanted_disk_byte, int count,
 379                             struct share_check *sc, gfp_t gfp_mask)
 380 {
 381         struct preftree *tree = &preftrees->indirect;
 382
 383         if (!key)
 384                 tree = &preftrees->indirect_missing_keys;
 385         return add_prelim_ref(fs_info, tree, root_id, key, level, 0,
 386                               wanted_disk_byte, count, sc, gfp_mask);
 387 }
 388
 389 static int is_shared_data_backref(struct preftrees *preftrees, u64 bytenr)
 390 {
 391         struct rb_node **p = &preftrees->direct.root.rb_root.rb_node;
 392         struct rb_node *parent = NULL;
 393         struct prelim_ref *ref = NULL;
 394         struct prelim_ref target = {};
 395         int result;
 396
 397         target.parent = bytenr;
 398
 399         while (*p) {
 400                 parent = *p;
 401                 ref = rb_entry(parent, struct prelim_ref, rbnode);
 402                 result = prelim_ref_compare(ref, &target);
 403
 404                 if (result < 0)
 405                         p = &(*p)->rb_left;
 406                 else if (result > 0)
 407                         p = &(*p)->rb_right;
 408                 else
 409                         return 1;
 410         }
 411         return 0;
 412 }
 413
 414 static int add_all_parents(struct btrfs_root *root, struct btrfs_path *path,
 415                            struct ulist *parents,
 416                            struct preftrees *preftrees, struct prelim_ref *ref,
 417                            int level, u64 time_seq, const u64 *extent_item_pos,
 418                            bool ignore_offset)
 419 {
 420         int ret = 0;
 421         int slot;
 422         struct extent_buffer *eb;
 423         struct btrfs_key key;
 424         struct btrfs_key *key_for_search = &ref->key_for_search;
 425         struct btrfs_file_extent_item *fi;
 426         struct extent_inode_elem *eie = NULL, *old = NULL;
 427         u64 disk_byte;
 428         u64 wanted_disk_byte = ref->wanted_disk_byte;
 429         u64 count = 0;
 430         u64 data_offset;
 431
 432         if (level != 0) {
 433                 eb = path->nodes[level];
 434                 ret = ulist_add(parents, eb->start, 0, GFP_NOFS);
 435                 if (ret < 0)
 436                         return ret;
 437                 return 0;
 438         }
 439
 440         /*
 441          * 1. We normally enter this function with the path already pointing to
 442          *    the first item to check. But sometimes, we may enter it with
 443          *    slot == nritems.
 444          * 2. We are searching for normal backref but bytenr of this leaf
 445          *    matches shared data backref
 446          * 3. The leaf owner is not equal to the root we are searching
 447          *
 448          * For these cases, go to the next leaf before we continue.
 449          */
 450         eb = path->nodes[0];
 451         if (path->slots[0] >= btrfs_header_nritems(eb) ||
 452             is_shared_data_backref(preftrees, eb->start) ||
 453             ref->root_id != btrfs_header_owner(eb)) {
 454                 if (time_seq == SEQ_LAST)
 455                         ret = btrfs_next_leaf(root, path);
 456                 else
 457                         ret = btrfs_next_old_leaf(root, path, time_seq);
 458         }
 459
 460         while (!ret && count < ref->count) {
 461                 eb = path->nodes[0];
 462                 slot = path->slots[0];
 463
 464                 btrfs_item_key_to_cpu(eb, &key, slot);
 465
 466                 if (key.objectid != key_for_search->objectid ||
 467                     key.type != BTRFS_EXTENT_DATA_KEY)
 468                         break;
 469
 470                 /*
 471                  * We are searching for normal backref but bytenr of this leaf
 472                  * matches shared data backref, OR
 473                  * the leaf owner is not equal to the root we are searching for
 474                  */
 475                 if (slot == 0 &&
 476                     (is_shared_data_backref(preftrees, eb->start) ||
 477                      ref->root_id != btrfs_header_owner(eb))) {
 478                         if (time_seq == SEQ_LAST)
 479                                 ret = btrfs_next_leaf(root, path);
 480                         else
 481                                 ret = btrfs_next_old_leaf(root, path, time_seq);
 482                         continue;
 483                 }
 484                 fi = btrfs_item_ptr(eb, slot, struct btrfs_file_extent_item);
 485                 disk_byte = btrfs_file_extent_disk_bytenr(eb, fi);
 486                 data_offset = btrfs_file_extent_offset(eb, fi);
 487
 488                 if (disk_byte == wanted_disk_byte) {
 489                         eie = NULL;
 490                         old = NULL;
 491                         if (ref->key_for_search.offset == key.offset - data_offset)
 492                                 count++;
 493                         else
 494                                 goto next;
 495                         if (extent_item_pos) {
 496                                 ret = check_extent_in_eb(&key, eb, fi,
 497                                                 *extent_item_pos,
 498                                                 &eie, ignore_offset);
 499                                 if (ret < 0)
 500                                         break;
 501                         }
 502                         if (ret > 0)
 503                                 goto next;
 504                         ret = ulist_add_merge_ptr(parents, eb->start,
 505                                                   eie, (void **)&old, GFP_NOFS);
 506                         if (ret < 0)
 507                                 break;
 508                         if (!ret && extent_item_pos) {
 509                                 while (old->next)
 510                                         old = old->next;
 511                                 old->next = eie;
 512                         }
 513                         eie = NULL;
 514                 }
 515 next:
 516                 if (time_seq == SEQ_LAST)
 517                         ret = btrfs_next_item(root, path);
 518                 else
 519                         ret = btrfs_next_old_item(root, path, time_seq);
 520         }
 521
 522         if (ret > 0)
 523                 ret = 0;
 524         else if (ret < 0)
 525                 free_inode_elem_list(eie);
 526         return ret;
 527 }
 528
 529 /*
 530  * resolve an indirect backref in the form (root_id, key, level)
 531  * to a logical address
 532  */
 533 static int resolve_indirect_ref(struct btrfs_fs_info *fs_info,
 534                                 struct btrfs_path *path, u64 time_seq,
 535                                 struct preftrees *preftrees,
 536                                 struct prelim_ref *ref, struct ulist *parents,
 537                                 const u64 *extent_item_pos, bool ignore_offset)
 538 {
 539         struct btrfs_root *root;
 540         struct btrfs_key root_key;
 541         struct extent_buffer *eb;
 542         int ret = 0;
 543         int root_level;
 544         int level = ref->level;
 545         struct btrfs_key search_key = ref->key_for_search;
 546
 547         root_key.objectid = ref->root_id;
 548         root_key.type = BTRFS_ROOT_ITEM_KEY;
 549         root_key.offset = (u64)-1;
 550
 551         root = btrfs_get_fs_root(fs_info, &root_key, false);
 552         if (IS_ERR(root)) {
 553                 ret = PTR_ERR(root);
 554                 goto out_free;
 555         }
 556
 557         if (!path->search_commit_root &&
 558             test_bit(BTRFS_ROOT_DELETING, &root->state)) {
 559                 ret = -ENOENT;
 560                 goto out;
 561         }
 562
 563         if (btrfs_is_testing(fs_info)) {
 564                 ret = -ENOENT;
 565                 goto out;
 566         }
 567
 568         if (path->search_commit_root)
 569                 root_level = btrfs_header_level(root->commit_root);
 570         else if (time_seq == SEQ_LAST)
 571                 root_level = btrfs_header_level(root->node);
 572         else
 573                 root_level = btrfs_old_root_level(root, time_seq);
 574
 575         if (root_level + 1 == level)
 576                 goto out;
 577
 578         /*
 579          * We can often find data backrefs with an offset that is too large
 580          * (>= LLONG_MAX, maximum allowed file offset) due to underflows when
 581          * subtracting a file's offset with the data offset of its
 582          * corresponding extent data item. This can happen for example in the
 583          * clone ioctl.
 584          *
 585          * So if we detect such case we set the search key's offset to zero to
 586          * make sure we will find the matching file extent item at
 587          * add_all_parents(), otherwise we will miss it because the offset
 588          * taken form the backref is much larger then the offset of the file
 589          * extent item. This can make us scan a very large number of file
 590          * extent items, but at least it will not make us miss any.
 591          *
 592          * This is an ugly workaround for a behaviour that should have never
 593          * existed, but it does and a fix for the clone ioctl would touch a lot
 594          * of places, cause backwards incompatibility and would not fix the
 595          * problem for extents cloned with older kernels.
 596          */
 597         if (search_key.type == BTRFS_EXTENT_DATA_KEY &&
 598             search_key.offset >= LLONG_MAX)
 599                 search_key.offset = 0;
 600         path->lowest_level = level;
 601         if (time_seq == SEQ_LAST)
 602                 ret = btrfs_search_slot(NULL, root, &search_key, path, 0, 0);
 603         else
 604                 ret = btrfs_search_old_slot(root, &search_key, path, time_seq);
 605
 606         btrfs_debug(fs_info,
 607                 "search slot in root %llu (level %d, ref count %d) returned %d for key (%llu %u %llu)",
 608                  ref->root_id, level, ref->count, ret,
 609                  ref->key_for_search.objectid, ref->key_for_search.type,
 610                  ref->key_for_search.offset);
 611         if (ret < 0)
 612                 goto out;
 613
 614         eb = path->nodes[level];
 615         while (!eb) {
 616                 if (WARN_ON(!level)) {
 617                         ret = 1;
 618                         goto out;
 619                 }
 620                 level--;
 621                 eb = path->nodes[level];
 622         }
 623
 624         ret = add_all_parents(root, path, parents, preftrees, ref, level,
 625                               time_seq, extent_item_pos, ignore_offset);
 626 out:
 627         btrfs_put_root(root);
 628 out_free:
 629         path->lowest_level = 0;
 630         btrfs_release_path(path);
 631         return ret;
 632 }
 633
 634 static struct extent_inode_elem *
 635 unode_aux_to_inode_list(struct ulist_node *node)
 636 {
 637         if (!node)
 638                 return NULL;
 639         return (struct extent_inode_elem *)(uintptr_t)node->aux;
 640 }
 641
 642 /*
 643  * We maintain three separate rbtrees: one for direct refs, one for
 644  * indirect refs which have a key, and one for indirect refs which do not
 645  * have a key. Each tree does merge on insertion.
