1 /* 2 * Copyright (C) 2007,2008 Oracle. All rights reserved. 3 * 4 * This program is free software; you can redistribute it and/or 5 * modify it under the terms of the GNU General Public 6 * License v2 as published by the Free Software Foundation. 7 * 8 * This program is distributed in the hope that it will be useful, 9 * but WITHOUT ANY WARRANTY; without even the implied warranty of 10 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU 11 * General Public License for more details. 12 * 13 * You should have received a copy of the GNU General Public 14 * License along with this program; if not, write to the 15 * Free Software Foundation, Inc., 59 Temple Place - Suite 330, 16 * Boston, MA 021110-1307, USA. 17 */ 18 19 #include <linux/sched.h> 20 #include <linux/slab.h> 21 #include <linux/rbtree.h> 22 #include <linux/vmalloc.h> 23 #include "ctree.h" 24 #include "disk-io.h" 25 #include "transaction.h" 26 #include "print-tree.h" 27 #include "locking.h" 28 29 static int split_node(struct btrfs_trans_handle *trans, struct btrfs_root 30 *root, struct btrfs_path *path, int level); 31 static int split_leaf(struct btrfs_trans_handle *trans, struct btrfs_root 32 *root, struct btrfs_key *ins_key, 33 struct btrfs_path *path, int data_size, int extend); 34 static int push_node_left(struct btrfs_trans_handle *trans, 35 struct btrfs_root *root, struct extent_buffer *dst, 36 struct extent_buffer *src, int empty); 37 static int balance_node_right(struct btrfs_trans_handle *trans, 38 struct btrfs_root *root, 39 struct extent_buffer *dst_buf, 40 struct extent_buffer *src_buf); 41 static void del_ptr(struct btrfs_root *root, struct btrfs_path *path, 42 int level, int slot); 43 static int tree_mod_log_free_eb(struct btrfs_fs_info *fs_info, 44 struct extent_buffer *eb); 45 46 struct btrfs_path *btrfs_alloc_path(void) 47 { 48 return kmem_cache_zalloc(btrfs_path_cachep, GFP_NOFS); 49 } 50 51 /* 52 * set all locked nodes in the path to blocking locks. This should 53 * be done before scheduling 54 */ 55 noinline void btrfs_set_path_blocking(struct btrfs_path *p) 56 { 57 int i; 58 for (i = 0; i < BTRFS_MAX_LEVEL; i++) { 59 if (!p->nodes[i] || !p->locks[i]) 60 continue; 61 btrfs_set_lock_blocking_rw(p->nodes[i], p->locks[i]); 62 if (p->locks[i] == BTRFS_READ_LOCK) 63 p->locks[i] = BTRFS_READ_LOCK_BLOCKING; 64 else if (p->locks[i] == BTRFS_WRITE_LOCK) 65 p->locks[i] = BTRFS_WRITE_LOCK_BLOCKING; 66 } 67 } 68 69 /* 70 * reset all the locked nodes in the patch to spinning locks. 71 * 72 * held is used to keep lockdep happy, when lockdep is enabled 73 * we set held to a blocking lock before we go around and 74 * retake all the spinlocks in the path. You can safely use NULL 75 * for held 76 */ 77 noinline void btrfs_clear_path_blocking(struct btrfs_path *p, 78 struct extent_buffer *held, int held_rw) 79 { 80 int i; 81 82 if (held) { 83 btrfs_set_lock_blocking_rw(held, held_rw); 84 if (held_rw == BTRFS_WRITE_LOCK) 85 held_rw = BTRFS_WRITE_LOCK_BLOCKING; 86 else if (held_rw == BTRFS_READ_LOCK) 87 held_rw = BTRFS_READ_LOCK_BLOCKING; 88 } 89 btrfs_set_path_blocking(p); 90 91 for (i = BTRFS_MAX_LEVEL - 1; i >= 0; i--) { 92 if (p->nodes[i] && p->locks[i]) { 93 btrfs_clear_lock_blocking_rw(p->nodes[i], p->locks[i]); 94 if (p->locks[i] == BTRFS_WRITE_LOCK_BLOCKING) 95 p->locks[i] = BTRFS_WRITE_LOCK; 96 else if (p->locks[i] == BTRFS_READ_LOCK_BLOCKING) 97 p->locks[i] = BTRFS_READ_LOCK; 98 } 99 } 100 101 if (held) 102 btrfs_clear_lock_blocking_rw(held, held_rw); 103 } 104 105 /* this also releases the path */ 106 void btrfs_free_path(struct btrfs_path *p) 107 { 108 if (!p) 109 return; 110 btrfs_release_path(p); 111 kmem_cache_free(btrfs_path_cachep, p); 112 } 113 114 /* 115 * path release drops references on the extent buffers in the path 116 * and it drops any locks held by this path 117 * 118 * It is safe to call this on paths that no locks or extent buffers held. 119 */ 120 noinline void btrfs_release_path(struct btrfs_path *p) 121 { 122 int i; 123 124 for (i = 0; i < BTRFS_MAX_LEVEL; i++) { 125 p->slots[i] = 0; 126 if (!p->nodes[i]) 127 continue; 128 if (p->locks[i]) { 129 btrfs_tree_unlock_rw(p->nodes[i], p->locks[i]); 130 p->locks[i] = 0; 131 } 132 free_extent_buffer(p->nodes[i]); 133 p->nodes[i] = NULL; 134 } 135 } 136 137 /* 138 * safely gets a reference on the root node of a tree. A lock 139 * is not taken, so a concurrent writer may put a different node 140 * at the root of the tree. See btrfs_lock_root_node for the 141 * looping required. 142 * 143 * The extent buffer returned by this has a reference taken, so 144 * it won't disappear. It may stop being the root of the tree 145 * at any time because there are no locks held. 146 */ 147 struct extent_buffer *btrfs_root_node(struct btrfs_root *root) 148 { 149 struct extent_buffer *eb; 150 151 while (1) { 152 rcu_read_lock(); 153 eb = rcu_dereference(root->node); 154 155 /* 156 * RCU really hurts here, we could free up the root node because 157 * it was COWed but we may not get the new root node yet so do 158 * the inc_not_zero dance and if it doesn't work then 159 * synchronize_rcu and try again. 160 */ 161 if (atomic_inc_not_zero(&eb->refs)) { 162 rcu_read_unlock(); 163 break; 164 } 165 rcu_read_unlock(); 166 synchronize_rcu(); 167 } 168 return eb; 169 } 170 171 /* loop around taking references on and locking the root node of the 172 * tree until you end up with a lock on the root. A locked buffer 173 * is returned, with a reference held. 174 */ 175 struct extent_buffer *btrfs_lock_root_node(struct btrfs_root *root) 176 { 177 struct extent_buffer *eb; 178 179 while (1) { 180 eb = btrfs_root_node(root); 181 btrfs_tree_lock(eb); 182 if (eb == root->node) 183 break; 184 btrfs_tree_unlock(eb); 185 free_extent_buffer(eb); 186 } 187 return eb; 188 } 189 190 /* loop around taking references on and locking the root node of the 191 * tree until you end up with a lock on the root. A locked buffer 192 * is returned, with a reference held. 193 */ 194 static struct extent_buffer *btrfs_read_lock_root_node(struct btrfs_root *root) 195 { 196 struct extent_buffer *eb; 197 198 while (1) { 199 eb = btrfs_root_node(root); 200 btrfs_tree_read_lock(eb); 201 if (eb == root->node) 202 break; 203 btrfs_tree_read_unlock(eb); 204 free_extent_buffer(eb); 205 } 206 return eb; 207 } 208 209 /* cowonly root (everything not a reference counted cow subvolume), just get 210 * put onto a simple dirty list. transaction.c walks this to make sure they 211 * get properly updated on disk. 212 */ 213 static void add_root_to_dirty_list(struct btrfs_root *root) 214 { 215 if (test_bit(BTRFS_ROOT_DIRTY, &root->state) || 216 !test_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state)) 217 return; 218 219 spin_lock(&root->fs_info->trans_lock); 220 if (!test_and_set_bit(BTRFS_ROOT_DIRTY, &root->state)) { 221 /* Want the extent tree to be the last on the list */ 222 if (root->objectid == BTRFS_EXTENT_TREE_OBJECTID) 223 list_move_tail(&root->dirty_list, 224 &root->fs_info->dirty_cowonly_roots); 225 else 226 list_move(&root->dirty_list, 227 &root->fs_info->dirty_cowonly_roots); 228 } 229 spin_unlock(&root->fs_info->trans_lock); 230 } 231 232 /* 233 * used by snapshot creation to make a copy of a root for a tree with 234 * a given objectid. The buffer with the new root node is returned in 235 * cow_ret, and this func returns zero on success or a negative error code. 236 */ 237 int btrfs_copy_root(struct btrfs_trans_handle *trans, 238 struct btrfs_root *root, 239 struct extent_buffer *buf, 240 struct extent_buffer **cow_ret, u64 new_root_objectid) 241 { 242 struct extent_buffer *cow; 243 int ret = 0; 244 int level; 245 struct btrfs_disk_key disk_key; 246 247 WARN_ON(test_bit(BTRFS_ROOT_REF_COWS, &root->state) && 248 trans->transid != root->fs_info->running_transaction->transid); 249 WARN_ON(test_bit(BTRFS_ROOT_REF_COWS, &root->state) && 250 trans->transid != root->last_trans); 251 252 level = btrfs_header_level(buf); 253 if (level == 0) 254 btrfs_item_key(buf, &disk_key, 0); 255 else 256 btrfs_node_key(buf, &disk_key, 0); 257 258 cow = btrfs_alloc_tree_block(trans, root, 0, new_root_objectid, 259 &disk_key, level, buf->start, 0); 260 if (IS_ERR(cow)) 261 return PTR_ERR(cow); 262 263 copy_extent_buffer(cow, buf, 0, 0, cow->len); 264 btrfs_set_header_bytenr(cow, cow->start); 265 btrfs_set_header_generation(cow, trans->transid); 266 btrfs_set_header_backref_rev(cow, BTRFS_MIXED_BACKREF_REV); 267 btrfs_clear_header_flag(cow, BTRFS_HEADER_FLAG_WRITTEN | 268 BTRFS_HEADER_FLAG_RELOC); 269 if (new_root_objectid == BTRFS_TREE_RELOC_OBJECTID) 270 btrfs_set_header_flag(cow, BTRFS_HEADER_FLAG_RELOC); 271 else 272 btrfs_set_header_owner(cow, new_root_objectid); 273 274 write_extent_buffer_fsid(cow, root->fs_info->fsid); 275 276 WARN_ON(btrfs_header_generation(buf) > trans->transid); 277 if (new_root_objectid == BTRFS_TREE_RELOC_OBJECTID) 278 ret = btrfs_inc_ref(trans, root, cow, 1); 279 else 280 ret = btrfs_inc_ref(trans, root, cow, 0); 281 282 if (ret) 283 return ret; 284 285 btrfs_mark_buffer_dirty(cow); 286 *cow_ret = cow; 287 return 0; 288 } 289 290 enum mod_log_op { 291 MOD_LOG_KEY_REPLACE, 292 MOD_LOG_KEY_ADD, 293 MOD_LOG_KEY_REMOVE, 294 MOD_LOG_KEY_REMOVE_WHILE_FREEING, 295 MOD_LOG_KEY_REMOVE_WHILE_MOVING, 296 MOD_LOG_MOVE_KEYS, 297 MOD_LOG_ROOT_REPLACE, 298 }; 299 300 struct tree_mod_move { 301 int dst_slot; 302 int nr_items; 303 }; 304 305 struct tree_mod_root { 306 u64 logical; 307 u8 level; 308 }; 309 310 struct tree_mod_elem { 311 struct rb_node node; 312 u64 logical; 313 u64 seq; 314 enum mod_log_op op; 315 316 /* this is used for MOD_LOG_KEY_* and MOD_LOG_MOVE_KEYS operations */ 317 int slot; 318 319 /* this is used for MOD_LOG_KEY* and MOD_LOG_ROOT_REPLACE */ 320 u64 generation; 321 322 /* those are used for op == MOD_LOG_KEY_{REPLACE,REMOVE} */ 323 struct btrfs_disk_key key; 324 u64 blockptr; 325 326 /* this is used for op == MOD_LOG_MOVE_KEYS */ 327 struct tree_mod_move move; 328 329 /* this is used for op == MOD_LOG_ROOT_REPLACE */ 330 struct tree_mod_root old_root; 331 }; 332 333 static inline void tree_mod_log_read_lock(struct btrfs_fs_info *fs_info) 334 { 335 read_lock(&fs_info->tree_mod_log_lock); 336 } 337 338 static inline void tree_mod_log_read_unlock(struct btrfs_fs_info *fs_info) 339 { 340 read_unlock(&fs_info->tree_mod_log_lock); 341 } 342 343 static inline void tree_mod_log_write_lock(struct btrfs_fs_info *fs_info) 344 { 345 write_lock(&fs_info->tree_mod_log_lock); 346 } 347 348 static inline void tree_mod_log_write_unlock(struct btrfs_fs_info *fs_info) 349 { 350 write_unlock(&fs_info->tree_mod_log_lock); 351 } 352 353 /* 354 * Pull a new tree mod seq number for our operation. 355 */ 356 static inline u64 btrfs_inc_tree_mod_seq(struct btrfs_fs_info *fs_info) 357 { 358 return atomic64_inc_return(&fs_info->tree_mod_seq); 359 } 360 361 /* 362 * This adds a new blocker to the tree mod log's blocker list if the @elem 363 * passed does not already have a sequence number set. So when a caller expects 364 * to record tree modifications, it should ensure to set elem->seq to zero 365 * before calling btrfs_get_tree_mod_seq. 366 * Returns a fresh, unused tree log modification sequence number, even if no new 367 * blocker was added. 