 646  *
 647  * Once all of the references are located, we iterate over the tree of
 648  * indirect refs with missing keys. An appropriate key is located and
 649  * the ref is moved onto the tree for indirect refs. After all missing
 650  * keys are thus located, we iterate over the indirect ref tree, resolve
 651  * each reference, and then insert the resolved reference onto the
 652  * direct tree (merging there too).
 653  *
 654  * New backrefs (i.e., for parent nodes) are added to the appropriate
 655  * rbtree as they are encountered. The new backrefs are subsequently
 656  * resolved as above.
 657  */
 658 static int resolve_indirect_refs(struct btrfs_fs_info *fs_info,
 659                                  struct btrfs_path *path, u64 time_seq,
 660                                  struct preftrees *preftrees,
 661                                  const u64 *extent_item_pos,
 662                                  struct share_check *sc, bool ignore_offset)
 663 {
 664         int err;
 665         int ret = 0;
 666         struct ulist *parents;
 667         struct ulist_node *node;
 668         struct ulist_iterator uiter;
 669         struct rb_node *rnode;
 670
 671         parents = ulist_alloc(GFP_NOFS);
 672         if (!parents)
 673                 return -ENOMEM;
 674
 675         /*
 676          * We could trade memory usage for performance here by iterating
 677          * the tree, allocating new refs for each insertion, and then
 678          * freeing the entire indirect tree when we're done.  In some test
 679          * cases, the tree can grow quite large (~200k objects).
 680          */
 681         while ((rnode = rb_first_cached(&preftrees->indirect.root))) {
 682                 struct prelim_ref *ref;
 683
 684                 ref = rb_entry(rnode, struct prelim_ref, rbnode);
 685                 if (WARN(ref->parent,
 686                          "BUG: direct ref found in indirect tree")) {
 687                         ret = -EINVAL;
 688                         goto out;
 689                 }
 690
 691                 rb_erase_cached(&ref->rbnode, &preftrees->indirect.root);
 692                 preftrees->indirect.count--;
 693
 694                 if (ref->count == 0) {
 695                         free_pref(ref);
 696                         continue;
 697                 }
 698
 699                 if (sc && sc->root_objectid &&
 700                     ref->root_id != sc->root_objectid) {
 701                         free_pref(ref);
 702                         ret = BACKREF_FOUND_SHARED;
 703                         goto out;
 704                 }
 705                 err = resolve_indirect_ref(fs_info, path, time_seq, preftrees,
 706                                            ref, parents, extent_item_pos,
 707                                            ignore_offset);
 708                 /*
 709                  * we can only tolerate ENOENT,otherwise,we should catch error
 710                  * and return directly.
 711                  */
 712                 if (err == -ENOENT) {
 713                         prelim_ref_insert(fs_info, &preftrees->direct, ref,
 714                                           NULL);
 715                         continue;
 716                 } else if (err) {
 717                         free_pref(ref);
 718                         ret = err;
 719                         goto out;
 720                 }
 721
 722                 /* we put the first parent into the ref at hand */
 723                 ULIST_ITER_INIT(&uiter);
 724                 node = ulist_next(parents, &uiter);
 725                 ref->parent = node ? node->val : 0;
 726                 ref->inode_list = unode_aux_to_inode_list(node);
 727
 728                 /* Add a prelim_ref(s) for any other parent(s). */
 729                 while ((node = ulist_next(parents, &uiter))) {
 730                         struct prelim_ref *new_ref;
 731
 732                         new_ref = kmem_cache_alloc(btrfs_prelim_ref_cache,
 733                                                    GFP_NOFS);
 734                         if (!new_ref) {
 735                                 free_pref(ref);
 736                                 ret = -ENOMEM;
 737                                 goto out;
 738                         }
 739                         memcpy(new_ref, ref, sizeof(*ref));
 740                         new_ref->parent = node->val;
 741                         new_ref->inode_list = unode_aux_to_inode_list(node);
 742                         prelim_ref_insert(fs_info, &preftrees->direct,
 743                                           new_ref, NULL);
 744                 }
 745
 746                 /*
 747                  * Now it's a direct ref, put it in the direct tree. We must
 748                  * do this last because the ref could be merged/freed here.
 749                  */
 750                 prelim_ref_insert(fs_info, &preftrees->direct, ref, NULL);
 751
 752                 ulist_reinit(parents);
 753                 cond_resched();
 754         }
 755 out:
 756         ulist_free(parents);
 757         return ret;
 758 }
 759
 760 /*
 761  * read tree blocks and add keys where required.
 762  */
 763 static int add_missing_keys(struct btrfs_fs_info *fs_info,
 764                             struct preftrees *preftrees, bool lock)
 765 {
 766         struct prelim_ref *ref;
 767         struct extent_buffer *eb;
 768         struct preftree *tree = &preftrees->indirect_missing_keys;
 769         struct rb_node *node;
 770
 771         while ((node = rb_first_cached(&tree->root))) {
 772                 ref = rb_entry(node, struct prelim_ref, rbnode);
 773                 rb_erase_cached(node, &tree->root);
 774
 775                 BUG_ON(ref->parent);    /* should not be a direct ref */
 776                 BUG_ON(ref->key_for_search.type);
 777                 BUG_ON(!ref->wanted_disk_byte);
 778
 779                 eb = read_tree_block(fs_info, ref->wanted_disk_byte, 0,
 780                                      ref->level - 1, NULL);
 781                 if (IS_ERR(eb)) {
 782                         free_pref(ref);
 783                         return PTR_ERR(eb);
 784                 } else if (!extent_buffer_uptodate(eb)) {
 785                         free_pref(ref);
 786                         free_extent_buffer(eb);
 787                         return -EIO;
 788                 }
 789                 if (lock)
 790                         btrfs_tree_read_lock(eb);
 791                 if (btrfs_header_level(eb) == 0)
 792                         btrfs_item_key_to_cpu(eb, &ref->key_for_search, 0);
 793                 else
 794                         btrfs_node_key_to_cpu(eb, &ref->key_for_search, 0);
 795                 if (lock)
 796                         btrfs_tree_read_unlock(eb);
 797                 free_extent_buffer(eb);
 798                 prelim_ref_insert(fs_info, &preftrees->indirect, ref, NULL);
 799                 cond_resched();
 800         }
 801         return 0;
 802 }
 803
 804 /*
 805  * add all currently queued delayed refs from this head whose seq nr is
 806  * smaller or equal that seq to the list
 807  */
 808 static int add_delayed_refs(const struct btrfs_fs_info *fs_info,
 809                             struct btrfs_delayed_ref_head *head, u64 seq,
 810                             struct preftrees *preftrees, struct share_check *sc)
 811 {
 812         struct btrfs_delayed_ref_node *node;
 813         struct btrfs_delayed_extent_op *extent_op = head->extent_op;
 814         struct btrfs_key key;
 815         struct btrfs_key tmp_op_key;
 816         struct rb_node *n;
 817         int count;
 818         int ret = 0;
 819
 820         if (extent_op && extent_op->update_key)
 821                 btrfs_disk_key_to_cpu(&tmp_op_key, &extent_op->key);
 822
 823         spin_lock(&head->lock);
 824         for (n = rb_first_cached(&head->ref_tree); n; n = rb_next(n)) {
 825                 node = rb_entry(n, struct btrfs_delayed_ref_node,
 826                                 ref_node);
 827                 if (node->seq > seq)
 828                         continue;
 829
 830                 switch (node->action) {
 831                 case BTRFS_ADD_DELAYED_EXTENT:
 832                 case BTRFS_UPDATE_DELAYED_HEAD:
 833                         WARN_ON(1);
 834                         continue;
 835                 case BTRFS_ADD_DELAYED_REF:
 836                         count = node->ref_mod;
 837                         break;
 838                 case BTRFS_DROP_DELAYED_REF:
 839                         count = node->ref_mod * -1;
 840                         break;
 841                 default:
 842                         BUG();
 843                 }
 844                 switch (node->type) {
 845                 case BTRFS_TREE_BLOCK_REF_KEY: {
 846                         /* NORMAL INDIRECT METADATA backref */
 847                         struct btrfs_delayed_tree_ref *ref;
 848
 849                         ref = btrfs_delayed_node_to_tree_ref(node);
 850                         ret = add_indirect_ref(fs_info, preftrees, ref->root,
 851                                                &tmp_op_key, ref->level + 1,
 852                                                node->bytenr, count, sc,
 853                                                GFP_ATOMIC);
 854                         break;
 855                 }
 856                 case BTRFS_SHARED_BLOCK_REF_KEY: {
 857                         /* SHARED DIRECT METADATA backref */
 858                         struct btrfs_delayed_tree_ref *ref;
 859
 860                         ref = btrfs_delayed_node_to_tree_ref(node);
 861
 862                         ret = add_direct_ref(fs_info, preftrees, ref->level + 1,
 863                                              ref->parent, node->bytenr, count,
 864                                              sc, GFP_ATOMIC);
 865                         break;
 866                 }
 867                 case BTRFS_EXTENT_DATA_REF_KEY: {
 868                         /* NORMAL INDIRECT DATA backref */
 869                         struct btrfs_delayed_data_ref *ref;
 870                         ref = btrfs_delayed_node_to_data_ref(node);
 871
 872                         key.objectid = ref->objectid;
 873                         key.type = BTRFS_EXTENT_DATA_KEY;
 874                         key.offset = ref->offset;
 875
 876                         /*
 877                          * Found a inum that doesn't match our known inum, we
 878                          * know it's shared.
 879                          */
 880                         if (sc && sc->inum && ref->objectid != sc->inum) {
 881                                 ret = BACKREF_FOUND_SHARED;
 882                                 goto out;
 883                         }
 884
 885                         ret = add_indirect_ref(fs_info, preftrees, ref->root,
 886                                                &key, 0, node->bytenr, count, sc,
 887                                                GFP_ATOMIC);
 888                         break;
 889                 }
 890                 case BTRFS_SHARED_DATA_REF_KEY: {
 891                         /* SHARED DIRECT FULL backref */
 892                         struct btrfs_delayed_data_ref *ref;
 893
 894                         ref = btrfs_delayed_node_to_data_ref(node);
 895
 896                         ret = add_direct_ref(fs_info, preftrees, 0, ref->parent,
 897                                              node->bytenr, count, sc,
 898                                              GFP_ATOMIC);
 899                         break;
 900                 }
 901                 default:
 902                         WARN_ON(1);
 903                 }
 904                 /*
 905                  * We must ignore BACKREF_FOUND_SHARED until all delayed
 906                  * refs have been checked.
 907                  */
 908                 if (ret && (ret != BACKREF_FOUND_SHARED))
 909                         break;
 910         }
 911         if (!ret)
 912                 ret = extent_is_shared(sc);
 913 out:
 914         spin_unlock(&head->lock);
 915         return ret;
 916 }
 917
 918 /*
 919  * add all inline backrefs for bytenr to the list
 920  *
 921  * Returns 0 on success, <0 on error, or BACKREF_FOUND_SHARED.
 922  */
 923 static int add_inline_refs(const struct btrfs_fs_info *fs_info,
 924                            struct btrfs_path *path, u64 bytenr,
 925                            int *info_level, struct preftrees *preftrees,
 926                            struct share_check *sc)
 927 {
 928         int ret = 0;
 929         int slot;
 930         struct extent_buffer *leaf;
 931         struct btrfs_key key;
 932         struct btrfs_key found_key;
 933         unsigned long ptr;
 934         unsigned long end;
 935         struct btrfs_extent_item *ei;
 936         u64 flags;
 937         u64 item_size;
 938
 939         /*
 940          * enumerate all inline refs
 941          */
 942         leaf = path->nodes[0];
 943         slot = path->slots[0];
 944
 945         item_size = btrfs_item_size_nr(leaf, slot);
 946         BUG_ON(item_size < sizeof(*ei));
 947
 948         ei = btrfs_item_ptr(leaf, slot, struct btrfs_extent_item);
 949         flags = btrfs_extent_flags(leaf, ei);
 950         btrfs_item_key_to_cpu(leaf, &found_key, slot);
 951
 952         ptr = (unsigned long)(ei + 1);
 953         end = (unsigned long)ei + item_size;
 954
 955         if (found_key.type == BTRFS_EXTENT_ITEM_KEY &&
 956             flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
 957                 struct btrfs_tree_block_info *info;
 958
 959                 info = (struct btrfs_tree_block_info *)ptr;
 960                 *info_level = btrfs_tree_block_level(leaf, info);
 961                 ptr += sizeof(struct btrfs_tree_block_info);
 962                 BUG_ON(ptr > end);
 963         } else if (found_key.type == BTRFS_METADATA_ITEM_KEY) {
 964                 *info_level = found_key.offset;
 965         } else {
 966                 BUG_ON(!(flags & BTRFS_EXTENT_FLAG_DATA));
 967         }
 968
 969         while (ptr < end) {
 970                 struct btrfs_extent_inline_ref *iref;
 971                 u64 offset;
 972                 int type;
 973
 974                 iref = (struct btrfs_extent_inline_ref *)ptr;
 975                 type = btrfs_get_extent_inline_ref_type(leaf, iref,
 976                                                         BTRFS_REF_TYPE_ANY);
 977                 if (type == BTRFS_REF_TYPE_INVALID)
 978                         return -EUCLEAN;
 979
 980                 offset = btrfs_extent_inline_ref_offset(leaf, iref);
 981
 982                 switch (type) {
 983                 case BTRFS_SHARED_BLOCK_REF_KEY:
 984                         ret = add_direct_ref(fs_info, preftrees,
 985                                              *info_level + 1, offset,
 986                                              bytenr, 1, NULL, GFP_NOFS);
 987                         break;
 988                 case BTRFS_SHARED_DATA_REF_KEY: {
 989                         struct btrfs_shared_data_ref *sdref;
 990                         int count;
 991
 992                         sdref = (struct btrfs_shared_data_ref *)(iref + 1);
 993                         count = btrfs_shared_data_ref_count(leaf, sdref);
 994
 995                         ret = add_direct_ref(fs_info, preftrees, 0, offset,
 996                                              bytenr, count, sc, GFP_NOFS);
 997                         break;
 998                 }
 999                 case BTRFS_TREE_BLOCK_REF_KEY:
1000                         ret = add_indirect_ref(fs_info, preftrees, offset,
1001                                                NULL, *info_level + 1,
1002                                                bytenr, 1, NULL, GFP_NOFS);
1003                         break;
1004                 case BTRFS_EXTENT_DATA_REF_KEY: {
1005                         struct btrfs_extent_data_ref *dref;
1006                         int count;
1007                         u64 root;
1008
1009                         dref = (struct btrfs_extent_data_ref *)(&iref->offset);
1010                         count = btrfs_extent_data_ref_count(leaf, dref);
1011                         key.objectid = btrfs_extent_data_ref_objectid(leaf,
1012                                                                       dref);
1013                         key.type = BTRFS_EXTENT_DATA_KEY;
1014                         key.offset = btrfs_extent_data_ref_offset(leaf, dref);
1015
1016                         if (sc && sc->inum && key.objectid != sc->inum) {
1017                                 ret = BACKREF_FOUND_SHARED;
1018                                 break;
1019                         }
1020
1021                         root = btrfs_extent_data_ref_root(leaf, dref);
1022
1023                         ret = add_indirect_ref(fs_info, preftrees, root,
1024                                                &key, 0, bytenr, count,
1025                                                sc, GFP_NOFS);
1026                         break;
1027                 }
1028                 default:
1029                         WARN_ON(1);
1030                 }
1031                 if (ret)
1032                         return ret;
1033                 ptr += btrfs_extent_inline_ref_size(type);
1034         }
1035
1036         return 0;
1037 }
1038
1039 /*
1040  * add all non-inline backrefs for bytenr to the list
1041  *
1042  * Returns 0 on success, <0 on error, or BACKREF_FOUND_SHARED.