368 */ 369 u64 btrfs_get_tree_mod_seq(struct btrfs_fs_info *fs_info, 370 struct seq_list *elem) 371 { 372 tree_mod_log_write_lock(fs_info); 373 spin_lock(&fs_info->tree_mod_seq_lock); 374 if (!elem->seq) { 375 elem->seq = btrfs_inc_tree_mod_seq(fs_info); 376 list_add_tail(&elem->list, &fs_info->tree_mod_seq_list); 377 } 378 spin_unlock(&fs_info->tree_mod_seq_lock); 379 tree_mod_log_write_unlock(fs_info); 380 381 return elem->seq; 382 } 383 384 void btrfs_put_tree_mod_seq(struct btrfs_fs_info *fs_info, 385 struct seq_list *elem) 386 { 387 struct rb_root *tm_root; 388 struct rb_node *node; 389 struct rb_node *next; 390 struct seq_list *cur_elem; 391 struct tree_mod_elem *tm; 392 u64 min_seq = (u64)-1; 393 u64 seq_putting = elem->seq; 394 395 if (!seq_putting) 396 return; 397 398 spin_lock(&fs_info->tree_mod_seq_lock); 399 list_del(&elem->list); 400 elem->seq = 0; 401 402 list_for_each_entry(cur_elem, &fs_info->tree_mod_seq_list, list) { 403 if (cur_elem->seq < min_seq) { 404 if (seq_putting > cur_elem->seq) { 405 /* 406 * blocker with lower sequence number exists, we 407 * cannot remove anything from the log 408 */ 409 spin_unlock(&fs_info->tree_mod_seq_lock); 410 return; 411 } 412 min_seq = cur_elem->seq; 413 } 414 } 415 spin_unlock(&fs_info->tree_mod_seq_lock); 416 417 /* 418 * anything that's lower than the lowest existing (read: blocked) 419 * sequence number can be removed from the tree. 420 */ 421 tree_mod_log_write_lock(fs_info); 422 tm_root = &fs_info->tree_mod_log; 423 for (node = rb_first(tm_root); node; node = next) { 424 next = rb_next(node); 425 tm = container_of(node, struct tree_mod_elem, node); 426 if (tm->seq > min_seq) 427 continue; 428 rb_erase(node, tm_root); 429 kfree(tm); 430 } 431 tree_mod_log_write_unlock(fs_info); 432 } 433 434 /* 435 * key order of the log: 436 * node/leaf start address -> sequence 437 * 438 * The 'start address' is the logical address of the *new* root node 439 * for root replace operations, or the logical address of the affected 440 * block for all other operations. 441 * 442 * Note: must be called with write lock (tree_mod_log_write_lock). 443 */ 444 static noinline int 445 __tree_mod_log_insert(struct btrfs_fs_info *fs_info, struct tree_mod_elem *tm) 446 { 447 struct rb_root *tm_root; 448 struct rb_node **new; 449 struct rb_node *parent = NULL; 450 struct tree_mod_elem *cur; 451 452 BUG_ON(!tm); 453 454 tm->seq = btrfs_inc_tree_mod_seq(fs_info); 455 456 tm_root = &fs_info->tree_mod_log; 457 new = &tm_root->rb_node; 458 while (*new) { 459 cur = container_of(*new, struct tree_mod_elem, node); 460 parent = *new; 461 if (cur->logical < tm->logical) 462 new = &((*new)->rb_left); 463 else if (cur->logical > tm->logical) 464 new = &((*new)->rb_right); 465 else if (cur->seq < tm->seq) 466 new = &((*new)->rb_left); 467 else if (cur->seq > tm->seq) 468 new = &((*new)->rb_right); 469 else 470 return -EEXIST; 471 } 472 473 rb_link_node(&tm->node, parent, new); 474 rb_insert_color(&tm->node, tm_root); 475 return 0; 476 } 477 478 /* 479 * Determines if logging can be omitted. Returns 1 if it can. Otherwise, it 480 * returns zero with the tree_mod_log_lock acquired. The caller must hold 481 * this until all tree mod log insertions are recorded in the rb tree and then 482 * call tree_mod_log_write_unlock() to release. 483 */ 484 static inline int tree_mod_dont_log(struct btrfs_fs_info *fs_info, 485 struct extent_buffer *eb) { 486 smp_mb(); 487 if (list_empty(&(fs_info)->tree_mod_seq_list)) 488 return 1; 489 if (eb && btrfs_header_level(eb) == 0) 490 return 1; 491 492 tree_mod_log_write_lock(fs_info); 493 if (list_empty(&(fs_info)->tree_mod_seq_list)) { 494 tree_mod_log_write_unlock(fs_info); 495 return 1; 496 } 497 498 return 0; 499 } 500 501 /* Similar to tree_mod_dont_log, but doesn't acquire any locks. */ 502 static inline int tree_mod_need_log(const struct btrfs_fs_info *fs_info, 503 struct extent_buffer *eb) 504 { 505 smp_mb(); 506 if (list_empty(&(fs_info)->tree_mod_seq_list)) 507 return 0; 508 if (eb && btrfs_header_level(eb) == 0) 509 return 0; 510 511 return 1; 512 } 513 514 static struct tree_mod_elem * 515 alloc_tree_mod_elem(struct extent_buffer *eb, int slot, 516 enum mod_log_op op, gfp_t flags) 517 { 518 struct tree_mod_elem *tm; 519 520 tm = kzalloc(sizeof(*tm), flags); 521 if (!tm) 522 return NULL; 523 524 tm->logical = eb->start; 525 if (op != MOD_LOG_KEY_ADD) { 526 btrfs_node_key(eb, &tm->key, slot); 527 tm->blockptr = btrfs_node_blockptr(eb, slot); 528 } 529 tm->op = op; 530 tm->slot = slot; 531 tm->generation = btrfs_node_ptr_generation(eb, slot); 532 RB_CLEAR_NODE(&tm->node); 533 534 return tm; 535 } 536 537 static noinline int 538 tree_mod_log_insert_key(struct btrfs_fs_info *fs_info, 539 struct extent_buffer *eb, int slot, 540 enum mod_log_op op, gfp_t flags) 541 { 542 struct tree_mod_elem *tm; 543 int ret; 544 545 if (!tree_mod_need_log(fs_info, eb)) 546 return 0; 547 548 tm = alloc_tree_mod_elem(eb, slot, op, flags); 549 if (!tm) 550 return -ENOMEM; 551 552 if (tree_mod_dont_log(fs_info, eb)) { 553 kfree(tm); 554 return 0; 555 } 556 557 ret = __tree_mod_log_insert(fs_info, tm); 558 tree_mod_log_write_unlock(fs_info); 559 if (ret) 560 kfree(tm); 561 562 return ret; 563 } 564 565 static noinline int 566 tree_mod_log_insert_move(struct btrfs_fs_info *fs_info, 567 struct extent_buffer *eb, int dst_slot, int src_slot, 568 int nr_items, gfp_t flags) 569 { 570 struct tree_mod_elem *tm = NULL; 571 struct tree_mod_elem **tm_list = NULL; 572 int ret = 0; 573 int i; 574 int locked = 0; 575 576 if (!tree_mod_need_log(fs_info, eb)) 577 return 0; 578 579 tm_list = kcalloc(nr_items, sizeof(struct tree_mod_elem *), flags); 580 if (!tm_list) 581 return -ENOMEM; 582 583 tm = kzalloc(sizeof(*tm), flags); 584 if (!tm) { 585 ret = -ENOMEM; 586 goto free_tms; 587 } 588 589 tm->logical = eb->start; 590 tm->slot = src_slot; 591 tm->move.dst_slot = dst_slot; 592 tm->move.nr_items = nr_items; 593 tm->op = MOD_LOG_MOVE_KEYS; 594 595 for (i = 0; i + dst_slot < src_slot && i < nr_items; i++) { 596 tm_list[i] = alloc_tree_mod_elem(eb, i + dst_slot, 597 MOD_LOG_KEY_REMOVE_WHILE_MOVING, flags); 598 if (!tm_list[i]) { 599 ret = -ENOMEM; 600 goto free_tms; 601 } 602 } 603 604 if (tree_mod_dont_log(fs_info, eb)) 605 goto free_tms; 606 locked = 1; 607 608 /* 609 * When we override something during the move, we log these removals. 610 * This can only happen when we move towards the beginning of the 611 * buffer, i.e. dst_slot < src_slot. 612 */ 613 for (i = 0; i + dst_slot < src_slot && i < nr_items; i++) { 614 ret = __tree_mod_log_insert(fs_info, tm_list[i]); 615 if (ret) 616 goto free_tms; 617 } 618 619 ret = __tree_mod_log_insert(fs_info, tm); 620 if (ret) 621 goto free_tms; 622 tree_mod_log_write_unlock(fs_info); 623 kfree(tm_list); 624 625 return 0; 626 free_tms: 627 for (i = 0; i < nr_items; i++) { 628 if (tm_list[i] && !RB_EMPTY_NODE(&tm_list[i]->node)) 629 rb_erase(&tm_list[i]->node, &fs_info->tree_mod_log); 630 kfree(tm_list[i]); 631 } 632 if (locked) 633 tree_mod_log_write_unlock(fs_info); 634 kfree(tm_list); 635 kfree(tm); 636 637 return ret; 638 } 639 640 static inline int 641 __tree_mod_log_free_eb(struct btrfs_fs_info *fs_info, 642 struct tree_mod_elem **tm_list, 643 int nritems) 644 { 645 int i, j; 646 int ret; 647 648 for (i = nritems - 1; i >= 0; i--) { 649 ret = __tree_mod_log_insert(fs_info, tm_list[i]); 650 if (ret) { 651 for (j = nritems - 1; j > i; j--) 652 rb_erase(&tm_list[j]->node, 653 &fs_info->tree_mod_log); 654 return ret; 655 } 656 } 657 658 return 0; 659 } 660 661 static noinline int 662 tree_mod_log_insert_root(struct btrfs_fs_info *fs_info, 663 struct extent_buffer *old_root, 664 struct extent_buffer *new_root, gfp_t flags, 665 int log_removal) 666 { 667 struct tree_mod_elem *tm = NULL; 668 struct tree_mod_elem **tm_list = NULL; 669 int nritems = 0; 670 int ret = 0; 671 int i; 672 673 if (!tree_mod_need_log(fs_info, NULL)) 674 return 0; 675 676 if (log_removal && btrfs_header_level(old_root) > 0) { 677 nritems = btrfs_header_nritems(old_root); 678 tm_list = kcalloc(nritems, sizeof(struct tree_mod_elem *), 679 flags); 680 if (!tm_list) { 681 ret = -ENOMEM; 682 goto free_tms; 683 } 684 for (i = 0; i < nritems; i++) { 685 tm_list[i] = alloc_tree_mod_elem(old_root, i, 686 MOD_LOG_KEY_REMOVE_WHILE_FREEING, flags); 687 if (!tm_list[i]) { 688 ret = -ENOMEM; 689 goto free_tms; 690 } 691 } 692 } 693 694 tm = kzalloc(sizeof(*tm), flags); 695 if (!tm) { 696 ret = -ENOMEM; 697 goto free_tms; 698 } 699 700 tm->logical = new_root->start; 701 tm->old_root.logical = old_root->start; 702 tm->old_root.level = btrfs_header_level(old_root); 703 tm->generation = btrfs_header_generation(old_root); 704 tm->op = MOD_LOG_ROOT_REPLACE; 705 706 if (tree_mod_dont_log(fs_info, NULL)) 707 goto free_tms; 708 709 if (tm_list) 710 ret = __tree_mod_log_free_eb(fs_info, tm_list, nritems); 711 if (!ret) 712 ret = __tree_mod_log_insert(fs_info, tm); 713 714 tree_mod_log_write_unlock(fs_info); 715 if (ret) 716 goto free_tms; 717 kfree(tm_list); 718 719 return ret; 720 721 free_tms: 722 if (tm_list) { 723 for (i = 0; i < nritems; i++) 724 kfree(tm_list[i]); 725 kfree(tm_list); 726 } 727 kfree(tm); 728 729 return ret; 730 } 731 732 static struct tree_mod_elem * 733 __tree_mod_log_search(struct btrfs_fs_info *fs_info, u64 start, u64 min_seq, 734 int smallest) 735 { 736 struct rb_root *tm_root; 737 struct rb_node *node; 738 struct tree_mod_elem *cur = NULL; 739 struct tree_mod_elem *found = NULL; 740 741 tree_mod_log_read_lock(fs_info); 742 tm_root = &fs_info->tree_mod_log; 743 node = tm_root->rb_node; 744 while (node) { 745 cur = container_of(node, struct tree_mod_elem, node); 746 if (cur->logical < start) { 747 node = node->rb_left; 748 } else if (cur->logical > start) { 749 node = node->rb_right; 750 } else if (cur->seq < min_seq) { 751 node = node->rb_left; 752 } else if (!smallest) { 753 /* we want the node with the highest seq */ 754 if (found) 755 BUG_ON(found->seq > cur->seq); 756 found = cur; 757 node = node->rb_left; 758 } else if (cur->seq > min_seq) { 759 /* we want the node with the smallest seq */ 760 if (found) 761 BUG_ON(found->seq < cur->seq); 762 found = cur; 763 node = node->rb_right; 764 } else { 765 found = cur; 766 break; 767 } 768 } 769 tree_mod_log_read_unlock(fs_info); 770 771 return found; 772 } 773 774 /* 775 * this returns the element from the log with the smallest time sequence 776 * value that's in the log (the oldest log item). any element with a time 777 * sequence lower than min_seq will be ignored. 778 */ 779 static struct tree_mod_elem * 780 tree_mod_log_search_oldest(struct btrfs_fs_info *fs_info, u64 start, 781 u64 min_seq) 782 { 783 return __tree_mod_log_search(fs_info, start, min_seq, 1); 784 } 785 786 /* 787 * this returns the element from the log with the largest time sequence 788 * value that's in the log (the most recent log item). any element with 789 * a time sequence lower than min_seq will be ignored. 