1043  */
1044 static int add_keyed_refs(struct btrfs_fs_info *fs_info,
1045                           struct btrfs_path *path, u64 bytenr,
1046                           int info_level, struct preftrees *preftrees,
1047                           struct share_check *sc)
1048 {
1049         struct btrfs_root *extent_root = fs_info->extent_root;
1050         int ret;
1051         int slot;
1052         struct extent_buffer *leaf;
1053         struct btrfs_key key;
1054
1055         while (1) {
1056                 ret = btrfs_next_item(extent_root, path);
1057                 if (ret < 0)
1058                         break;
1059                 if (ret) {
1060                         ret = 0;
1061                         break;
1062                 }
1063
1064                 slot = path->slots[0];
1065                 leaf = path->nodes[0];
1066                 btrfs_item_key_to_cpu(leaf, &key, slot);
1067
1068                 if (key.objectid != bytenr)
1069                         break;
1070                 if (key.type < BTRFS_TREE_BLOCK_REF_KEY)
1071                         continue;
1072                 if (key.type > BTRFS_SHARED_DATA_REF_KEY)
1073                         break;
1074
1075                 switch (key.type) {
1076                 case BTRFS_SHARED_BLOCK_REF_KEY:
1077                         /* SHARED DIRECT METADATA backref */
1078                         ret = add_direct_ref(fs_info, preftrees,
1079                                              info_level + 1, key.offset,
1080                                              bytenr, 1, NULL, GFP_NOFS);
1081                         break;
1082                 case BTRFS_SHARED_DATA_REF_KEY: {
1083                         /* SHARED DIRECT FULL backref */
1084                         struct btrfs_shared_data_ref *sdref;
1085                         int count;
1086
1087                         sdref = btrfs_item_ptr(leaf, slot,
1088                                               struct btrfs_shared_data_ref);
1089                         count = btrfs_shared_data_ref_count(leaf, sdref);
1090                         ret = add_direct_ref(fs_info, preftrees, 0,
1091                                              key.offset, bytenr, count,
1092                                              sc, GFP_NOFS);
1093                         break;
1094                 }
1095                 case BTRFS_TREE_BLOCK_REF_KEY:
1096                         /* NORMAL INDIRECT METADATA backref */
1097                         ret = add_indirect_ref(fs_info, preftrees, key.offset,
1098                                                NULL, info_level + 1, bytenr,
1099                                                1, NULL, GFP_NOFS);
1100                         break;
1101                 case BTRFS_EXTENT_DATA_REF_KEY: {
1102                         /* NORMAL INDIRECT DATA backref */
1103                         struct btrfs_extent_data_ref *dref;
1104                         int count;
1105                         u64 root;
1106
1107                         dref = btrfs_item_ptr(leaf, slot,
1108                                               struct btrfs_extent_data_ref);
1109                         count = btrfs_extent_data_ref_count(leaf, dref);
1110                         key.objectid = btrfs_extent_data_ref_objectid(leaf,
1111                                                                       dref);
1112                         key.type = BTRFS_EXTENT_DATA_KEY;
1113                         key.offset = btrfs_extent_data_ref_offset(leaf, dref);
1114
1115                         if (sc && sc->inum && key.objectid != sc->inum) {
1116                                 ret = BACKREF_FOUND_SHARED;
1117                                 break;
1118                         }
1119
1120                         root = btrfs_extent_data_ref_root(leaf, dref);
1121                         ret = add_indirect_ref(fs_info, preftrees, root,
1122                                                &key, 0, bytenr, count,
1123                                                sc, GFP_NOFS);
1124                         break;
1125                 }
1126                 default:
1127                         WARN_ON(1);
1128                 }
1129                 if (ret)
1130                         return ret;
1131
1132         }
1133
1134         return ret;
1135 }
1136
1137 /*
1138  * this adds all existing backrefs (inline backrefs, backrefs and delayed
1139  * refs) for the given bytenr to the refs list, merges duplicates and resolves
1140  * indirect refs to their parent bytenr.
1141  * When roots are found, they're added to the roots list
1142  *
1143  * If time_seq is set to SEQ_LAST, it will not search delayed_refs, and behave
1144  * much like trans == NULL case, the difference only lies in it will not
1145  * commit root.
1146  * The special case is for qgroup to search roots in commit_transaction().
1147  *
1148  * @sc - if !NULL, then immediately return BACKREF_FOUND_SHARED when a
1149  * shared extent is detected.
1150  *
1151  * Otherwise this returns 0 for success and <0 for an error.
1152  *
1153  * If ignore_offset is set to false, only extent refs whose offsets match
1154  * extent_item_pos are returned.  If true, every extent ref is returned
1155  * and extent_item_pos is ignored.
1156  *
1157  * FIXME some caching might speed things up
1158  */
1159 static int find_parent_nodes(struct btrfs_trans_handle *trans,
1160                              struct btrfs_fs_info *fs_info, u64 bytenr,
1161                              u64 time_seq, struct ulist *refs,
1162                              struct ulist *roots, const u64 *extent_item_pos,
1163                              struct share_check *sc, bool ignore_offset)
1164 {
1165         struct btrfs_key key;
1166         struct btrfs_path *path;
1167         struct btrfs_delayed_ref_root *delayed_refs = NULL;
1168         struct btrfs_delayed_ref_head *head;
1169         int info_level = 0;
1170         int ret;
1171         struct prelim_ref *ref;
1172         struct rb_node *node;
1173         struct extent_inode_elem *eie = NULL;
1174         struct preftrees preftrees = {
1175                 .direct = PREFTREE_INIT,
1176                 .indirect = PREFTREE_INIT,
1177                 .indirect_missing_keys = PREFTREE_INIT
1178         };
1179
1180         key.objectid = bytenr;
1181         key.offset = (u64)-1;
1182         if (btrfs_fs_incompat(fs_info, SKINNY_METADATA))
1183                 key.type = BTRFS_METADATA_ITEM_KEY;
1184         else
1185                 key.type = BTRFS_EXTENT_ITEM_KEY;
1186
1187         path = btrfs_alloc_path();
1188         if (!path)
1189                 return -ENOMEM;
1190         if (!trans) {
1191                 path->search_commit_root = 1;
1192                 path->skip_locking = 1;
1193         }
1194
1195         if (time_seq == SEQ_LAST)
1196                 path->skip_locking = 1;
1197
1198         /*
1199          * grab both a lock on the path and a lock on the delayed ref head.
1200          * We need both to get a consistent picture of how the refs look
1201          * at a specified point in time
1202          */
1203 again:
1204         head = NULL;
1205
1206         ret = btrfs_search_slot(trans, fs_info->extent_root, &key, path, 0, 0);
1207         if (ret < 0)
1208                 goto out;
1209         BUG_ON(ret == 0);
1210
1211 #ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
1212         if (trans && likely(trans->type != __TRANS_DUMMY) &&
1213             time_seq != SEQ_LAST) {
1214 #else
1215         if (trans && time_seq != SEQ_LAST) {
1216 #endif
1217                 /*
1218                  * look if there are updates for this ref queued and lock the
1219                  * head
1220                  */
1221                 delayed_refs = &trans->transaction->delayed_refs;
1222                 spin_lock(&delayed_refs->lock);
1223                 head = btrfs_find_delayed_ref_head(delayed_refs, bytenr);
1224                 if (head) {
1225                         if (!mutex_trylock(&head->mutex)) {
1226                                 refcount_inc(&head->refs);
1227                                 spin_unlock(&delayed_refs->lock);
1228
1229                                 btrfs_release_path(path);
1230
1231                                 /*
1232                                  * Mutex was contended, block until it's
1233                                  * released and try again
1234                                  */
1235                                 mutex_lock(&head->mutex);
1236                                 mutex_unlock(&head->mutex);
1237                                 btrfs_put_delayed_ref_head(head);
1238                                 goto again;
1239                         }
1240                         spin_unlock(&delayed_refs->lock);
1241                         ret = add_delayed_refs(fs_info, head, time_seq,
1242                                                &preftrees, sc);
1243                         mutex_unlock(&head->mutex);
1244                         if (ret)
1245                                 goto out;
1246                 } else {
1247                         spin_unlock(&delayed_refs->lock);
1248                 }
1249         }
1250
1251         if (path->slots[0]) {
1252                 struct extent_buffer *leaf;
1253                 int slot;
1254
1255                 path->slots[0]--;
1256                 leaf = path->nodes[0];
1257                 slot = path->slots[0];
1258                 btrfs_item_key_to_cpu(leaf, &key, slot);
1259                 if (key.objectid == bytenr &&
1260                     (key.type == BTRFS_EXTENT_ITEM_KEY ||
1261                      key.type == BTRFS_METADATA_ITEM_KEY)) {
1262                         ret = add_inline_refs(fs_info, path, bytenr,
1263                                               &info_level, &preftrees, sc);
1264                         if (ret)
1265                                 goto out;
1266                         ret = add_keyed_refs(fs_info, path, bytenr, info_level,
1267                                              &preftrees, sc);
1268                         if (ret)
1269                                 goto out;
1270                 }
1271         }
1272
1273         btrfs_release_path(path);
1274
1275         ret = add_missing_keys(fs_info, &preftrees, path->skip_locking == 0);
1276         if (ret)
1277                 goto out;
1278
1279         WARN_ON(!RB_EMPTY_ROOT(&preftrees.indirect_missing_keys.root.rb_root));
1280
1281         ret = resolve_indirect_refs(fs_info, path, time_seq, &preftrees,
1282                                     extent_item_pos, sc, ignore_offset);
1283         if (ret)
1284                 goto out;
1285
1286         WARN_ON(!RB_EMPTY_ROOT(&preftrees.indirect.root.rb_root));
1287
1288         /*
1289          * This walks the tree of merged and resolved refs. Tree blocks are
1290          * read in as needed. Unique entries are added to the ulist, and
1291          * the list of found roots is updated.
1292          *
1293          * We release the entire tree in one go before returning.
1294          */
1295         node = rb_first_cached(&preftrees.direct.root);
1296         while (node) {
1297                 ref = rb_entry(node, struct prelim_ref, rbnode);
1298                 node = rb_next(&ref->rbnode);
1299                 /*
1300                  * ref->count < 0 can happen here if there are delayed
1301                  * refs with a node->action of BTRFS_DROP_DELAYED_REF.
1302                  * prelim_ref_insert() relies on this when merging
1303                  * identical refs to keep the overall count correct.
1304                  * prelim_ref_insert() will merge only those refs
1305                  * which compare identically.  Any refs having
1306                  * e.g. different offsets would not be merged,
1307                  * and would retain their original ref->count < 0.
1308                  */
1309                 if (roots && ref->count && ref->root_id && ref->parent == 0) {
1310                         if (sc && sc->root_objectid &&
1311                             ref->root_id != sc->root_objectid) {
1312                                 ret = BACKREF_FOUND_SHARED;
1313                                 goto out;
1314                         }
1315
1316                         /* no parent == root of tree */
1317                         ret = ulist_add(roots, ref->root_id, 0, GFP_NOFS);
1318                         if (ret < 0)
1319                                 goto out;
1320                 }
1321                 if (ref->count && ref->parent) {
1322                         if (extent_item_pos && !ref->inode_list &&
1323                             ref->level == 0) {
1324                                 struct extent_buffer *eb;
1325
1326                                 eb = read_tree_block(fs_info, ref->parent, 0,
1327                                                      ref->level, NULL);
1328                                 if (IS_ERR(eb)) {
1329                                         ret = PTR_ERR(eb);
1330                                         goto out;
1331                                 } else if (!extent_buffer_uptodate(eb)) {
1332                                         free_extent_buffer(eb);
1333                                         ret = -EIO;
1334                                         goto out;
1335                                 }
1336
1337                                 if (!path->skip_locking) {
1338                                         btrfs_tree_read_lock(eb);
1339                                         btrfs_set_lock_blocking_read(eb);
1340                                 }
1341                                 ret = find_extent_in_eb(eb, bytenr,
1342                                                         *extent_item_pos, &eie, ignore_offset);
1343                                 if (!path->skip_locking)
1344                                         btrfs_tree_read_unlock_blocking(eb);
1345                                 free_extent_buffer(eb);
1346                                 if (ret < 0)
1347                                         goto out;
1348                                 ref->inode_list = eie;
1349                         }
1350                         ret = ulist_add_merge_ptr(refs, ref->parent,
1351                                                   ref->inode_list,
1352                                                   (void **)&eie, GFP_NOFS);
1353                         if (ret < 0)
1354                                 goto out;
1355                         if (!ret && extent_item_pos) {
1356                                 /*
1357                                  * we've recorded that parent, so we must extend
1358                                  * its inode list here
1359                                  */
1360                                 BUG_ON(!eie);
1361                                 while (eie->next)
1362                                         eie = eie->next;
1363                                 eie->next = ref->inode_list;
1364                         }
1365                         eie = NULL;
1366                 }
1367                 cond_resched();
1368         }
1369
1370 out:
1371         btrfs_free_path(path);
1372
1373         prelim_release(&preftrees.direct);
1374         prelim_release(&preftrees.indirect);
1375         prelim_release(&preftrees.indirect_missing_keys);
1376
1377         if (ret < 0)
1378                 free_inode_elem_list(eie);
1379         return ret;
1380 }
1381
1382 static void free_leaf_list(struct ulist *blocks)
1383 {
1384         struct ulist_node *node = NULL;
1385         struct extent_inode_elem *eie;
1386         struct ulist_iterator uiter;
1387
1388         ULIST_ITER_INIT(&uiter);
1389         while ((node = ulist_next(blocks, &uiter))) {
1390                 if (!node->aux)
1391                         continue;
1392                 eie = unode_aux_to_inode_list(node);
1393                 free_inode_elem_list(eie);
1394                 node->aux = 0;
1395         }
1396
1397         ulist_free(blocks);
1398 }
1399
1400 /*
1401  * Finds all leafs with a reference to the specified combination of bytenr and
1402  * offset. key_list_head will point to a list of corresponding keys (caller must
1403  * free each list element). The leafs will be stored in the leafs ulist, which
1404  * must be freed with ulist_free.
1405  *
1406  * returns 0 on success, <0 on error
1407  */
1408 int btrfs_find_all_leafs(struct btrfs_trans_handle *trans,
1409                          struct btrfs_fs_info *fs_info, u64 bytenr,
1410                          u64 time_seq, struct ulist **leafs,
1411                          const u64 *extent_item_pos, bool ignore_offset)
1412 {
1413         int ret;
1414
1415         *leafs = ulist_alloc(GFP_NOFS);
1416         if (!*leafs)
1417                 return -ENOMEM;
1418
1419         ret = find_parent_nodes(trans, fs_info, bytenr, time_seq,
1420                                 *leafs, NULL, extent_item_pos, NULL, ignore_offset);
1421         if (ret < 0 && ret != -ENOENT) {
1422                 free_leaf_list(*leafs);
1423                 return ret;
1424         }
1425
1426         return 0;
1427 }
1428
1429 /*
1430  * walk all backrefs for a given extent to find all roots that reference this
1431  * extent. Walking a backref means finding all extents that reference this
1432  * extent and in turn walk the backrefs of those, too. Naturally this is a
1433  * recursive process, but here it is implemented in an iterative fashion: We
1434  * find all referencing extents for the extent in question and put them on a
1435  * list. In turn, we find all referencing extents for those, further appending
1436  * to the list. The way we iterate the list allows adding more elements after
1437  * the current while iterating. The process stops when we reach the end of the
1438  * list. Found roots are added to the roots list.
1439  *
1440  * returns 0 on success, < 0 on error.
1441  */
1442 static int btrfs_find_all_roots_safe(struct btrfs_trans_handle *trans,
1443                                      struct btrfs_fs_info *fs_info, u64 bytenr,
1444                                      u64 time_seq, struct ulist **roots,
1445                                      bool ignore_offset)
1446 {
1447         struct ulist *tmp;
1448         struct ulist_node *node = NULL;
1449         struct ulist_iterator uiter;
1450         int ret;
1451
1452         tmp = ulist_alloc(GFP_NOFS);
1453         if (!tmp)
1454                 return -ENOMEM;
1455         *roots = ulist_alloc(GFP_NOFS);
1456         if (!*roots) {
1457                 ulist_free(tmp);
1458                 return -ENOMEM;
1459         }
1460
1461         ULIST_ITER_INIT(&uiter);
1462         while (1) {
1463                 ret = find_parent_nodes(trans, fs_info, bytenr, time_seq,
1464                                         tmp, *roots, NULL, NULL, ignore_offset);
1465                 if (ret < 0 && ret != -ENOENT) {
1466                         ulist_free(tmp);
1467                         ulist_free(*roots);
1468                         return ret;
1469                 }
1470                 node = ulist_next(tmp, &uiter);
1471                 if (!node)
1472                         break;
1473                 bytenr = node->val;
1474                 cond_resched();
1475         }
1476
1477         ulist_free(tmp);
1478         return 0;
1479 }
1480
1481 int btrfs_find_all_roots(struct btrfs_trans_handle *trans,
1482                          struct btrfs_fs_info *fs_info, u64 bytenr,
1483                          u64 time_seq, struct ulist **roots,
1484                          bool ignore_offset)
1485 {
1486         int ret;
1487
1488         if (!trans)
1489                 down_read(&fs_info->commit_root_sem);
1490         ret = btrfs_find_all_roots_safe(trans, fs_info, bytenr,
1491                                         time_seq, roots, ignore_offset);
1492         if (!trans)
1493                 up_read(&fs_info->commit_root_sem);
1494         return ret;
1495 }
1496
1497 /**
1498  * btrfs_check_shared - tell us whether an extent is shared
1499  *
1500  * btrfs_check_shared uses the backref walking code but will short
1501  * circuit as soon as it finds a root or inode that doesn't match the
1502  * one passed in. This provides a significant performance benefit for
1503  * callers (such as fiemap) which want to know whether the extent is
1504  * shared but do not need a ref count.
1505  *
1506  * This attempts to attach to the running transaction in order to account for
1507  * delayed refs, but continues on even when no running transaction exists.
1508  *
1509  * Return: 0 if extent is not shared, 1 if it is shared, < 0 on error.
1510  */
1511 int btrfs_check_shared(struct btrfs_root *root, u64 inum, u64 bytenr,
1512                 struct ulist *roots, struct ulist *tmp)
1513 {
1514         struct btrfs_fs_info *fs_info = root->fs_info;
1515         struct btrfs_trans_handle *trans;
1516         struct ulist_iterator uiter;
1517         struct ulist_node *node;
1518         struct seq_list elem = SEQ_LIST_INIT(elem);
1519         int ret = 0;
1520         struct share_check shared = {
1521                 .root_objectid = root->root_key.objectid,
1522                 .inum = inum,
1523                 .share_count = 0,
1524         };
1525
1526         ulist_init(roots);
1527         ulist_init(tmp);
1528
1529         trans = btrfs_join_transaction_nostart(root);
1530         if (IS_ERR(trans)) {
1531                 if (PTR_ERR(trans) != -ENOENT && PTR_ERR(trans) != -EROFS) {
1532                         ret = PTR_ERR(trans);
1533                         goto out;
1534                 }
1535                 trans = NULL;
1536                 down_read(&fs_info->commit_root_sem);
1537         } else {
1538                 btrfs_get_tree_mod_seq(fs_info, &elem);
1539         }
1540
1541         ULIST_ITER_INIT(&uiter);
1542         while (1) {
1543                 ret = find_parent_nodes(trans, fs_info, bytenr, elem.seq, tmp,
1544                                         roots, NULL, &shared, false);
1545                 if (ret == BACKREF_FOUND_SHARED) {
1546                         /* this is the only condition under which we return 1 */
1547                         ret = 1;
1548                         break;
1549                 }
1550                 if (ret < 0 && ret != -ENOENT)
1551                         break;
1552                 ret = 0;
1553                 node = ulist_next(tmp, &uiter);
1554                 if (!node)
1555                         break;
1556                 bytenr = node->val;
1557                 shared.share_count = 0;
1558                 cond_resched();
1559         }
1560
1561         if (trans) {
1562                 btrfs_put_tree_mod_seq(fs_info, &elem);
1563                 btrfs_end_transaction(trans);
1564         } else {
1565                 up_read(&fs_info->commit_root_sem);
1566         }
1567 out:
1568         ulist_release(roots);
1569         ulist_release(tmp);
1570         return ret;
1571 }
1572
1573 int btrfs_find_one_extref(struct btrfs_root *root, u64 inode_objectid,
1574                           u64 start_off, struct btrfs_path *path,
1575                           struct btrfs_inode_extref **ret_extref,
1576                           u64 *found_off)
1577 {
1578         int ret, slot;
1579         struct btrfs_key key;
1580         struct btrfs_key found_key;
1581         struct btrfs_inode_extref *extref;
1582         const struct extent_buffer *leaf;
1583         unsigned long ptr;
1584
1585         key.objectid = inode_objectid;
1586         key.type = BTRFS_INODE_EXTREF_KEY;
1587         key.offset = start_off;
1588
1589         ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
1590         if (ret < 0)
1591                 return ret;
1592
1593         while (1) {
1594                 leaf = path->nodes[0];
1595                 slot = path->slots[0];
1596                 if (slot >= btrfs_header_nritems(leaf)) {
1597                         /*
1598                          * If the item at offset is not found,
1599                          * btrfs_search_slot will point us to the slot
1600                          * where it should be inserted. In our case
1601                          * that will be the slot directly before the
1602                          * next INODE_REF_KEY_V2 item. In the case
1603                          * that we're pointing to the last slot in a
1604                          * leaf, we must move one leaf over.