790 */ 791 static struct tree_mod_elem * 792 tree_mod_log_search(struct btrfs_fs_info *fs_info, u64 start, u64 min_seq) 793 { 794 return __tree_mod_log_search(fs_info, start, min_seq, 0); 795 } 796 797 static noinline int 798 tree_mod_log_eb_copy(struct btrfs_fs_info *fs_info, struct extent_buffer *dst, 799 struct extent_buffer *src, unsigned long dst_offset, 800 unsigned long src_offset, int nr_items) 801 { 802 int ret = 0; 803 struct tree_mod_elem **tm_list = NULL; 804 struct tree_mod_elem **tm_list_add, **tm_list_rem; 805 int i; 806 int locked = 0; 807 808 if (!tree_mod_need_log(fs_info, NULL)) 809 return 0; 810 811 if (btrfs_header_level(dst) == 0 && btrfs_header_level(src) == 0) 812 return 0; 813 814 tm_list = kcalloc(nr_items * 2, sizeof(struct tree_mod_elem *), 815 GFP_NOFS); 816 if (!tm_list) 817 return -ENOMEM; 818 819 tm_list_add = tm_list; 820 tm_list_rem = tm_list + nr_items; 821 for (i = 0; i < nr_items; i++) { 822 tm_list_rem[i] = alloc_tree_mod_elem(src, i + src_offset, 823 MOD_LOG_KEY_REMOVE, GFP_NOFS); 824 if (!tm_list_rem[i]) { 825 ret = -ENOMEM; 826 goto free_tms; 827 } 828 829 tm_list_add[i] = alloc_tree_mod_elem(dst, i + dst_offset, 830 MOD_LOG_KEY_ADD, GFP_NOFS); 831 if (!tm_list_add[i]) { 832 ret = -ENOMEM; 833 goto free_tms; 834 } 835 } 836 837 if (tree_mod_dont_log(fs_info, NULL)) 838 goto free_tms; 839 locked = 1; 840 841 for (i = 0; i < nr_items; i++) { 842 ret = __tree_mod_log_insert(fs_info, tm_list_rem[i]); 843 if (ret) 844 goto free_tms; 845 ret = __tree_mod_log_insert(fs_info, tm_list_add[i]); 846 if (ret) 847 goto free_tms; 848 } 849 850 tree_mod_log_write_unlock(fs_info); 851 kfree(tm_list); 852 853 return 0; 854 855 free_tms: 856 for (i = 0; i < nr_items * 2; i++) { 857 if (tm_list[i] && !RB_EMPTY_NODE(&tm_list[i]->node)) 858 rb_erase(&tm_list[i]->node, &fs_info->tree_mod_log); 859 kfree(tm_list[i]); 860 } 861 if (locked) 862 tree_mod_log_write_unlock(fs_info); 863 kfree(tm_list); 864 865 return ret; 866 } 867 868 static inline void 869 tree_mod_log_eb_move(struct btrfs_fs_info *fs_info, struct extent_buffer *dst, 870 int dst_offset, int src_offset, int nr_items) 871 { 872 int ret; 873 ret = tree_mod_log_insert_move(fs_info, dst, dst_offset, src_offset, 874 nr_items, GFP_NOFS); 875 BUG_ON(ret < 0); 876 } 877 878 static noinline void 879 tree_mod_log_set_node_key(struct btrfs_fs_info *fs_info, 880 struct extent_buffer *eb, int slot, int atomic) 881 { 882 int ret; 883 884 ret = tree_mod_log_insert_key(fs_info, eb, slot, 885 MOD_LOG_KEY_REPLACE, 886 atomic ? GFP_ATOMIC : GFP_NOFS); 887 BUG_ON(ret < 0); 888 } 889 890 static noinline int 891 tree_mod_log_free_eb(struct btrfs_fs_info *fs_info, struct extent_buffer *eb) 892 { 893 struct tree_mod_elem **tm_list = NULL; 894 int nritems = 0; 895 int i; 896 int ret = 0; 897 898 if (btrfs_header_level(eb) == 0) 899 return 0; 900 901 if (!tree_mod_need_log(fs_info, NULL)) 902 return 0; 903 904 nritems = btrfs_header_nritems(eb); 905 tm_list = kcalloc(nritems, sizeof(struct tree_mod_elem *), GFP_NOFS); 906 if (!tm_list) 907 return -ENOMEM; 908 909 for (i = 0; i < nritems; i++) { 910 tm_list[i] = alloc_tree_mod_elem(eb, i, 911 MOD_LOG_KEY_REMOVE_WHILE_FREEING, GFP_NOFS); 912 if (!tm_list[i]) { 913 ret = -ENOMEM; 914 goto free_tms; 915 } 916 } 917 918 if (tree_mod_dont_log(fs_info, eb)) 919 goto free_tms; 920 921 ret = __tree_mod_log_free_eb(fs_info, tm_list, nritems); 922 tree_mod_log_write_unlock(fs_info); 923 if (ret) 924 goto free_tms; 925 kfree(tm_list); 926 927 return 0; 928 929 free_tms: 930 for (i = 0; i < nritems; i++) 931 kfree(tm_list[i]); 932 kfree(tm_list); 933 934 return ret; 935 } 936 937 static noinline void 938 tree_mod_log_set_root_pointer(struct btrfs_root *root, 939 struct extent_buffer *new_root_node, 940 int log_removal) 941 { 942 int ret; 943 ret = tree_mod_log_insert_root(root->fs_info, root->node, 944 new_root_node, GFP_NOFS, log_removal); 945 BUG_ON(ret < 0); 946 } 947 948 /* 949 * check if the tree block can be shared by multiple trees 950 */ 951 int btrfs_block_can_be_shared(struct btrfs_root *root, 952 struct extent_buffer *buf) 953 { 954 /* 955 * Tree blocks not in reference counted trees and tree roots 956 * are never shared. If a block was allocated after the last 957 * snapshot and the block was not allocated by tree relocation, 958 * we know the block is not shared. 959 */ 960 if (test_bit(BTRFS_ROOT_REF_COWS, &root->state) && 961 buf != root->node && buf != root->commit_root && 962 (btrfs_header_generation(buf) <= 963 btrfs_root_last_snapshot(&root->root_item) || 964 btrfs_header_flag(buf, BTRFS_HEADER_FLAG_RELOC))) 965 return 1; 966 #ifdef BTRFS_COMPAT_EXTENT_TREE_V0 967 if (test_bit(BTRFS_ROOT_REF_COWS, &root->state) && 968 btrfs_header_backref_rev(buf) < BTRFS_MIXED_BACKREF_REV) 969 return 1; 970 #endif 971 return 0; 972 } 973 974 static noinline int update_ref_for_cow(struct btrfs_trans_handle *trans, 975 struct btrfs_root *root, 976 struct extent_buffer *buf, 977 struct extent_buffer *cow, 978 int *last_ref) 979 { 980 u64 refs; 981 u64 owner; 982 u64 flags; 983 u64 new_flags = 0; 984 int ret; 985 986 /* 987 * Backrefs update rules: 988 * 989 * Always use full backrefs for extent pointers in tree block 990 * allocated by tree relocation. 991 * 992 * If a shared tree block is no longer referenced by its owner 993 * tree (btrfs_header_owner(buf) == root->root_key.objectid), 994 * use full backrefs for extent pointers in tree block. 995 * 996 * If a tree block is been relocating 997 * (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID), 998 * use full backrefs for extent pointers in tree block. 999 * The reason for this is some operations (such as drop tree) 1000 * are only allowed for blocks use full backrefs. 1001 */ 1002 1003 if (btrfs_block_can_be_shared(root, buf)) { 1004 ret = btrfs_lookup_extent_info(trans, root, buf->start, 1005 btrfs_header_level(buf), 1, 1006 &refs, &flags); 1007 if (ret) 1008 return ret; 1009 if (refs == 0) { 1010 ret = -EROFS; 1011 btrfs_handle_fs_error(root->fs_info, ret, NULL); 1012 return ret; 1013 } 1014 } else { 1015 refs = 1; 1016 if (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID || 1017 btrfs_header_backref_rev(buf) < BTRFS_MIXED_BACKREF_REV) 1018 flags = BTRFS_BLOCK_FLAG_FULL_BACKREF; 1019 else 1020 flags = 0; 1021 } 1022 1023 owner = btrfs_header_owner(buf); 1024 BUG_ON(owner == BTRFS_TREE_RELOC_OBJECTID && 1025 !(flags & BTRFS_BLOCK_FLAG_FULL_BACKREF)); 1026 1027 if (refs > 1) { 1028 if ((owner == root->root_key.objectid || 1029 root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID) && 1030 !(flags & BTRFS_BLOCK_FLAG_FULL_BACKREF)) { 1031 ret = btrfs_inc_ref(trans, root, buf, 1); 1032 BUG_ON(ret); /* -ENOMEM */ 1033 1034 if (root->root_key.objectid == 1035 BTRFS_TREE_RELOC_OBJECTID) { 1036 ret = btrfs_dec_ref(trans, root, buf, 0); 1037 BUG_ON(ret); /* -ENOMEM */ 1038 ret = btrfs_inc_ref(trans, root, cow, 1); 1039 BUG_ON(ret); /* -ENOMEM */ 1040 } 1041 new_flags |= BTRFS_BLOCK_FLAG_FULL_BACKREF; 1042 } else { 1043 1044 if (root->root_key.objectid == 1045 BTRFS_TREE_RELOC_OBJECTID) 1046 ret = btrfs_inc_ref(trans, root, cow, 1); 1047 else 1048 ret = btrfs_inc_ref(trans, root, cow, 0); 1049 BUG_ON(ret); /* -ENOMEM */ 1050 } 1051 if (new_flags != 0) { 1052 int level = btrfs_header_level(buf); 1053 1054 ret = btrfs_set_disk_extent_flags(trans, root, 1055 buf->start, 1056 buf->len, 1057 new_flags, level, 0); 1058 if (ret) 1059 return ret; 1060 } 1061 } else { 1062 if (flags & BTRFS_BLOCK_FLAG_FULL_BACKREF) { 1063 if (root->root_key.objectid == 1064 BTRFS_TREE_RELOC_OBJECTID) 1065 ret = btrfs_inc_ref(trans, root, cow, 1); 1066 else 1067 ret = btrfs_inc_ref(trans, root, cow, 0); 1068 BUG_ON(ret); /* -ENOMEM */ 1069 ret = btrfs_dec_ref(trans, root, buf, 1); 1070 BUG_ON(ret); /* -ENOMEM */ 1071 } 1072 clean_tree_block(trans, root->fs_info, buf); 1073 *last_ref = 1; 1074 } 1075 return 0; 1076 } 1077 1078 /* 1079 * does the dirty work in cow of a single block. The parent block (if 1080 * supplied) is updated to point to the new cow copy. The new buffer is marked 1081 * dirty and returned locked. If you modify the block it needs to be marked 1082 * dirty again. 1083 * 1084 * search_start -- an allocation hint for the new block 1085 * 1086 * empty_size -- a hint that you plan on doing more cow. This is the size in 1087 * bytes the allocator should try to find free next to the block it returns. 1088 * This is just a hint and may be ignored by the allocator. 1089 */ 1090 static noinline int __btrfs_cow_block(struct btrfs_trans_handle *trans, 1091 struct btrfs_root *root, 1092 struct extent_buffer *buf, 1093 struct extent_buffer *parent, int parent_slot, 1094 struct extent_buffer **cow_ret, 1095 u64 search_start, u64 empty_size) 1096 { 1097 struct btrfs_disk_key disk_key; 1098 struct extent_buffer *cow; 1099 int level, ret; 1100 int last_ref = 0; 1101 int unlock_orig = 0; 1102 u64 parent_start = 0; 1103 1104 if (*cow_ret == buf) 1105 unlock_orig = 1; 1106 1107 btrfs_assert_tree_locked(buf); 1108 1109 WARN_ON(test_bit(BTRFS_ROOT_REF_COWS, &root->state) && 1110 trans->transid != root->fs_info->running_transaction->transid); 1111 WARN_ON(test_bit(BTRFS_ROOT_REF_COWS, &root->state) && 1112 trans->transid != root->last_trans); 1113 1114 level = btrfs_header_level(buf); 1115 1116 if (level == 0) 1117 btrfs_item_key(buf, &disk_key, 0); 1118 else 1119 btrfs_node_key(buf, &disk_key, 0); 1120 1121 if ((root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID) && parent) 1122 parent_start = parent->start; 1123 1124 cow = btrfs_alloc_tree_block(trans, root, parent_start, 1125 root->root_key.objectid, &disk_key, level, 1126 search_start, empty_size); 1127 if (IS_ERR(cow)) 1128 return PTR_ERR(cow); 1129 1130 /* cow is set to blocking by btrfs_init_new_buffer */ 1131 1132 copy_extent_buffer(cow, buf, 0, 0, cow->len); 1133 btrfs_set_header_bytenr(cow, cow->start); 1134 btrfs_set_header_generation(cow, trans->transid); 1135 btrfs_set_header_backref_rev(cow, BTRFS_MIXED_BACKREF_REV); 1136 btrfs_clear_header_flag(cow, BTRFS_HEADER_FLAG_WRITTEN | 1137 BTRFS_HEADER_FLAG_RELOC); 1138 if (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID) 1139 btrfs_set_header_flag(cow, BTRFS_HEADER_FLAG_RELOC); 1140 else 1141 btrfs_set_header_owner(cow, root->root_key.objectid); 1142 1143 write_extent_buffer_fsid(cow, root->fs_info->fsid); 1144 1145 ret = update_ref_for_cow(trans, root, buf, cow, &last_ref); 1146 if (ret) { 1147 btrfs_abort_transaction(trans, ret); 1148 return ret; 1149 } 1150 1151 if (test_bit(BTRFS_ROOT_REF_COWS, &root->state)) { 1152 ret = btrfs_reloc_cow_block(trans, root, buf, cow); 1153 if (ret) { 1154 btrfs_abort_transaction(trans, ret); 1155 return ret; 1156 } 1157 } 1158 1159 if (buf == root->node) { 1160 WARN_ON(parent && parent != buf); 1161 if (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID || 1162 btrfs_header_backref_rev(buf) < BTRFS_MIXED_BACKREF_REV) 1163 parent_start = buf->start; 1164 1165 extent_buffer_get(cow); 1166 tree_mod_log_set_root_pointer(root, cow, 1); 1167 rcu_assign_pointer(root->node, cow); 1168 1169 btrfs_free_tree_block(trans, root, buf, parent_start, 1170 last_ref); 1171 free_extent_buffer(buf); 1172 add_root_to_dirty_list(root); 1173 } else { 1174 WARN_ON(trans->transid != btrfs_header_generation(parent)); 1175 tree_mod_log_insert_key(root->fs_info, parent, parent_slot, 1176 MOD_LOG_KEY_REPLACE, GFP_NOFS); 1177 btrfs_set_node_blockptr(parent, parent_slot, 1178 cow->start); 1179 btrfs_set_node_ptr_generation(parent, parent_slot, 1180 trans->transid); 1181 btrfs_mark_buffer_dirty(parent); 1182 if (last_ref) { 1183 ret = tree_mod_log_free_eb(root->fs_info, buf); 1184 if (ret) { 1185 btrfs_abort_transaction(trans, ret); 1186 return ret; 1187 } 1188 } 1189 btrfs_free_tree_block(trans, root, buf, parent_start, 1190 last_ref); 1191 } 1192 if (unlock_orig) 1193 btrfs_tree_unlock(buf); 1194 free_extent_buffer_stale(buf); 1195 btrfs_mark_buffer_dirty(cow); 1196 *cow_ret = cow; 1197 return 0; 1198 } 1199 1200 /* 1201 * returns the logical address of the oldest predecessor of the given root. 1202 * entries older than time_seq are ignored. 