1605                          */
1606                         ret = btrfs_next_leaf(root, path);
1607                         if (ret) {
1608                                 if (ret >= 1)
1609                                         ret = -ENOENT;
1610                                 break;
1611                         }
1612                         continue;
1613                 }
1614
1615                 btrfs_item_key_to_cpu(leaf, &found_key, slot);
1616
1617                 /*
1618                  * Check that we're still looking at an extended ref key for
1619                  * this particular objectid. If we have different
1620                  * objectid or type then there are no more to be found
1621                  * in the tree and we can exit.
1622                  */
1623                 ret = -ENOENT;
1624                 if (found_key.objectid != inode_objectid)
1625                         break;
1626                 if (found_key.type != BTRFS_INODE_EXTREF_KEY)
1627                         break;
1628
1629                 ret = 0;
1630                 ptr = btrfs_item_ptr_offset(leaf, path->slots[0]);
1631                 extref = (struct btrfs_inode_extref *)ptr;
1632                 *ret_extref = extref;
1633                 if (found_off)
1634                         *found_off = found_key.offset;
1635                 break;
1636         }
1637
1638         return ret;
1639 }
1640
1641 /*
1642  * this iterates to turn a name (from iref/extref) into a full filesystem path.
1643  * Elements of the path are separated by '/' and the path is guaranteed to be
1644  * 0-terminated. the path is only given within the current file system.
1645  * Therefore, it never starts with a '/'. the caller is responsible to provide
1646  * "size" bytes in "dest". the dest buffer will be filled backwards. finally,
1647  * the start point of the resulting string is returned. this pointer is within
1648  * dest, normally.
1649  * in case the path buffer would overflow, the pointer is decremented further
1650  * as if output was written to the buffer, though no more output is actually
1651  * generated. that way, the caller can determine how much space would be
1652  * required for the path to fit into the buffer. in that case, the returned
1653  * value will be smaller than dest. callers must check this!
1654  */
1655 char *btrfs_ref_to_path(struct btrfs_root *fs_root, struct btrfs_path *path,
1656                         u32 name_len, unsigned long name_off,
1657                         struct extent_buffer *eb_in, u64 parent,
1658                         char *dest, u32 size)
1659 {
1660         int slot;
1661         u64 next_inum;
1662         int ret;
1663         s64 bytes_left = ((s64)size) - 1;
1664         struct extent_buffer *eb = eb_in;
1665         struct btrfs_key found_key;
1666         int leave_spinning = path->leave_spinning;
1667         struct btrfs_inode_ref *iref;
1668
1669         if (bytes_left >= 0)
1670                 dest[bytes_left] = '\0';
1671
1672         path->leave_spinning = 1;
1673         while (1) {
1674                 bytes_left -= name_len;
1675                 if (bytes_left >= 0)
1676                         read_extent_buffer(eb, dest + bytes_left,
1677                                            name_off, name_len);
1678                 if (eb != eb_in) {
1679                         if (!path->skip_locking)
1680                                 btrfs_tree_read_unlock_blocking(eb);
1681                         free_extent_buffer(eb);
1682                 }
1683                 ret = btrfs_find_item(fs_root, path, parent, 0,
1684                                 BTRFS_INODE_REF_KEY, &found_key);
1685                 if (ret > 0)
1686                         ret = -ENOENT;
1687                 if (ret)
1688                         break;
1689
1690                 next_inum = found_key.offset;
1691
1692                 /* regular exit ahead */
1693                 if (parent == next_inum)
1694                         break;
1695
1696                 slot = path->slots[0];
1697                 eb = path->nodes[0];
1698                 /* make sure we can use eb after releasing the path */
1699                 if (eb != eb_in) {
1700                         if (!path->skip_locking)
1701                                 btrfs_set_lock_blocking_read(eb);
1702                         path->nodes[0] = NULL;
1703                         path->locks[0] = 0;
1704                 }
1705                 btrfs_release_path(path);
1706                 iref = btrfs_item_ptr(eb, slot, struct btrfs_inode_ref);
1707
1708                 name_len = btrfs_inode_ref_name_len(eb, iref);
1709                 name_off = (unsigned long)(iref + 1);
1710
1711                 parent = next_inum;
1712                 --bytes_left;
1713                 if (bytes_left >= 0)
1714                         dest[bytes_left] = '/';
1715         }
1716
1717         btrfs_release_path(path);
1718         path->leave_spinning = leave_spinning;
1719
1720         if (ret)
1721                 return ERR_PTR(ret);
1722
1723         return dest + bytes_left;
1724 }
1725
1726 /*
1727  * this makes the path point to (logical EXTENT_ITEM *)
1728  * returns BTRFS_EXTENT_FLAG_DATA for data, BTRFS_EXTENT_FLAG_TREE_BLOCK for
1729  * tree blocks and <0 on error.
1730  */
1731 int extent_from_logical(struct btrfs_fs_info *fs_info, u64 logical,
1732                         struct btrfs_path *path, struct btrfs_key *found_key,
1733                         u64 *flags_ret)
1734 {
1735         int ret;
1736         u64 flags;
1737         u64 size = 0;
1738         u32 item_size;
1739         const struct extent_buffer *eb;
1740         struct btrfs_extent_item *ei;
1741         struct btrfs_key key;
1742
1743         if (btrfs_fs_incompat(fs_info, SKINNY_METADATA))
1744                 key.type = BTRFS_METADATA_ITEM_KEY;
1745         else
1746                 key.type = BTRFS_EXTENT_ITEM_KEY;
1747         key.objectid = logical;
1748         key.offset = (u64)-1;
1749
1750         ret = btrfs_search_slot(NULL, fs_info->extent_root, &key, path, 0, 0);
1751         if (ret < 0)
1752                 return ret;
1753
1754         ret = btrfs_previous_extent_item(fs_info->extent_root, path, 0);
1755         if (ret) {
1756                 if (ret > 0)
1757                         ret = -ENOENT;
1758                 return ret;
1759         }
1760         btrfs_item_key_to_cpu(path->nodes[0], found_key, path->slots[0]);
1761         if (found_key->type == BTRFS_METADATA_ITEM_KEY)
1762                 size = fs_info->nodesize;
1763         else if (found_key->type == BTRFS_EXTENT_ITEM_KEY)
1764                 size = found_key->offset;
1765
1766         if (found_key->objectid > logical ||
1767             found_key->objectid + size <= logical) {
1768                 btrfs_debug(fs_info,
1769                         "logical %llu is not within any extent", logical);
1770                 return -ENOENT;
1771         }
1772
1773         eb = path->nodes[0];
1774         item_size = btrfs_item_size_nr(eb, path->slots[0]);
1775         BUG_ON(item_size < sizeof(*ei));
1776
1777         ei = btrfs_item_ptr(eb, path->slots[0], struct btrfs_extent_item);
1778         flags = btrfs_extent_flags(eb, ei);
1779
1780         btrfs_debug(fs_info,
1781                 "logical %llu is at position %llu within the extent (%llu EXTENT_ITEM %llu) flags %#llx size %u",
1782                  logical, logical - found_key->objectid, found_key->objectid,
1783                  found_key->offset, flags, item_size);
1784
1785         WARN_ON(!flags_ret);
1786         if (flags_ret) {
1787                 if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK)
1788                         *flags_ret = BTRFS_EXTENT_FLAG_TREE_BLOCK;
1789                 else if (flags & BTRFS_EXTENT_FLAG_DATA)
1790                         *flags_ret = BTRFS_EXTENT_FLAG_DATA;
1791                 else
1792                         BUG();
1793                 return 0;
1794         }
1795
1796         return -EIO;
1797 }
1798
1799 /*
1800  * helper function to iterate extent inline refs. ptr must point to a 0 value
1801  * for the first call and may be modified. it is used to track state.
1802  * if more refs exist, 0 is returned and the next call to
1803  * get_extent_inline_ref must pass the modified ptr parameter to get the
1804  * next ref. after the last ref was processed, 1 is returned.