1203 */ 1204 static struct tree_mod_elem * 1205 __tree_mod_log_oldest_root(struct btrfs_fs_info *fs_info, 1206 struct extent_buffer *eb_root, u64 time_seq) 1207 { 1208 struct tree_mod_elem *tm; 1209 struct tree_mod_elem *found = NULL; 1210 u64 root_logical = eb_root->start; 1211 int looped = 0; 1212 1213 if (!time_seq) 1214 return NULL; 1215 1216 /* 1217 * the very last operation that's logged for a root is the 1218 * replacement operation (if it is replaced at all). this has 1219 * the logical address of the *new* root, making it the very 1220 * first operation that's logged for this root. 1221 */ 1222 while (1) { 1223 tm = tree_mod_log_search_oldest(fs_info, root_logical, 1224 time_seq); 1225 if (!looped && !tm) 1226 return NULL; 1227 /* 1228 * if there are no tree operation for the oldest root, we simply 1229 * return it. this should only happen if that (old) root is at 1230 * level 0. 1231 */ 1232 if (!tm) 1233 break; 1234 1235 /* 1236 * if there's an operation that's not a root replacement, we 1237 * found the oldest version of our root. normally, we'll find a 1238 * MOD_LOG_KEY_REMOVE_WHILE_FREEING operation here. 1239 */ 1240 if (tm->op != MOD_LOG_ROOT_REPLACE) 1241 break; 1242 1243 found = tm; 1244 root_logical = tm->old_root.logical; 1245 looped = 1; 1246 } 1247 1248 /* if there's no old root to return, return what we found instead */ 1249 if (!found) 1250 found = tm; 1251 1252 return found; 1253 } 1254 1255 /* 1256 * tm is a pointer to the first operation to rewind within eb. then, all 1257 * previous operations will be rewound (until we reach something older than 1258 * time_seq). 1259 */ 1260 static void 1261 __tree_mod_log_rewind(struct btrfs_fs_info *fs_info, struct extent_buffer *eb, 1262 u64 time_seq, struct tree_mod_elem *first_tm) 1263 { 1264 u32 n; 1265 struct rb_node *next; 1266 struct tree_mod_elem *tm = first_tm; 1267 unsigned long o_dst; 1268 unsigned long o_src; 1269 unsigned long p_size = sizeof(struct btrfs_key_ptr); 1270 1271 n = btrfs_header_nritems(eb); 1272 tree_mod_log_read_lock(fs_info); 1273 while (tm && tm->seq >= time_seq) { 1274 /* 1275 * all the operations are recorded with the operator used for 1276 * the modification. as we're going backwards, we do the 1277 * opposite of each operation here. 1278 */ 1279 switch (tm->op) { 1280 case MOD_LOG_KEY_REMOVE_WHILE_FREEING: 1281 BUG_ON(tm->slot < n); 1282 /* Fallthrough */ 1283 case MOD_LOG_KEY_REMOVE_WHILE_MOVING: 1284 case MOD_LOG_KEY_REMOVE: 1285 btrfs_set_node_key(eb, &tm->key, tm->slot); 1286 btrfs_set_node_blockptr(eb, tm->slot, tm->blockptr); 1287 btrfs_set_node_ptr_generation(eb, tm->slot, 1288 tm->generation); 1289 n++; 1290 break; 1291 case MOD_LOG_KEY_REPLACE: 1292 BUG_ON(tm->slot >= n); 1293 btrfs_set_node_key(eb, &tm->key, tm->slot); 1294 btrfs_set_node_blockptr(eb, tm->slot, tm->blockptr); 1295 btrfs_set_node_ptr_generation(eb, tm->slot, 1296 tm->generation); 1297 break; 1298 case MOD_LOG_KEY_ADD: 1299 /* if a move operation is needed it's in the log */ 1300 n--; 1301 break; 1302 case MOD_LOG_MOVE_KEYS: 1303 o_dst = btrfs_node_key_ptr_offset(tm->slot); 1304 o_src = btrfs_node_key_ptr_offset(tm->move.dst_slot); 1305 memmove_extent_buffer(eb, o_dst, o_src, 1306 tm->move.nr_items * p_size); 1307 break; 1308 case MOD_LOG_ROOT_REPLACE: 1309 /* 1310 * this operation is special. for roots, this must be 1311 * handled explicitly before rewinding. 1312 * for non-roots, this operation may exist if the node 1313 * was a root: root A -> child B; then A gets empty and 1314 * B is promoted to the new root. in the mod log, we'll 1315 * have a root-replace operation for B, a tree block 1316 * that is no root. we simply ignore that operation. 1317 */ 1318 break; 1319 } 1320 next = rb_next(&tm->node); 1321 if (!next) 1322 break; 1323 tm = container_of(next, struct tree_mod_elem, node); 1324 if (tm->logical != first_tm->logical) 1325 break; 1326 } 1327 tree_mod_log_read_unlock(fs_info); 1328 btrfs_set_header_nritems(eb, n); 1329 } 1330 1331 /* 1332 * Called with eb read locked. If the buffer cannot be rewound, the same buffer 1333 * is returned. If rewind operations happen, a fresh buffer is returned. The 1334 * returned buffer is always read-locked. If the returned buffer is not the 1335 * input buffer, the lock on the input buffer is released and the input buffer 1336 * is freed (its refcount is decremented). 1337 */ 1338 static struct extent_buffer * 1339 tree_mod_log_rewind(struct btrfs_fs_info *fs_info, struct btrfs_path *path, 1340 struct extent_buffer *eb, u64 time_seq) 1341 { 1342 struct extent_buffer *eb_rewin; 1343 struct tree_mod_elem *tm; 1344 1345 if (!time_seq) 1346 return eb; 1347 1348 if (btrfs_header_level(eb) == 0) 1349 return eb; 1350 1351 tm = tree_mod_log_search(fs_info, eb->start, time_seq); 1352 if (!tm) 1353 return eb; 1354 1355 btrfs_set_path_blocking(path); 1356 btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK); 1357 1358 if (tm->op == MOD_LOG_KEY_REMOVE_WHILE_FREEING) { 1359 BUG_ON(tm->slot != 0); 1360 eb_rewin = alloc_dummy_extent_buffer(fs_info, eb->start, 1361 eb->len); 1362 if (!eb_rewin) { 1363 btrfs_tree_read_unlock_blocking(eb); 1364 free_extent_buffer(eb); 1365 return NULL; 1366 } 1367 btrfs_set_header_bytenr(eb_rewin, eb->start); 1368 btrfs_set_header_backref_rev(eb_rewin, 1369 btrfs_header_backref_rev(eb)); 1370 btrfs_set_header_owner(eb_rewin, btrfs_header_owner(eb)); 1371 btrfs_set_header_level(eb_rewin, btrfs_header_level(eb)); 1372 } else { 1373 eb_rewin = btrfs_clone_extent_buffer(eb); 1374 if (!eb_rewin) { 1375 btrfs_tree_read_unlock_blocking(eb); 1376 free_extent_buffer(eb); 1377 return NULL; 1378 } 1379 } 1380 1381 btrfs_clear_path_blocking(path, NULL, BTRFS_READ_LOCK); 1382 btrfs_tree_read_unlock_blocking(eb); 1383 free_extent_buffer(eb); 1384 1385 extent_buffer_get(eb_rewin); 1386 btrfs_tree_read_lock(eb_rewin); 1387 __tree_mod_log_rewind(fs_info, eb_rewin, time_seq, tm); 1388 WARN_ON(btrfs_header_nritems(eb_rewin) > 1389 BTRFS_NODEPTRS_PER_BLOCK(fs_info->tree_root)); 1390 1391 return eb_rewin; 1392 } 1393 1394 /* 1395 * get_old_root() rewinds the state of @root's root node to the given @time_seq 1396 * value. If there are no changes, the current root->root_node is returned. If 1397 * anything changed in between, there's a fresh buffer allocated on which the 1398 * rewind operations are done. In any case, the returned buffer is read locked. 1399 * Returns NULL on error (with no locks held). 1400 */ 1401 static inline struct extent_buffer * 1402 get_old_root(struct btrfs_root *root, u64 time_seq) 1403 { 1404 struct tree_mod_elem *tm; 1405 struct extent_buffer *eb = NULL; 1406 struct extent_buffer *eb_root; 1407 struct extent_buffer *old; 1408 struct tree_mod_root *old_root = NULL; 1409 u64 old_generation = 0; 1410 u64 logical; 1411 1412 eb_root = btrfs_read_lock_root_node(root); 1413 tm = __tree_mod_log_oldest_root(root->fs_info, eb_root, time_seq); 1414 if (!tm) 1415 return eb_root; 1416 1417 if (tm->op == MOD_LOG_ROOT_REPLACE) { 1418 old_root = &tm->old_root; 1419 old_generation = tm->generation; 1420 logical = old_root->logical; 1421 } else { 1422 logical = eb_root->start; 1423 } 1424 1425 tm = tree_mod_log_search(root->fs_info, logical, time_seq); 1426 if (old_root && tm && tm->op != MOD_LOG_KEY_REMOVE_WHILE_FREEING) { 1427 btrfs_tree_read_unlock(eb_root); 1428 free_extent_buffer(eb_root); 1429 old = read_tree_block(root, logical, 0); 1430 if (WARN_ON(IS_ERR(old) || !extent_buffer_uptodate(old))) { 1431 if (!IS_ERR(old)) 1432 free_extent_buffer(old); 1433 btrfs_warn(root->fs_info, 1434 "failed to read tree block %llu from get_old_root", logical); 1435 } else { 1436 eb = btrfs_clone_extent_buffer(old); 1437 free_extent_buffer(old); 1438 } 1439 } else if (old_root) { 1440 btrfs_tree_read_unlock(eb_root); 1441 free_extent_buffer(eb_root); 1442 eb = alloc_dummy_extent_buffer(root->fs_info, logical, 1443 root->nodesize); 1444 } else { 1445 btrfs_set_lock_blocking_rw(eb_root, BTRFS_READ_LOCK); 1446 eb = btrfs_clone_extent_buffer(eb_root); 1447 btrfs_tree_read_unlock_blocking(eb_root); 1448 free_extent_buffer(eb_root); 1449 } 1450 1451 if (!eb) 1452 return NULL; 1453 extent_buffer_get(eb); 1454 btrfs_tree_read_lock(eb); 1455 if (old_root) { 1456 btrfs_set_header_bytenr(eb, eb->start); 1457 btrfs_set_header_backref_rev(eb, BTRFS_MIXED_BACKREF_REV); 1458 btrfs_set_header_owner(eb, btrfs_header_owner(eb_root)); 1459 btrfs_set_header_level(eb, old_root->level); 1460 btrfs_set_header_generation(eb, old_generation); 1461 } 1462 if (tm) 1463 __tree_mod_log_rewind(root->fs_info, eb, time_seq, tm); 1464 else 1465 WARN_ON(btrfs_header_level(eb) != 0); 1466 WARN_ON(btrfs_header_nritems(eb) > BTRFS_NODEPTRS_PER_BLOCK(root)); 1467 1468 return eb; 1469 } 1470 1471 int btrfs_old_root_level(struct btrfs_root *root, u64 time_seq) 1472 { 1473 struct tree_mod_elem *tm; 1474 int level; 1475 struct extent_buffer *eb_root = btrfs_root_node(root); 1476 1477 tm = __tree_mod_log_oldest_root(root->fs_info, eb_root, time_seq); 1478 if (tm && tm->op == MOD_LOG_ROOT_REPLACE) { 1479 level = tm->old_root.level; 1480 } else { 1481 level = btrfs_header_level(eb_root); 1482 } 1483 free_extent_buffer(eb_root); 1484 1485 return level; 1486 } 1487 1488 static inline int should_cow_block(struct btrfs_trans_handle *trans, 1489 struct btrfs_root *root, 1490 struct extent_buffer *buf) 1491 { 1492 if (btrfs_is_testing(root->fs_info)) 1493 return 0; 1494 1495 /* ensure we can see the force_cow */ 1496 smp_rmb(); 1497 1498 /* 1499 * We do not need to cow a block if 1500 * 1) this block is not created or changed in this transaction; 1501 * 2) this block does not belong to TREE_RELOC tree; 1502 * 3) the root is not forced COW. 1503 * 1504 * What is forced COW: 1505 * when we create snapshot during committing the transaction, 1506 * after we've finished coping src root, we must COW the shared 1507 * block to ensure the metadata consistency. 1508 */ 1509 if (btrfs_header_generation(buf) == trans->transid && 1510 !btrfs_header_flag(buf, BTRFS_HEADER_FLAG_WRITTEN) && 1511 !(root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID && 1512 btrfs_header_flag(buf, BTRFS_HEADER_FLAG_RELOC)) && 1513 !test_bit(BTRFS_ROOT_FORCE_COW, &root->state)) 1514 return 0; 1515 return 1; 1516 } 1517 1518 /* 1519 * cows a single block, see __btrfs_cow_block for the real work. 1520 * This version of it has extra checks so that a block isn't COWed more than 1521 * once per transaction, as long as it hasn't been written yet 1522 */ 1523 noinline int btrfs_cow_block(struct btrfs_trans_handle *trans, 1524 struct btrfs_root *root, struct extent_buffer *buf, 1525 struct extent_buffer *parent, int parent_slot, 1526 struct extent_buffer **cow_ret) 1527 { 1528 u64 search_start; 1529 int ret; 1530 1531 if (trans->transaction != root->fs_info->running_transaction) 1532 WARN(1, KERN_CRIT "trans %llu running %llu\n", 1533 trans->transid, 1534 root->fs_info->running_transaction->transid); 1535 1536 if (trans->transid != root->fs_info->generation) 1537 WARN(1, KERN_CRIT "trans %llu running %llu\n", 1538 trans->transid, root->fs_info->generation); 1539 1540 if (!should_cow_block(trans, root, buf)) { 1541 trans->dirty = true; 1542 *cow_ret = buf; 1543 return 0; 1544 } 1545 1546 search_start = buf->start & ~((u64)SZ_1G - 1); 1547 1548 if (parent) 1549 btrfs_set_lock_blocking(parent); 1550 btrfs_set_lock_blocking(buf); 1551 1552 ret = __btrfs_cow_block(trans, root, buf, parent, 1553 parent_slot, cow_ret, search_start, 0); 1554 1555 trace_btrfs_cow_block(root, buf, *cow_ret); 1556 1557 return ret; 1558 } 1559 1560 /* 1561 * helper function for defrag to decide if two blocks pointed to by a 1562 * node are actually close by 1563 */ 1564 static int close_blocks(u64 blocknr, u64 other, u32 blocksize) 1565 { 1566 if (blocknr < other && other - (blocknr + blocksize) < 32768) 1567 return 1; 1568 if (blocknr > other && blocknr - (other + blocksize) < 32768) 1569 return 1; 1570 return 0; 1571 } 1572 1573 /* 1574 * compare two keys in a memcmp fashion 1575 */ 1576 static int comp_keys(struct btrfs_disk_key *disk, struct btrfs_key *k2) 1577 { 1578 struct btrfs_key k1; 1579 1580 btrfs_disk_key_to_cpu(&k1, disk); 1581 1582 return btrfs_comp_cpu_keys(&k1, k2); 1583 } 1584 1585 /* 1586 * same as comp_keys only with two btrfs_key's 1587 */ 1588 int btrfs_comp_cpu_keys(struct btrfs_key *k1, struct btrfs_key *k2) 1589 { 1590 if (k1->objectid > k2->objectid) 1591 return 1; 1592 if (k1->objectid < k2->objectid) 1593 return -1; 1594 if (k1->type > k2->type) 1595 return 1; 1596 if (k1->type < k2->type) 1597 return -1; 1598 if (k1->offset > k2->offset) 1599 return 1; 1600 if (k1->offset < k2->offset) 1601 return -1; 1602 return 0; 1603 } 1604 1605 /* 1606 * this is used by the defrag code to go through all the 1607 * leaves pointed to by a node and reallocate them so that 1608 * disk order is close to key order 1609 */ 1610 int btrfs_realloc_node(struct btrfs_trans_handle *trans, 1611 struct btrfs_root *root, struct extent_buffer *parent, 1612 int start_slot, u64 *last_ret, 1613 struct btrfs_key *progress) 1614 { 1615 struct extent_buffer *cur; 1616 u64 blocknr; 1617 u64 gen; 1618 u64 search_start = *last_ret; 1619 u64 last_block = 0; 1620 u64 other; 1621 u32 parent_nritems; 1622 int end_slot; 1623 int i; 1624 int err = 0; 1625 int parent_level; 1626 int uptodate; 1627 u32 blocksize; 1628 int progress_passed = 0; 1629 struct btrfs_disk_key disk_key; 1630 1631 parent_level = btrfs_header_level(parent); 1632 1633 WARN_ON(trans->transaction != root->fs_info->running_transaction); 1634 WARN_ON(trans->transid != root->fs_info->generation); 1635 1636 parent_nritems = btrfs_header_nritems(parent); 1637 blocksize = root->nodesize; 1638 end_slot = parent_nritems - 1; 1639 1640 if (parent_nritems <= 1) 1641 return 0; 1642 1643 btrfs_set_lock_blocking(parent); 1644 1645 for (i = start_slot; i <= end_slot; i++) { 1646 int close = 1; 1647 1648 btrfs_node_key(parent, &disk_key, i); 1649 if (!progress_passed && comp_keys(&disk_key, progress) < 0) 1650 continue; 1651 1652 progress_passed = 1; 1653 blocknr = btrfs_node_blockptr(parent, i); 1654 gen = btrfs_node_ptr_generation(parent, i); 1655 if (last_block == 0) 1656 last_block = blocknr; 1657 1658 if (i > 0) { 1659 other = btrfs_node_blockptr(parent, i - 1); 1660 close = close_blocks(blocknr, other, blocksize); 1661 } 1662 if (!close && i < end_slot) { 1663 other = btrfs_node_blockptr(parent, i + 1); 1664 close = close_blocks(blocknr, other, blocksize); 1665 } 1666 if (close) { 1667 last_block = blocknr; 1668 continue; 1669 } 1670 1671 cur = find_extent_buffer(root->fs_info, blocknr); 1672 if (cur) 1673 uptodate = btrfs_buffer_uptodate(cur, gen, 0); 1674 else 1675 uptodate = 0; 1676 if (!cur || !uptodate) { 1677 if (!cur) { 1678 cur = read_tree_block(root, blocknr, gen); 1679 if (IS_ERR(cur)) { 1680 return PTR_ERR(cur); 1681 } else if (!extent_buffer_uptodate(cur)) { 1682 free_extent_buffer(cur); 1683 return -EIO; 1684 } 1685 } else if (!uptodate) { 1686 err = btrfs_read_buffer(cur, gen); 1687 if (err) { 1688 free_extent_buffer(cur); 1689 return err; 1690 } 1691 } 1692 } 1693 if (search_start == 0) 1694 search_start = last_block; 1695 1696 btrfs_tree_lock(cur); 1697 btrfs_set_lock_blocking(cur); 1698 err = __btrfs_cow_block(trans, root, cur, parent, i, 1699 &cur, search_start, 1700 min(16 * blocksize, 1701 (end_slot - i) * blocksize)); 1702 if (err) { 1703 btrfs_tree_unlock(cur); 1704 free_extent_buffer(cur); 1705 break; 1706 } 1707 search_start = cur->start; 1708 last_block = cur->start; 1709 *last_ret = search_start; 1710 btrfs_tree_unlock(cur); 1711 free_extent_buffer(cur); 1712 } 1713 return err; 1714 } 1715 1716 1717 /* 1718 * search for key in the extent_buffer. The items start at offset p, 1719 * and they are item_size apart. There are 'max' items in p. 1720 * 1721 * the slot in the array is returned via slot, and it points to 1722 * the place where you would insert key if it is not found in 1723 * the array. 1724 * 1725 * slot may point to max if the key is bigger than all of the keys 1726 */ 1727 static noinline int generic_bin_search(struct extent_buffer *eb, 1728 unsigned long p, 1729 int item_size, struct btrfs_key *key, 1730 int max, int *slot) 1731 { 1732 int low = 0; 1733 int high = max; 1734 int mid; 1735 int ret; 1736 struct btrfs_disk_key *tmp = NULL; 1737 struct btrfs_disk_key unaligned; 1738 unsigned long offset; 1739 char *kaddr = NULL; 1740 unsigned long map_start = 0; 1741 unsigned long map_len = 0; 1742 int err; 1743 1744 if (low > high) { 1745 btrfs_err(eb->fs_info, 1746 "%s: low (%d) > high (%d) eb %llu owner %llu level %d", 1747 __func__, low, high, eb->start, 1748 btrfs_header_owner(eb), btrfs_header_level(eb)); 1749 return -EINVAL; 1750 } 1751 1752 while (low < high) { 1753 mid = (low + high) / 2; 1754 offset = p + mid * item_size; 1755 1756 if (!kaddr || offset < map_start || 1757 (offset + sizeof(struct btrfs_disk_key)) > 1758 map_start + map_len) { 1759 1760 err = map_private_extent_buffer(eb, offset, 1761 sizeof(struct btrfs_disk_key), 1762 &kaddr, &map_start, &map_len); 1763 1764 if (!err) { 1765 tmp = (struct btrfs_disk_key *)(kaddr + offset - 1766 map_start); 1767 } else if (err == 1) { 1768 read_extent_buffer(eb, &unaligned, 1769 offset, sizeof(unaligned)); 1770 tmp = &unaligned; 1771 } else { 1772 return err; 1773 } 1774 1775 } else { 1776 tmp = (struct btrfs_disk_key *)(kaddr + offset - 1777 map_start); 1778 } 1779 ret = comp_keys(tmp, key); 1780 1781 if (ret < 0) 1782 low = mid + 1; 1783 else if (ret > 0) 1784 high = mid; 1785 else { 1786 *slot = mid; 1787 return 0; 1788 } 1789 } 1790 *slot = low; 1791 return 1; 1792 } 1793 1794 /* 1795 * simple bin_search frontend that does the right thing for 1796 * leaves vs nodes 1797 */ 1798 static int bin_search(struct extent_buffer *eb, struct btrfs_key *key, 1799 int level, int *slot) 1800 { 1801 if (level == 0) 1802 return generic_bin_search(eb, 1803 offsetof(struct btrfs_leaf, items), 1804 sizeof(struct btrfs_item), 1805 key, btrfs_header_nritems(eb), 1806 slot); 1807 else 1808 return generic_bin_search(eb, 1809 offsetof(struct btrfs_node, ptrs), 1810 sizeof(struct btrfs_key_ptr), 1811 key, btrfs_header_nritems(eb), 1812 slot); 1813 } 1814 1815 int btrfs_bin_search(struct extent_buffer *eb, struct btrfs_key *key, 1816 int level, int *slot) 1817 { 1818 return bin_search(eb, key, level, slot); 1819 } 1820 1821 static void root_add_used(struct btrfs_root *root, u32 size) 1822 { 1823 spin_lock(&root->accounting_lock); 1824 btrfs_set_root_used(&root->root_item, 1825 btrfs_root_used(&root->root_item) + size); 1826 spin_unlock(&root->accounting_lock); 1827 } 1828 1829 static void root_sub_used(struct btrfs_root *root, u32 size) 1830 { 1831 spin_lock(&root->accounting_lock); 1832 btrfs_set_root_used(&root->root_item, 1833 btrfs_root_used(&root->root_item) - size); 1834 spin_unlock(&root->accounting_lock); 1835 } 1836 1837 /* given a node and slot number, this reads the blocks it points to. The 1838 * extent buffer is returned with a reference taken (but unlocked). 1839 */ 1840 static noinline struct extent_buffer *read_node_slot(struct btrfs_root *root, 1841 struct extent_buffer *parent, int slot) 1842 { 1843 int level = btrfs_header_level(parent); 1844 struct extent_buffer *eb; 1845 1846 if (slot < 0 || slot >= btrfs_header_nritems(parent)) 1847 return ERR_PTR(-ENOENT); 1848 1849 BUG_ON(level == 0); 1850 1851 eb = read_tree_block(root, btrfs_node_blockptr(parent, slot), 1852 btrfs_node_ptr_generation(parent, slot)); 1853 if (!IS_ERR(eb) && !extent_buffer_uptodate(eb)) { 1854 free_extent_buffer(eb); 1855 eb = ERR_PTR(-EIO); 1856 } 1857 1858 return eb; 1859 } 1860 1861 /* 1862 * node level balancing, used to make sure nodes are in proper order for 1863 * item deletion. We balance from the top down, so we have to make sure 1864 * that a deletion won't leave an node completely empty later on. 1865 */ 1866 static noinline int balance_level(struct btrfs_trans_handle *trans, 1867 struct btrfs_root *root, 1868 struct btrfs_path *path, int level) 1869 { 1870 struct extent_buffer *right = NULL; 1871 struct extent_buffer *mid; 1872 struct extent_buffer *left = NULL; 1873 struct extent_buffer *parent = NULL; 1874 int ret = 0; 1875 int wret; 1876 int pslot; 1877 int orig_slot = path->slots[level]; 1878 u64 orig_ptr; 1879 1880 if (level == 0) 1881 return 0; 1882 1883 mid = path->nodes[level]; 1884 1885 WARN_ON(path->locks[level] != BTRFS_WRITE_LOCK && 1886 path->locks[level] != BTRFS_WRITE_LOCK_BLOCKING); 1887 WARN_ON(btrfs_header_generation(mid) != trans->transid); 1888 1889 orig_ptr = btrfs_node_blockptr(mid, orig_slot); 1890 1891 if (level < BTRFS_MAX_LEVEL - 1) { 1892 parent = path->nodes[level + 1]; 1893 pslot = path->slots[level + 1]; 1894 } 1895 1896 /* 1897 * deal with the case where there is only one pointer in the root 1898 * by promoting the node below to a root 1899 */ 1900 if (!parent) { 1901 struct extent_buffer *child; 1902 1903 if (btrfs_header_nritems(mid) != 1) 1904 return 0; 1905 1906 /* promote the child to a root */ 1907 child = read_node_slot(root, mid, 0); 1908 if (IS_ERR(child)) { 1909 ret = PTR_ERR(child); 1910 btrfs_handle_fs_error(root->fs_info, ret, NULL); 1911 goto enospc; 1912 } 1913 1914 btrfs_tree_lock(child); 1915 btrfs_set_lock_blocking(child); 1916 ret = btrfs_cow_block(trans, root, child, mid, 0, &child); 1917 if (ret) { 1918 btrfs_tree_unlock(child); 1919 free_extent_buffer(child); 1920 goto enospc; 1921 } 1922 1923 tree_mod_log_set_root_pointer(root, child, 1); 1924 rcu_assign_pointer(root->node, child); 1925 1926 add_root_to_dirty_list(root); 1927 btrfs_tree_unlock(child); 1928 1929 path->locks[level] = 0; 1930 path->nodes[level] = NULL; 1931 clean_tree_block(trans, root->fs_info, mid); 1932 btrfs_tree_unlock(mid); 1933 /* once for the path */ 1934 free_extent_buffer(mid); 1935 1936 root_sub_used(root, mid->len); 1937 btrfs_free_tree_block(trans, root, mid, 0, 1); 1938 /* once for the root ptr */ 1939 free_extent_buffer_stale(mid); 1940 return 0; 1941 } 1942 if (btrfs_header_nritems(mid) > 1943 