1805  * returns <0 on error
1806  */
1807 static int get_extent_inline_ref(unsigned long *ptr,
1808                                  const struct extent_buffer *eb,
1809                                  const struct btrfs_key *key,
1810                                  const struct btrfs_extent_item *ei,
1811                                  u32 item_size,
1812                                  struct btrfs_extent_inline_ref **out_eiref,
1813                                  int *out_type)
1814 {
1815         unsigned long end;
1816         u64 flags;
1817         struct btrfs_tree_block_info *info;
1818
1819         if (!*ptr) {
1820                 /* first call */
1821                 flags = btrfs_extent_flags(eb, ei);
1822                 if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
1823                         if (key->type == BTRFS_METADATA_ITEM_KEY) {
1824                                 /* a skinny metadata extent */
1825                                 *out_eiref =
1826                                      (struct btrfs_extent_inline_ref *)(ei + 1);
1827                         } else {
1828                                 WARN_ON(key->type != BTRFS_EXTENT_ITEM_KEY);
1829                                 info = (struct btrfs_tree_block_info *)(ei + 1);
1830                                 *out_eiref =
1831                                    (struct btrfs_extent_inline_ref *)(info + 1);
1832                         }
1833                 } else {
1834                         *out_eiref = (struct btrfs_extent_inline_ref *)(ei + 1);
1835                 }
1836                 *ptr = (unsigned long)*out_eiref;
1837                 if ((unsigned long)(*ptr) >= (unsigned long)ei + item_size)
1838                         return -ENOENT;
1839         }
1840
1841         end = (unsigned long)ei + item_size;
1842         *out_eiref = (struct btrfs_extent_inline_ref *)(*ptr);
1843         *out_type = btrfs_get_extent_inline_ref_type(eb, *out_eiref,
1844                                                      BTRFS_REF_TYPE_ANY);
1845         if (*out_type == BTRFS_REF_TYPE_INVALID)
1846                 return -EUCLEAN;
1847
1848         *ptr += btrfs_extent_inline_ref_size(*out_type);
1849         WARN_ON(*ptr > end);
1850         if (*ptr == end)
1851                 return 1; /* last */
1852
1853         return 0;
1854 }
1855
1856 /*
1857  * reads the tree block backref for an extent. tree level and root are returned
1858  * through out_level and out_root. ptr must point to a 0 value for the first
1859  * call and may be modified (see get_extent_inline_ref comment).
1860  * returns 0 if data was provided, 1 if there was no more data to provide or
1861  * <0 on error.
1862  */
1863 int tree_backref_for_extent(unsigned long *ptr, struct extent_buffer *eb,
1864                             struct btrfs_key *key, struct btrfs_extent_item *ei,
1865                             u32 item_size, u64 *out_root, u8 *out_level)
1866 {
1867         int ret;
1868         int type;
1869         struct btrfs_extent_inline_ref *eiref;
1870
1871         if (*ptr == (unsigned long)-1)
1872                 return 1;
1873
1874         while (1) {
1875                 ret = get_extent_inline_ref(ptr, eb, key, ei, item_size,
1876                                               &eiref, &type);
1877                 if (ret < 0)
1878                         return ret;
1879
1880                 if (type == BTRFS_TREE_BLOCK_REF_KEY ||
1881                     type == BTRFS_SHARED_BLOCK_REF_KEY)
1882                         break;
1883
1884                 if (ret == 1)
1885                         return 1;
1886         }
1887
1888         /* we can treat both ref types equally here */
1889         *out_root = btrfs_extent_inline_ref_offset(eb, eiref);
1890
1891         if (key->type == BTRFS_EXTENT_ITEM_KEY) {
1892                 struct btrfs_tree_block_info *info;
1893
1894                 info = (struct btrfs_tree_block_info *)(ei + 1);
1895                 *out_level = btrfs_tree_block_level(eb, info);
1896         } else {
1897                 ASSERT(key->type == BTRFS_METADATA_ITEM_KEY);
1898                 *out_level = (u8)key->offset;
1899         }
1900
1901         if (ret == 1)
1902                 *ptr = (unsigned long)-1;
1903
1904         return 0;
1905 }
1906
1907 static int iterate_leaf_refs(struct btrfs_fs_info *fs_info,
1908                              struct extent_inode_elem *inode_list,
1909                              u64 root, u64 extent_item_objectid,
1910                              iterate_extent_inodes_t *iterate, void *ctx)
1911 {
1912         struct extent_inode_elem *eie;
1913         int ret = 0;
1914
1915         for (eie = inode_list; eie; eie = eie->next) {
1916                 btrfs_debug(fs_info,
1917                             "ref for %llu resolved, key (%llu EXTEND_DATA %llu), root %llu",
1918                             extent_item_objectid, eie->inum,
1919                             eie->offset, root);
1920                 ret = iterate(eie->inum, eie->offset, root, ctx);
1921                 if (ret) {
1922                         btrfs_debug(fs_info,
1923                                     "stopping iteration for %llu due to ret=%d",
1924                                     extent_item_objectid, ret);
1925                         break;
1926                 }
1927         }
1928
1929         return ret;
1930 }
1931
1932 /*
1933  * calls iterate() for every inode that references the extent identified by
1934  * the given parameters.
1935  * when the iterator function returns a non-zero value, iteration stops.
1936  */
1937 int iterate_extent_inodes(struct btrfs_fs_info *fs_info,
1938                                 u64 extent_item_objectid, u64 extent_item_pos,
1939                                 int search_commit_root,
1940                                 iterate_extent_inodes_t *iterate, void *ctx,
1941                                 bool ignore_offset)
1942 {
1943         int ret;
1944         struct btrfs_trans_handle *trans = NULL;
1945         struct ulist *refs = NULL;
1946         struct ulist *roots = NULL;
1947         struct ulist_node *ref_node = NULL;
1948         struct ulist_node *root_node = NULL;
1949         struct seq_list tree_mod_seq_elem = SEQ_LIST_INIT(tree_mod_seq_elem);
1950         struct ulist_iterator ref_uiter;
1951         struct ulist_iterator root_uiter;
1952
1953         btrfs_debug(fs_info, "resolving all inodes for extent %llu",
1954                         extent_item_objectid);
1955
1956         if (!search_commit_root) {
1957                 trans = btrfs_attach_transaction(fs_info->extent_root);
1958                 if (IS_ERR(trans)) {
1959                         if (PTR_ERR(trans) != -ENOENT &&
1960                             PTR_ERR(trans) != -EROFS)
1961                                 return PTR_ERR(trans);
1962                         trans = NULL;
1963                 }
1964         }
1965
1966         if (trans)
1967                 btrfs_get_tree_mod_seq(fs_info, &tree_mod_seq_elem);
1968         else
1969                 down_read(&fs_info->commit_root_sem);
1970
1971         ret = btrfs_find_all_leafs(trans, fs_info, extent_item_objectid,
1972                                    tree_mod_seq_elem.seq, &refs,
1973                                    &extent_item_pos, ignore_offset);
1974         if (ret)
1975                 goto out;
1976
1977         ULIST_ITER_INIT(&ref_uiter);
1978         while (!ret && (ref_node = ulist_next(refs, &ref_uiter))) {
1979                 ret = btrfs_find_all_roots_safe(trans, fs_info, ref_node->val,
1980                                                 tree_mod_seq_elem.seq, &roots,
1981                                                 ignore_offset);
1982                 if (ret)
1983                         break;
1984                 ULIST_ITER_INIT(&root_uiter);
1985                 while (!ret && (root_node = ulist_next(roots, &root_uiter))) {
1986                         btrfs_debug(fs_info,
1987                                     "root %llu references leaf %llu, data list %#llx",
1988                                     root_node->val, ref_node->val,
1989                                     ref_node->aux);
1990                         ret = iterate_leaf_refs(fs_info,
1991                                                 (struct extent_inode_elem *)
1992                                                 (uintptr_t)ref_node->aux,
1993                                                 root_node->val,
1994                                                 extent_item_objectid,
1995                                                 iterate, ctx);
1996                 }
1997                 ulist_free(roots);
1998         }
1999
2000         free_leaf_list(refs);
2001 out:
2002         if (trans) {
2003                 btrfs_put_tree_mod_seq(fs_info, &tree_mod_seq_elem);
2004                 btrfs_end_transaction(trans);
2005         } else {
2006                 up_read(&fs_info->commit_root_sem);
2007         }
2008
2009         return ret;
2010 }
2011
2012 int iterate_inodes_from_logical(u64 logical, struct btrfs_fs_info *fs_info,
2013                                 struct btrfs_path *path,
2014                                 iterate_extent_inodes_t *iterate, void *ctx,
2015                                 bool ignore_offset)
2016 {
2017         int ret;
2018         u64 extent_item_pos;
2019         u64 flags = 0;
2020         struct btrfs_key found_key;
2021         int search_commit_root = path->search_commit_root;
2022
2023         ret = extent_from_logical(fs_info, logical, path, &found_key, &flags);
2024         btrfs_release_path(path);
2025         if (ret < 0)
2026                 return ret;
2027         if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK)
2028                 return -EINVAL;
2029
2030         extent_item_pos = logical - found_key.objectid;
2031         ret = iterate_extent_inodes(fs_info, found_key.objectid,
2032                                         extent_item_pos, search_commit_root,
2033                                         iterate, ctx, ignore_offset);
2034
2035         return ret;
2036 }
2037
2038 typedef int (iterate_irefs_t)(u64 parent, u32 name_len, unsigned long name_off,
2039                               struct extent_buffer *eb, void *ctx);
2040
2041 static int iterate_inode_refs(u64 inum, struct btrfs_root *fs_root,
2042                               struct btrfs_path *path,
2043                               iterate_irefs_t *iterate, void *ctx)
2044 {
2045         int ret = 0;
2046         int slot;
2047         u32 cur;
2048         u32 len;
2049         u32 name_len;
2050         u64 parent = 0;
2051         int found = 0;
2052         struct extent_buffer *eb;
2053         struct btrfs_item *item;
2054         struct btrfs_inode_ref *iref;
2055         struct btrfs_key found_key;
2056
2057         while (!ret) {
2058                 ret = btrfs_find_item(fs_root, path, inum,
2059                                 parent ? parent + 1 : 0, BTRFS_INODE_REF_KEY,
2060                                 &found_key);
2061
2062                 if (ret < 0)
2063                         break;
2064                 if (ret) {
2065                         ret = found ? 0 : -ENOENT;
2066                         break;
2067                 }
2068                 ++found;
2069
2070                 parent = found_key.offset;
2071                 slot = path->slots[0];
2072                 eb = btrfs_clone_extent_buffer(path->nodes[0]);
2073                 if (!eb) {
2074                         ret = -ENOMEM;
2075                         break;
2076                 }
2077                 btrfs_release_path(path);
2078
2079                 item = btrfs_item_nr(slot);
2080                 iref = btrfs_item_ptr(eb, slot, struct btrfs_inode_ref);
2081
2082                 for (cur = 0; cur < btrfs_item_size(eb, item); cur += len) {
2083                         name_len = btrfs_inode_ref_name_len(eb, iref);
2084                         /* path must be released before calling iterate()! */
2085                         btrfs_debug(fs_root->fs_info,
2086                                 "following ref at offset %u for inode %llu in tree %llu",
2087                                 cur, found_key.objectid,
2088                                 fs_root->root_key.objectid);
2089                         ret = iterate(parent, name_len,
2090                                       (unsigned long)(iref + 1), eb, ctx);
2091                         if (ret)
2092                                 break;
2093                         len = sizeof(*iref) + name_len;
2094                         iref = (struct btrfs_inode_ref *)((char *)iref + len);
2095                 }
2096                 free_extent_buffer(eb);
2097         }
2098
2099         btrfs_release_path(path);
2100
2101         return ret;
2102 }
2103
2104 static int iterate_inode_extrefs(u64 inum, struct btrfs_root *fs_root,
2105                                  struct btrfs_path *path,
2106                                  iterate_irefs_t *iterate, void *ctx)
2107 {
2108         int ret;
2109         int slot;
2110         u64 offset = 0;
2111         u64 parent;
2112         int found = 0;
2113         struct extent_buffer *eb;
2114         struct btrfs_inode_extref *extref;
2115         u32 item_size;
2116         u32 cur_offset;
2117         unsigned long ptr;
2118
2119         while (1) {
2120                 ret = btrfs_find_one_extref(fs_root, inum, offset, path, &extref,
2121                                             &offset);
2122                 if (ret < 0)
2123                         break;
2124                 if (ret) {
2125                         ret = found ? 0 : -ENOENT;
2126                         break;
2127                 }
2128                 ++found;
2129
2130                 slot = path->slots[0];
2131                 eb = btrfs_clone_extent_buffer(path->nodes[0]);
2132                 if (!eb) {
2133                         ret = -ENOMEM;
2134                         break;
2135                 }
2136                 btrfs_release_path(path);
2137
2138                 item_size = btrfs_item_size_nr(eb, slot);
2139                 ptr = btrfs_item_ptr_offset(eb, slot);
2140                 cur_offset = 0;
2141
2142                 while (cur_offset < item_size) {
2143                         u32 name_len;
2144
2145                         extref = (struct btrfs_inode_extref *)(ptr + cur_offset);
2146                         parent = btrfs_inode_extref_parent(eb, extref);
2147                         name_len = btrfs_inode_extref_name_len(eb, extref);
2148                         ret = iterate(parent, name_len,
2149                                       (unsigned long)&extref->name, eb, ctx);
2150                         if (ret)
2151                                 break;
2152
2153                         cur_offset += btrfs_inode_extref_name_len(eb, extref);
2154                         cur_offset += sizeof(*extref);
2155                 }
2156                 free_extent_buffer(eb);
2157
2158                 offset++;
2159         }
2160
2161         btrfs_release_path(path);
2162
2163         return ret;
2164 }
2165
2166 static int iterate_irefs(u64 inum, struct btrfs_root *fs_root,
2167                          struct btrfs_path *path, iterate_irefs_t *iterate,
2168                          void *ctx)
2169 {
2170         int ret;
2171         int found_refs = 0;
2172
2173         ret = iterate_inode_refs(inum, fs_root, path, iterate, ctx);
2174         if (!ret)
2175                 ++found_refs;
2176         else if (ret != -ENOENT)
2177                 return ret;
2178
2179         ret = iterate_inode_extrefs(inum, fs_root, path, iterate, ctx);
2180         if (ret == -ENOENT && found_refs)
2181                 return 0;
2182
2183         return ret;
2184 }
2185
2186 /*
2187  * returns 0 if the path could be dumped (probably truncated)
2188  * returns <0 in case of an error
2189  */
2190 static int inode_to_path(u64 inum, u32 name_len, unsigned long name_off,
2191                          struct extent_buffer *eb, void *ctx)
2192 {
2193         struct inode_fs_paths *ipath = ctx;
2194         char *fspath;
2195         char *fspath_min;
2196         int i = ipath->fspath->elem_cnt;
2197         const int s_ptr = sizeof(char *);
2198         u32 bytes_left;
2199
2200         bytes_left = ipath->fspath->bytes_left > s_ptr ?
2201                                         ipath->fspath->bytes_left - s_ptr : 0;
2202
2203         fspath_min = (char *)ipath->fspath->val + (i + 1) * s_ptr;
2204         fspath = btrfs_ref_to_path(ipath->fs_root, ipath->btrfs_path, name_len,
2205                                    name_off, eb, inum, fspath_min, bytes_left);
2206         if (IS_ERR(fspath))
2207                 return PTR_ERR(fspath);
2208
2209         if (fspath > fspath_min) {
2210                 ipath->fspath->val[i] = (u64)(unsigned long)fspath;
2211                 ++ipath->fspath->elem_cnt;
2212                 ipath->fspath->bytes_left = fspath - fspath_min;
2213         } else {
2214                 ++ipath->fspath->elem_missed;
2215                 ipath->fspath->bytes_missing += fspath_min - fspath;
2216                 ipath->fspath->bytes_left = 0;
2217         }
2218
2219         return 0;
2220 }
2221
2222 /*
2223  * this dumps all file system paths to the inode into the ipath struct, provided
2224  * is has been created large enough. each path is zero-terminated and accessed
2225  * from ipath->fspath->val[i].
2226  * when it returns, there are ipath->fspath->elem_cnt number of paths available
2227  * in ipath->fspath->val[]. when the allocated space wasn't sufficient, the
2228  * number of missed paths is recorded in ipath->fspath->elem_missed, otherwise,
2229  * it's zero. ipath->fspath->bytes_missing holds the number of bytes that would
2230  * have been needed to return all paths.
2231  */
2232 int paths_from_inode(u64 inum, struct inode_fs_paths *ipath)
2233 {
2234         return iterate_irefs(inum, ipath->fs_root, ipath->btrfs_path,
2235                              inode_to_path, ipath);
2236 }
2237
2238 struct btrfs_data_container *init_data_container(u32 total_bytes)
2239 {
2240         struct btrfs_data_container *data;
2241         size_t alloc_bytes;
2242
2243         alloc_bytes = max_t(size_t, total_bytes, sizeof(*data));
2244         data = kvmalloc(alloc_bytes, GFP_KERNEL);
2245         if (!data)
2246                 return ERR_PTR(-ENOMEM);
2247
2248         if (total_bytes >= sizeof(*data)) {
2249                 data->bytes_left = total_bytes - sizeof(*data);
2250                 data->bytes_missing = 0;
2251         } else {
2252                 data->bytes_missing = sizeof(*data) - total_bytes;
2253                 data->bytes_left = 0;
2254         }
2255
2256         data->elem_cnt = 0;
2257         data->elem_missed = 0;
2258
2259         return data;
2260 }
2261
2262 /*
2263  * allocates space to return multiple file system paths for an inode.
2264  * total_bytes to allocate are passed, note that space usable for actual path
2265  * information will be total_bytes - sizeof(struct inode_fs_paths).
2266  * the returned pointer must be freed with free_ipath() in the end.
2267  */
2268 struct inode_fs_paths *init_ipath(s32 total_bytes, struct btrfs_root *fs_root,
2269                                         struct btrfs_path *path)
2270 {
2271         struct inode_fs_paths *ifp;
2272         struct btrfs_data_container *fspath;
2273
2274         fspath = init_data_container(total_bytes);
2275         if (IS_ERR(fspath))
2276                 return ERR_CAST(fspath);
2277
2278         ifp = kmalloc(sizeof(*ifp), GFP_KERNEL);
2279         if (!ifp) {
2280                 kvfree(fspath);
2281                 return ERR_PTR(-ENOMEM);
2282         }
2283
2284         ifp->btrfs_path = path;
2285         ifp->fspath = fspath;
2286         ifp->fs_root = fs_root;
2287
2288         return ifp;
2289 }
2290
2291 void free_ipath(struct inode_fs_paths *ipath)
2292 {
2293         if (!ipath)
2294                 return;
2295         kvfree(ipath->fspath);
2296         kfree(ipath);
2297 }