BTRFS_NODEPTRS_PER_BLOCK(root) / 4) 1944 return 0; 1945 1946 left = read_node_slot(root, parent, pslot - 1); 1947 if (IS_ERR(left)) 1948 left = NULL; 1949 1950 if (left) { 1951 btrfs_tree_lock(left); 1952 btrfs_set_lock_blocking(left); 1953 wret = btrfs_cow_block(trans, root, left, 1954 parent, pslot - 1, &left); 1955 if (wret) { 1956 ret = wret; 1957 goto enospc; 1958 } 1959 } 1960 1961 right = read_node_slot(root, parent, pslot + 1); 1962 if (IS_ERR(right)) 1963 right = NULL; 1964 1965 if (right) { 1966 btrfs_tree_lock(right); 1967 btrfs_set_lock_blocking(right); 1968 wret = btrfs_cow_block(trans, root, right, 1969 parent, pslot + 1, &right); 1970 if (wret) { 1971 ret = wret; 1972 goto enospc; 1973 } 1974 } 1975 1976 /* first, try to make some room in the middle buffer */ 1977 if (left) { 1978 orig_slot += btrfs_header_nritems(left); 1979 wret = push_node_left(trans, root, left, mid, 1); 1980 if (wret < 0) 1981 ret = wret; 1982 } 1983 1984 /* 1985 * then try to empty the right most buffer into the middle 1986 */ 1987 if (right) { 1988 wret = push_node_left(trans, root, mid, right, 1); 1989 if (wret < 0 && wret != -ENOSPC) 1990 ret = wret; 1991 if (btrfs_header_nritems(right) == 0) { 1992 clean_tree_block(trans, root->fs_info, right); 1993 btrfs_tree_unlock(right); 1994 del_ptr(root, path, level + 1, pslot + 1); 1995 root_sub_used(root, right->len); 1996 btrfs_free_tree_block(trans, root, right, 0, 1); 1997 free_extent_buffer_stale(right); 1998 right = NULL; 1999 } else { 2000 struct btrfs_disk_key right_key; 2001 btrfs_node_key(right, &right_key, 0); 2002 tree_mod_log_set_node_key(root->fs_info, parent, 2003 pslot + 1, 0); 2004 btrfs_set_node_key(parent, &right_key, pslot + 1); 2005 btrfs_mark_buffer_dirty(parent); 2006 } 2007 } 2008 if (btrfs_header_nritems(mid) == 1) { 2009 /* 2010 * we're not allowed to leave a node with one item in the 2011 * tree during a delete. A deletion from lower in the tree 2012 * could try to delete the only pointer in this node. 2013 * So, pull some keys from the left. 2014 * There has to be a left pointer at this point because 2015 * otherwise we would have pulled some pointers from the 2016 * right 2017 */ 2018 if (!left) { 2019 ret = -EROFS; 2020 btrfs_handle_fs_error(root->fs_info, ret, NULL); 2021 goto enospc; 2022 } 2023 wret = balance_node_right(trans, root, mid, left); 2024 if (wret < 0) { 2025 ret = wret; 2026 goto enospc; 2027 } 2028 if (wret == 1) { 2029 wret = push_node_left(trans, root, left, mid, 1); 2030 if (wret < 0) 2031 ret = wret; 2032 } 2033 BUG_ON(wret == 1); 2034 } 2035 if (btrfs_header_nritems(mid) == 0) { 2036 clean_tree_block(trans, root->fs_info, mid); 2037 btrfs_tree_unlock(mid); 2038 del_ptr(root, path, level + 1, pslot); 2039 root_sub_used(root, mid->len); 2040 btrfs_free_tree_block(trans, root, mid, 0, 1); 2041 free_extent_buffer_stale(mid); 2042 mid = NULL; 2043 } else { 2044 /* update the parent key to reflect our changes */ 2045 struct btrfs_disk_key mid_key; 2046 btrfs_node_key(mid, &mid_key, 0); 2047 tree_mod_log_set_node_key(root->fs_info, parent, 2048 pslot, 0); 2049 btrfs_set_node_key(parent, &mid_key, pslot); 2050 btrfs_mark_buffer_dirty(parent); 2051 } 2052 2053 /* update the path */ 2054 if (left) { 2055 if (btrfs_header_nritems(left) > orig_slot) { 2056 extent_buffer_get(left); 2057 /* left was locked after cow */ 2058 path->nodes[level] = left; 2059 path->slots[level + 1] -= 1; 2060 path->slots[level] = orig_slot; 2061 if (mid) { 2062 btrfs_tree_unlock(mid); 2063 free_extent_buffer(mid); 2064 } 2065 } else { 2066 orig_slot -= btrfs_header_nritems(left); 2067 path->slots[level] = orig_slot; 2068 } 2069 } 2070 /* double check we haven't messed things up */ 2071 if (orig_ptr != 2072 btrfs_node_blockptr(path->nodes[level], path->slots[level])) 2073 BUG(); 2074 enospc: 2075 if (right) { 2076 btrfs_tree_unlock(right); 2077 free_extent_buffer(right); 2078 } 2079 if (left) { 2080 if (path->nodes[level] != left) 2081 btrfs_tree_unlock(left); 2082 free_extent_buffer(left); 2083 } 2084 return ret; 2085 } 2086 2087 /* Node balancing for insertion. Here we only split or push nodes around 2088 * when they are completely full. This is also done top down, so we 2089 * have to be pessimistic. 2090 */ 2091 static noinline int push_nodes_for_insert(struct btrfs_trans_handle *trans, 2092 struct btrfs_root *root, 2093 struct btrfs_path *path, int level) 2094 { 2095 struct extent_buffer *right = NULL; 2096 struct extent_buffer *mid; 2097 struct extent_buffer *left = NULL; 2098 struct extent_buffer *parent = NULL; 2099 int ret = 0; 2100 int wret; 2101 int pslot; 2102 int orig_slot = path->slots[level]; 2103 2104 if (level == 0) 2105 return 1; 2106 2107 mid = path->nodes[level]; 2108 WARN_ON(btrfs_header_generation(mid) != trans->transid); 2109 2110 if (level < BTRFS_MAX_LEVEL - 1) { 2111 parent = path->nodes[level + 1]; 2112 pslot = path->slots[level + 1]; 2113 } 2114 2115 if (!parent) 2116 return 1; 2117 2118 left = read_node_slot(root, parent, pslot - 1); 2119 if (IS_ERR(left)) 2120 left = NULL; 2121 2122 /* first, try to make some room in the middle buffer */ 2123 if (left) { 2124 u32 left_nr; 2125 2126 btrfs_tree_lock(left); 2127 btrfs_set_lock_blocking(left); 2128 2129 left_nr = btrfs_header_nritems(left); 2130 if (left_nr >= BTRFS_NODEPTRS_PER_BLOCK(root) - 1) { 2131 wret = 1; 2132 } else { 2133 ret = btrfs_cow_block(trans, root, left, parent, 2134 pslot - 1, &left); 2135 if (ret) 2136 wret = 1; 2137 else { 2138 wret = push_node_left(trans, root, 2139 left, mid, 0); 2140 } 2141 } 2142 if (wret < 0) 2143 ret = wret; 2144 if (wret == 0) { 2145 struct btrfs_disk_key disk_key; 2146 orig_slot += left_nr; 2147 btrfs_node_key(mid, &disk_key, 0); 2148 tree_mod_log_set_node_key(root->fs_info, parent, 2149 pslot, 0); 2150 btrfs_set_node_key(parent, &disk_key, pslot); 2151 btrfs_mark_buffer_dirty(parent); 2152 if (btrfs_header_nritems(left) > orig_slot) { 2153 path->nodes[level] = left; 2154 path->slots[level + 1] -= 1; 2155 path->slots[level] = orig_slot; 2156 btrfs_tree_unlock(mid); 2157 free_extent_buffer(mid); 2158 } else { 2159 orig_slot -= 2160 btrfs_header_nritems(left); 2161 path->slots[level] = orig_slot; 2162 btrfs_tree_unlock(left); 2163 free_extent_buffer(left); 2164 } 2165 return 0; 2166 } 2167 btrfs_tree_unlock(left); 2168 free_extent_buffer(left); 2169 } 2170 right = read_node_slot(root, parent, pslot + 1); 2171 if (IS_ERR(right)) 2172 right = NULL; 2173 2174 /* 2175 * then try to empty the right most buffer into the middle 2176 */ 2177 if (right) { 2178 u32 right_nr; 2179 2180 btrfs_tree_lock(right); 2181 btrfs_set_lock_blocking(right); 2182 2183 right_nr = btrfs_header_nritems(right); 2184 if (right_nr >= BTRFS_NODEPTRS_PER_BLOCK(root) - 1) { 2185 wret = 1; 2186 } else { 2187 ret = btrfs_cow_block(trans, root, right, 2188 parent, pslot + 1, 2189 &right); 2190 if (ret) 2191 wret = 1; 2192 else { 2193 wret = balance_node_right(trans, root, 2194 right, mid); 2195 } 2196 } 2197 if (wret < 0) 2198 ret = wret; 2199 if (wret == 0) { 2200 struct btrfs_disk_key disk_key; 2201 2202 btrfs_node_key(right, &disk_key, 0); 2203 tree_mod_log_set_node_key(root->fs_info, parent, 2204 pslot + 1, 0); 2205 btrfs_set_node_key(parent, &disk_key, pslot + 1); 2206 btrfs_mark_buffer_dirty(parent); 2207 2208 if (btrfs_header_nritems(mid) <= orig_slot) { 2209 path->nodes[level] = right; 2210 path->slots[level + 1] += 1; 2211 path->slots[level] = orig_slot - 2212 btrfs_header_nritems(mid); 2213 btrfs_tree_unlock(mid); 2214 free_extent_buffer(mid); 2215 } else { 2216 btrfs_tree_unlock(right); 2217 free_extent_buffer(right); 2218 } 2219 return 0; 2220 } 2221 btrfs_tree_unlock(right); 2222 free_extent_buffer(right); 2223 } 2224 return 1; 2225 } 2226 2227 /* 2228 * readahead one full node of leaves, finding things that are close 2229 * to the block in 'slot', and triggering ra on them. 2230 */ 2231 static void reada_for_search(struct btrfs_root *root, 2232 struct btrfs_path *path, 2233 int level, int slot, u64 objectid) 2234 { 2235 struct extent_buffer *node; 2236 struct btrfs_disk_key disk_key; 2237 u32 nritems; 2238 u64 search; 2239 u64 target; 2240 u64 nread = 0; 2241 struct extent_buffer *eb; 2242 u32 nr; 2243 u32 blocksize; 2244 u32 nscan = 0; 2245 2246 if (level != 1) 2247 return; 2248 2249 if (!path->nodes[level]) 2250 return; 2251 2252 node = path->nodes[level]; 2253 2254 search = btrfs_node_blockptr(node, slot); 2255 blocksize = root->nodesize; 2256 eb = find_extent_buffer(root->fs_info, search); 2257 if (eb) { 2258 free_extent_buffer(eb); 2259 return; 2260 } 2261 2262 target = search; 2263 2264 nritems = btrfs_header_nritems(node); 2265 nr = slot; 2266 2267 while (1) { 2268 if (path->reada == READA_BACK) { 2269 if (nr == 0) 2270 break; 2271 nr--; 2272 } else if (path->reada == READA_FORWARD) { 2273 nr++; 2274 if (nr >= nritems) 2275 break; 2276 } 2277 if (path->reada == READA_BACK && objectid) { 2278 btrfs_node_key(node, &disk_key, nr); 2279 if (btrfs_disk_key_objectid(&disk_key) != objectid) 2280 break; 2281 } 2282 search = btrfs_node_blockptr(node, nr); 2283 if ((search <= target && target - search <= 65536) || 2284 (search > target && search - target <= 65536)) { 2285 readahead_tree_block(root, search); 2286 nread += blocksize; 2287 } 2288 nscan++; 2289 if ((nread > 65536 || nscan > 32)) 2290 break; 2291 } 2292 } 2293 2294 static noinline void reada_for_balance(struct btrfs_root *root, 2295 struct btrfs_path *path, int level) 2296 { 2297 int slot; 2298 int nritems; 2299 struct extent_buffer *parent; 2300 struct extent_buffer *eb; 2301 u64 gen; 2302 u64 block1 = 0; 2303 u64 block2 = 0; 2304 2305 parent = path->nodes[level + 1]; 2306 if (!parent) 2307 return; 2308 2309 nritems = btrfs_header_nritems(parent); 2310 slot = path->slots[level + 1]; 2311 2312 if (slot > 0) { 2313 block1 = btrfs_node_blockptr(parent, slot - 1); 2314 gen = btrfs_node_ptr_generation(parent, slot - 1); 2315 eb = find_extent_buffer(root->fs_info, block1); 2316 /* 2317 * if we get -eagain from btrfs_buffer_uptodate, we 2318 * don't want to return eagain here. That will loop 2319 * forever 2320 */ 2321 if (eb && btrfs_buffer_uptodate(eb, gen, 1) != 0) 2322 block1 = 0; 2323 free_extent_buffer(eb); 2324 } 2325 if (slot + 1 < nritems) { 2326 block2 = btrfs_node_blockptr(parent, slot + 1); 2327 gen = btrfs_node_ptr_generation(parent, slot + 1); 2328 eb = find_extent_buffer(root->fs_info, block2); 2329 if (eb && btrfs_buffer_uptodate(eb, gen, 1) != 0) 2330 block2 = 0; 2331 free_extent_buffer(eb); 2332 } 2333 2334 if (block1) 2335 readahead_tree_block(root, block1); 2336 if (block2) 2337 readahead_tree_block(root, block2); 2338 } 2339 2340 2341 /* 2342 * when we walk down the tree, it is usually safe to unlock the higher layers 2343 * in the tree. The exceptions are when our path goes through slot 0, because 2344 * operations on the tree might require changing key pointers higher up in the 2345 * tree. 2346 * 2347 * callers might also have set path->keep_locks, which tells this code to keep 2348 * the lock if the path points to the last slot in the block. This is part of 2349 * walking through the tree, and selecting the next slot in the higher block. 2350 * 2351 * lowest_unlock sets the lowest level in the tree we're allowed to unlock. so 2352 * if lowest_unlock is 1, level 0 won't be unlocked 2353 */ 2354 static noinline void unlock_up(struct btrfs_path *path, int level, 2355 int lowest_unlock, int min_write_lock_level, 2356 int *write_lock_level) 2357 { 2358 int i; 2359 int skip_level = level; 2360 int no_skips = 0; 2361 struct extent_buffer *t; 2362 2363 for (i = level; i < BTRFS_MAX_LEVEL; i++) { 2364 if (!path->nodes[i]) 2365 break; 2366 if (!path->locks[i]) 2367 break; 2368 if (!no_skips && path->slots[i] == 0) { 2369 skip_level = i + 1; 2370 continue; 2371 } 2372 if (!no_skips && path->keep_locks) { 2373 u32 nritems; 2374 t = path->nodes[i]; 2375 nritems = btrfs_header_nritems(t); 2376 if (nritems < 1 || path->slots[i] >= nritems - 1) { 2377 skip_level = i + 1; 2378 continue; 2379 } 2380 } 2381 if (skip_level < i && i >= lowest_unlock) 2382 no_skips = 1; 2383 2384 t = path->nodes[i]; 2385 if (i >= lowest_unlock && i > skip_level && path->locks[i]) { 2386 btrfs_tree_unlock_rw(t, path->locks[i]); 2387 path->locks[i] = 0; 2388 if (write_lock_level && 2389 i > min_write_lock_level && 2390 i <= *write_lock_level) { 2391 *write_lock_level = i - 1; 2392 } 2393 } 2394 } 2395 } 2396 2397 /* 2398 * This releases any locks held in the path starting at level and 2399 * going all the way up to the root. 2400 * 2401 * btrfs_search_slot will keep the lock held on higher nodes in a few 2402 * corner cases, such as COW of the block at slot zero in the node. This 2403 * ignores those rules, and it should only be called when there are no 2404 * more updates to be done higher up in the tree. 2405 */ 2406 noinline void btrfs_unlock_up_safe(struct btrfs_path *path, int level) 2407 { 2408 int i; 2409 2410 if (path->keep_locks) 2411 return; 2412 2413 for (i = level; i < BTRFS_MAX_LEVEL; i++) { 2414 if (!path->nodes[i]) 2415 continue; 2416 if (!path->locks[i]) 2417 continue; 2418 btrfs_tree_unlock_rw(path->nodes[i], path->locks[i]); 2419 path->locks[i] = 0; 2420 } 2421 } 2422 2423 /* 2424 * helper function for btrfs_search_slot. The goal is to find a block 2425 * in cache without setting the path to blocking. If we find the block 2426 * we return zero and the path is unchanged. 2427 * 2428 * If we can't find the block, we set the path blocking and do some 2429 * reada. -EAGAIN is returned and the search must be repeated. 2430 */ 2431 static int 2432 read_block_for_search(struct btrfs_trans_handle *trans, 2433 struct btrfs_root *root, struct btrfs_path *p, 2434 struct extent_buffer **eb_ret, int level, int slot, 2435 struct btrfs_key *key, u64 time_seq) 2436 { 2437 u64 blocknr; 2438 u64 gen; 2439 struct extent_buffer *b = *eb_ret; 2440 struct extent_buffer *tmp; 2441 int ret; 2442 2443 blocknr = btrfs_node_blockptr(b, slot); 2444 gen = btrfs_node_ptr_generation(b, slot); 2445 2446 tmp = find_extent_buffer(root->fs_info, blocknr); 2447 if (tmp) { 2448 /* first we do an atomic uptodate check */ 2449 if (btrfs_buffer_uptodate(tmp, gen, 1) > 0) { 2450 *eb_ret = tmp; 2451 return 0; 2452 } 2453 2454 /* the pages were up to date, but we failed 2455 * the generation number check. Do a full 2456 * read for the generation number that is correct. 2457 * We must do this without dropping locks so 2458 * we can trust our generation number 2459 */ 2460 btrfs_set_path_blocking(p); 2461 2462 /* now we're allowed to do a blocking uptodate check */ 2463 ret = btrfs_read_buffer(tmp, gen); 2464 if (!ret) { 2465 *eb_ret = tmp; 2466 return 0; 2467 } 2468 free_extent_buffer(tmp); 2469 btrfs_release_path(p); 2470 return -EIO; 2471 } 2472 2473 /* 2474 * reduce lock contention at high levels 2475 * of the btree by dropping locks before 2476 * we read. Don't release the lock on the current 2477 * level because we need to walk this node to figure 2478 * out which blocks to read. 2479 */ 2480 btrfs_unlock_up_safe(p, level + 1); 2481 btrfs_set_path_blocking(p); 2482 2483 free_extent_buffer(tmp); 2484 if (p->reada != READA_NONE) 2485 reada_for_search(root, p, level, slot, key->objectid); 2486 2487 btrfs_release_path(p); 2488 2489 ret = -EAGAIN; 2490 tmp = read_tree_block(root, blocknr, 0); 2491 if (!IS_ERR(tmp)) { 2492 /* 2493 * If the read above didn't mark this buffer up to date, 2494 * it will never end up being up to date. Set ret to EIO now 2495 * and give up so that our caller doesn't loop forever 2496 * on our EAGAINs. 2497 */ 2498 if (!btrfs_buffer_uptodate(tmp, 0, 0)) 2499 ret = -EIO; 2500 free_extent_buffer(tmp); 2501 } else { 2502 ret = PTR_ERR(tmp); 2503 } 2504 return ret; 2505 } 2506 2507 /* 2508 * helper function for btrfs_search_slot. This does all of the checks 2509 * for node-level blocks and does any balancing required based on 2510 * the ins_len. 2511 * 2512 * If no extra work was required, zero is returned. If we had to 2513 * drop the path, -EAGAIN is returned and btrfs_search_slot must 2514 * start over 2515 */ 2516 static int 2517 setup_nodes_for_search(struct btrfs_trans_handle *trans, 2518 struct btrfs_root *root, struct btrfs_path *p, 2519 struct extent_buffer *b, int level, int ins_len, 2520 int *write_lock_level) 2521 { 2522 int ret; 2523 if ((p->search_for_split || ins_len > 0) && btrfs_header_nritems(b) >= 2524 BTRFS_NODEPTRS_PER_BLOCK(root) - 3) { 2525 int sret; 2526 2527 if (*write_lock_level < level + 1) { 2528 *write_lock_level = level + 1; 2529 btrfs_release_path(p); 2530 goto again; 2531 } 2532 2533 btrfs_set_path_blocking(p); 2534 reada_for_balance(root, p, level); 2535 sret = split_node(trans, root, p, level); 2536 btrfs_clear_path_blocking(p, NULL, 0); 2537 2538 BUG_ON(sret > 0); 2539 if (sret) { 2540 ret = sret; 2541 goto done; 2542 } 2543 b = p->nodes[level]; 2544 } else if (ins_len < 0 && btrfs_header_nritems(b) < 2545 BTRFS_NODEPTRS_PER_BLOCK(root) / 2) { 2546 int sret; 2547 2548 if (*write_lock_level < level + 1) { 2549 *write_lock_level = level + 1; 2550 btrfs_release_path(p); 2551 goto again; 2552 } 2553 2554 btrfs_set_path_blocking(p); 2555 reada_for_balance(root, p, level); 2556 sret = balance_level(trans, root, p, level); 2557 btrfs_clear_path_blocking(p, NULL, 0); 2558 2559 if (sret) { 2560 ret = sret; 2561 goto done; 2562 } 2563 b = p->nodes[level]; 2564 if (!b) { 2565 btrfs_release_path(p); 2566 goto again; 2567 } 2568 BUG_ON(btrfs_header_nritems(b) == 1); 2569 } 2570 return 0; 2571 2572 again: 2573 ret = -EAGAIN; 2574 done: 2575 return ret; 2576 } 2577 2578 static void key_search_validate(struct extent_buffer *b, 2579 struct btrfs_key *key, 2580 int level) 2581 { 2582 #ifdef CONFIG_BTRFS_ASSERT 2583 struct btrfs_disk_key disk_key; 2584 2585 btrfs_cpu_key_to_disk(&disk_key, key); 2586 2587 if (level == 0) 2588 ASSERT(!memcmp_extent_buffer(b, &disk_key, 2589 offsetof(struct btrfs_leaf, items[0].key), 2590 sizeof(disk_key))); 2591 else 2592 ASSERT(!memcmp_extent_buffer(b, &disk_key, 2593 offsetof(struct btrfs_node, ptrs[0].key), 2594 sizeof(disk_key))); 2595 #endif 2596 } 2597 2598 static int key_search(struct extent_buffer *b, struct btrfs_key *key, 2599 int level, int *prev_cmp, int *slot) 2600 { 2601 if (*prev_cmp != 0) { 2602 *prev_cmp = bin_search(b, key, level, slot); 2603 return *prev_cmp; 2604 } 2605 2606 key_search_validate(b, key, level); 2607 *slot = 0; 2608 2609 return 0; 2610 } 2611 2612 int btrfs_find_item(struct btrfs_root *fs_root, struct btrfs_path *path, 2613 u64 iobjectid, u64 ioff, u8 key_type, 2614 struct btrfs_key *found_key) 2615 { 2616 int ret; 2617 struct btrfs_key key; 2618 struct extent_buffer *eb; 2619 2620 ASSERT(path); 2621 ASSERT(found_key); 2622 2623 key.type = key_type; 2624 key.objectid = iobjectid; 2625 key.offset = ioff; 2626 2627 ret = btrfs_search_slot(NULL, fs_root, &key, path, 0, 0); 2628 if (ret < 0) 2629 return ret; 2630 2631 eb = path->nodes[0]; 2632 if (ret && path->slots[0] >= btrfs_header_nritems(eb)) { 2633 ret = btrfs_next_leaf(fs_root, path); 2634 if (ret) 2635 return ret; 2636 eb = path->nodes[0]; 2637 } 2638 2639 btrfs_item_key_to_cpu(eb, found_key, path->slots[0]); 2640 if (found_key->type != key.type || 2641 found_key->objectid != key.objectid) 2642 return 1; 2643 2644 return 0; 2645 } 2646 2647 /* 2648 * look for key in the tree. path is filled in with nodes along the way 2649 * if key is found, we return zero and you can find the item in the leaf 2650 * level of the path (level 0) 2651 * 2652 * If the key isn't found, the path points to the slot where it should 2653 * be inserted, and 1 is returned. If there are other errors during the 2654 * search a negative error number is returned. 2655 * 2656 * if ins_len > 0, nodes and leaves will be split as we walk down the 2657 * tree. if ins_len < 0, nodes will be merged as we walk down the tree (if 2658 * possible) 2659 */ 2660 int btrfs_search_slot(struct btrfs_trans_handle *trans, struct btrfs_root 2661 *root, struct btrfs_key *key, struct btrfs_path *p, int 2662 ins_len, int cow) 2663 { 2664 struct extent_buffer *b; 2665 int slot; 2666 int ret; 2667 int err; 2668 int level; 2669 int lowest_unlock = 1; 2670 int root_lock; 2671 /* everything at write_lock_level or lower must be write locked */ 2672 int write_lock_level = 0; 2673 u8 lowest_level = 0; 2674 int min_write_lock_level; 2675 int prev_cmp; 2676 2677 lowest_level = p->lowest_level; 2678 WARN_ON(lowest_level && ins_len > 0); 2679 WARN_ON(p->nodes[0] != NULL); 2680 BUG_ON(!cow && ins_len); 2681 2682 if (ins_len < 0) { 2683 lowest_unlock = 2; 2684 2685 /* when we are removing items, we might have to go up to level 2686 * two as we update tree pointers Make sure we keep write 2687 * for those levels as well 2688 */ 2689 write_lock_level = 2; 2690 } else if (ins_len > 0) { 2691 /* 2692 * for inserting items, make sure we have a write lock on 2693 * level 1 so we can update keys 2694 */ 2695 write_lock_level = 1; 2696 } 2697 2698 if (!cow) 2699 write_lock_level = -1; 2700 2701 if (cow && (p->keep_locks || p->lowest_level)) 2702 write_lock_level = BTRFS_MAX_LEVEL; 2703 2704 min_write_lock_level = write_lock_level; 2705 2706 again: 2707 prev_cmp = -1; 2708 /* 2709 * we try very hard to do read locks on the root 2710 */ 2711 root_lock = BTRFS_READ_LOCK; 2712 level = 0; 2713 if (p->search_commit_root) { 2714 /* 2715 * the commit roots are read only 2716 * so we always do read locks 2717 */ 2718 if (p->need_commit_sem) 2719 down_read(&root->fs_info->commit_root_sem); 2720 b = root->commit_root; 2721 extent_buffer_get(b); 2722 level = btrfs_header_level(b); 2723 if (p->need_commit_sem) 2724 up_read(&root->fs_info->commit_root_sem); 2725 if (!p->skip_locking) 2726 btrfs_tree_read_lock(b); 2727 } else { 2728 if (p->skip_locking) { 2729 b = btrfs_root_node(root); 2730 level = btrfs_header_level(b); 2731 } else { 2732 /* we don't know the level of the root node 2733 * until we actually have it read locked 2734 */ 2735 b = btrfs_read_lock_root_node(root); 2736 level = btrfs_header_level(b); 2737 if (level <= write_lock_level) { 2738 /* whoops, must trade for write lock */ 2739 btrfs_tree_read_unlock(b); 2740 free_extent_buffer(b); 2741 b = btrfs_lock_root_node(root); 2742 root_lock = BTRFS_WRITE_LOCK; 2743 2744 /* the level might have changed, check again */ 2745 level = btrfs_header_level(b); 2746 } 2747 } 2748 } 2749 p->nodes[level] = b; 2750 if (!p->skip_locking) 2751 p->locks[level] = root_lock; 2752 2753 while (b) { 2754 level = btrfs_header_level(b); 2755 2756 /* 2757 * setup the path here so we can release it under lock 2758 * contention with the cow code 2759 */ 2760 if (cow) { 2761 /* 2762 * if we don't really need to cow this block 2763 * then we don't want to set the path blocking, 2764 * so we test it here 2765 */ 2766 if (!should_cow_block(trans, root, b)) { 2767 trans->dirty = true; 2768 goto cow_done; 2769 } 2770 2771 /* 2772 * must have write locks on this node and the 2773 * parent 2774 */ 2775 if (level > write_lock_level || 2776 (level + 1 > write_lock_level && 2777 level + 1 < BTRFS_MAX_LEVEL && 2778 p->nodes[level + 1])) { 2779 write_lock_level = level + 1; 2780 btrfs_release_path(p); 2781 goto again; 2782 } 2783 2784 btrfs_set_path_blocking(p); 2785 err = btrfs_cow_block(trans, root, b, 2786 p->nodes[level + 1], 2787 p->slots[level + 1], &b); 2788 if (err) { 2789 ret = err; 2790 goto done; 2791 } 2792 } 2793 cow_done: 2794 p->nodes[level] = b; 2795 btrfs_clear_path_blocking(p, NULL, 0); 2796 2797 /* 2798 * we have a lock on b and as long as we aren't changing 2799 * the tree, there is no way to for the items in b to change. 2800 * It is safe to drop the lock on our parent before we 2801 * go through the expensive btree search on b. 2802 * 2803 * If we're inserting or deleting (ins_len != 0), then we might 2804 * be changing slot zero, which may require changing the parent. 2805 * So, we can't drop the lock until after we know which slot 2806 * we're operating on. 2807 */ 2808 if (!ins_len && !p->keep_locks) { 2809 int u = level + 1; 2810 2811 if (u < BTRFS_MAX_LEVEL && p->locks[u]) { 2812 btrfs_tree_unlock_rw(p->nodes[u], p->locks[u]); 2813 p->locks[u] = 0; 2814 } 2815 } 2816 2817 ret = key_search(b, key, level, &prev_cmp, &slot); 2818 if (ret < 0) 2819 goto done; 2820 2821 if (level != 0) { 2822 int dec = 0; 2823 if (ret && slot > 0) { 2824 dec = 1; 2825 slot -= 1; 2826 } 2827 p->slots[level] = slot; 2828 err = setup_nodes_for_search(trans, root, p, b, level, 2829 ins_len, &write_lock_level); 2830 if (err == -EAGAIN) 2831 goto again; 2832 if (err) { 2833 ret = err; 2834 goto done; 2835 } 2836 b = p->nodes[level]; 2837 slot = p->slots[level]; 2838 2839 /* 2840 * slot 0 is special, if we change the key 2841 * we have to update the parent pointer 2842 * which means we must have a write lock 2843 * on the parent 2844 */ 2845 if (slot == 0 && ins_len && 2846 write_lock_level < level + 1) { 2847 write_lock_level = level + 1; 2848 btrfs_release_path(p); 2849 goto again; 2850 } 2851 2852 unlock_up(p, level, lowest_unlock, 2853 min_write_lock_level, &write_lock_level); 2854 2855 if (level == lowest_level) { 2856 if (dec) 2857 p->slots[level]++; 2858 goto done; 2859 } 2860 2861 err = read_block_for_search(trans, root, p, 2862 &b, level, slot, key, 0); 2863 if (err == -EAGAIN) 2864 goto again; 2865 if (err) { 2866 ret = err; 2867 goto done; 2868 } 2869 2870 if (!p->skip_locking) { 2871 level = btrfs_header_level(b); 2872 if (level <= write_lock_level) { 2873 err = btrfs_try_tree_write_lock(b); 2874 if (!err) { 2875 btrfs_set_path_blocking(p); 2876 btrfs_tree_lock(b); 2877 btrfs_clear_path_blocking(p, b, 2878 BTRFS_WRITE_LOCK); 2879 } 2880 p->locks[level] = BTRFS_WRITE_LOCK; 2881 } else { 2882 err = btrfs_tree_read_lock_atomic(b); 2883 if (!err) { 2884 btrfs_set_path_blocking(p); 2885 btrfs_tree_read_lock(b); 2886 btrfs_clear_path_blocking(p, b, 2887 BTRFS_READ_LOCK); 2888 } 2889 p->locks[level] = BTRFS_READ_LOCK; 2890 } 2891 p->nodes[level] = b; 2892 } 2893 } else { 2894 p->slots[level] = slot; 2895 if (ins_len > 0 && 2896 btrfs_leaf_free_space(root, b) < ins_len) { 2897 if (write_lock_level < 1) { 2898 write_lock_level = 1; 2899 btrfs_release_path(p); 2900 goto again; 2901 } 2902 2903 btrfs_set_path_blocking(p); 2904 err = split_leaf(trans, root, key, 2905 p, ins_len, ret == 0); 2906 btrfs_clear_path_blocking(p, NULL, 0); 2907 2908 BUG_ON(err > 0); 2909 if (err) { 2910 ret = err; 2911 goto done; 2912 } 2913 } 2914 if (!p->search_for_split) 2915 unlock_up(p, level, lowest_unlock, 2916 min_write_lock_level, &write_lock_level); 2917 goto done; 2918 } 2919 } 2920 ret = 1; 2921 done: 2922 /* 2923 * we don't really know what they plan on doing with the path 2924 * from here on, so for now just mark it as blocking 2925 */ 2926 if (!p->leave_spinning) 2927 btrfs_set_path_blocking(p); 2928 if (ret < 0 && !p->skip_release_on_error) 2929 btrfs_release_path(p); 2930 return ret; 2931 } 2932 2933 /* 2934 * Like btrfs_search_slot, this looks for a key in the given tree. It uses the 2935 * current state of the tree together with the operations recorded in the tree 2936 * modification log to search for the key in a previous version of this tree, as 2937 * denoted by the time_seq parameter. 2938 * 2939 * Naturally, there is no support for insert, delete or cow operations. 2940 * 2941 * The resulting path and return value will be set up as if we called 2942 * btrfs_search_slot at that point in time with ins_len and cow both set to 0. 2943 */ 2944 int btrfs_search_old_slot(struct btrfs_root *root, struct btrfs_key *key, 2945 struct btrfs_path *p, u64 time_seq) 2946 { 2947 struct extent_buffer *b; 2948 int slot; 2949 int ret; 2950 int err; 2951 int level; 2952 int lowest_unlock = 1; 2953 u8 lowest_level = 0; 2954 int prev_cmp = -1; 2955 2956 lowest_level = p->lowest_level; 2957 WARN_ON(p->nodes[0] != NULL); 2958 2959 if (p->search_commit_root) { 2960 BUG_ON(time_seq); 2961 return btrfs_search_slot(NULL, root, key, p, 0, 0); 2962 } 2963 2964 again: 2965 b = get_old_root(root, time_seq); 2966 level = btrfs_header_level(b); 2967 p->locks[level] = BTRFS_READ_LOCK; 2968 2969 while (b) { 2970 level = btrfs_header_level(b); 2971 p->nodes[level] = b; 2972 btrfs_clear_path_blocking(p, NULL, 0); 2973 2974 /* 2975 * we have a lock on b and as long as we aren't changing 2976 * the tree, there is no way to for the items in b to change. 2977 * It is safe to drop the lock on our parent before we 2978 * go through the expensive btree search on b. 2979 */ 2980 btrfs_unlock_up_safe(p, level + 1); 2981 2982 /* 2983 * Since we can unwind ebs we want to do a real search every 2984 * time. 2985 */ 2986 prev_cmp = -1; 2987 ret = key_search(b, key, level, &prev_cmp, &slot); 2988 2989 if (level != 0) { 2990 int dec = 0; 2991 if (ret && slot > 0) { 2992 dec = 1; 2993 slot -= 1; 2994 } 2995 p->slots[level] = slot; 2996 unlock_up(p, level, lowest_unlock, 0, NULL); 2997 2998 if (level == lowest_level) { 2999 if (dec) 3000 p->slots[level]++; 3001 goto done; 3002 } 3003 3004 err = read_block_for_search(NULL, root, p, &b, level, 3005 slot, key, time_seq); 3006 if (err == -EAGAIN) 3007 goto again; 3008 if (err) { 3009 ret = err; 3010 goto done; 3011 } 3012 3013 level = btrfs_header_level(b); 3014 err = btrfs_tree_read_lock_atomic(b); 3015 if (!err) { 3016 btrfs_set_path_blocking(p); 3017 btrfs_tree_read_lock(b); 3018 btrfs_clear_path_blocking(p, b, 3019 BTRFS_READ_LOCK); 3020 } 3021 b = tree_mod_log_rewind(root->fs_info, p, b, time_seq); 3022 if (!b) { 3023 ret = -ENOMEM; 3024 goto done; 3025 } 3026 p->locks[level] = BTRFS_READ_LOCK; 3027 p->nodes[level] = b; 3028 } else { 3029 p->slots[level] = slot; 3030 unlock_up(p, level, lowest_unlock, 0, NULL); 3031 goto done; 3032 } 3033 } 3034 ret = 1; 3035 done: 3036 if (!p->leave_spinning) 3037 btrfs_set_path_blocking(p); 3038 if (ret < 0) 3039 btrfs_release_path(p); 3040 3041 return ret; 3042 } 3043 3044 /* 3045 * helper to use instead of search slot if no exact match is needed but 3046 * instead the next or previous item should be returned. 3047 * When find_higher is true, the next higher item is returned, the next lower 3048 * otherwise. 3049 * When return_any and find_higher are both true, and no higher item is found, 3050 * return the next lower instead. 3051 * When return_any is true and find_higher is false, and no lower item is found, 3052 * return the next higher instead. 3053 * It returns 0 if any item is found, 1 if none is found (tree empty), and 3054 * < 0 on error 3055 */ 3056 int btrfs_search_slot_for_read(struct btrfs_root *root, 3057 struct btrfs_key *key, struct btrfs_path *p, 3058 int find_higher, int return_any) 3059 { 3060 int ret; 3061 struct extent_buffer *leaf; 3062 3063 again: 3064 ret = btrfs_search_slot(NULL, root, key, p, 0, 0); 3065 if (ret <= 0) 3066 return ret; 3067 /* 3068 * a return value of 1 means the path is at the position where the 3069 * item should be inserted. Normally this is the next bigger item, 3070 * but in case the previous item is the last in a leaf, path points 3071 * to the first free slot in the previous leaf, i.e. at an invalid 3072 * item. 3073 */ 3074 leaf = p->nodes[0]; 3075 3076 if (find_higher) { 3077 if (p->slots[0] >= btrfs_header_nritems(leaf)) { 3078 ret = btrfs_next_leaf(root, p); 3079 if (ret <= 0) 3080 return ret; 3081 if (!return_any) 3082 return 1; 3083 /* 3084 * no higher item found, return the next 3085 * lower instead 3086 */ 3087 return_any = 0; 3088 find_higher = 0; 3089 btrfs_release_path(p); 3090 goto again; 3091 } 3092 } else { 3093 if (p->slots[0] == 0) { 3094 ret = btrfs_prev_leaf(root, p); 3095 if (ret < 0) 3096 return ret; 3097 if (!ret) { 3098 leaf = p->nodes[0]; 3099 if (p->slots[0] == btrfs_header_nritems(leaf)) 3100 p->slots[0]--; 3101 return 0; 3102 } 3103 if (!return_any) 3104 return 1; 3105 /* 3106 * no lower item found, return the next 3107 * higher instead 3108 */ 3109 return_any = 0; 3110 find_higher = 1; 3111 btrfs_release_path(p); 3112 goto again; 3113 } else { 3114 --p->slots[0]; 3115 } 3116 } 3117 return 0; 3118 } 3119 3120 /* 3121 * adjust the pointers going up the tree, starting at level 3122 * making sure the right key of each node is points to 'key'. 3123 * This is used after shifting pointers to the left, so it stops 3124 * fixing up pointers when a given leaf/node is not in slot 0 of the 3125 * higher levels 3126 * 3127 */ 3128 static void fixup_low_keys(struct btrfs_fs_info *fs_info, 3129 struct btrfs_path *path, 3130 struct btrfs_disk_key *key, int level) 3131 { 3132 int i; 3133 struct extent_buffer *t; 3134 3135 for (i = level; i < BTRFS_MAX_LEVEL; i++) { 3136 int tslot = path->slots[i]; 3137 if (!path->nodes[i]) 3138 break; 3139 t = path->nodes[i]; 3140 tree_mod_log_set_node_key(fs_info, t, tslot, 1); 3141 btrfs_set_node_key(t, key, tslot); 3142 btrfs_mark_buffer_dirty(path->nodes[i]); 3143 if (tslot != 0) 3144 break; 3145 } 3146 } 3147 3148 /* 3149 * update item key. 3150 * 3151 * This function isn't completely safe. It's the caller's responsibility 3152 * that the new key won't break the order 3153 */ 3154 void btrfs_set_item_key_safe(struct btrfs_fs_info *fs_info, 3155 struct btrfs_path *path, 3156 struct btrfs_key *new_key) 3157 { 3158 struct btrfs_disk_key disk_key; 3159 struct extent_buffer *eb; 3160 int slot; 3161 3162 eb = path->nodes[0]; 3163 slot = path->slots[0]; 3164 if (slot > 0) { 3165 btrfs_item_key(eb, &disk_key, slot - 1); 3166 BUG_ON(comp_keys(&disk_key, new_key) >= 0); 3167 } 3168 if (slot < btrfs_header_nritems(eb) - 1) { 3169 btrfs_item_key(eb, &disk_key, slot + 1); 3170 BUG_ON(comp_keys(&disk_key, new_key) <= 0); 3171 } 3172 3173 btrfs_cpu_key_to_disk(&disk_key, new_key); 3174 btrfs_set_item_key(eb, &disk_key, slot); 3175 btrfs_mark_buffer_dirty(eb); 3176 if (slot == 0) 3177 fixup_low_keys(fs_info, path, &disk_key, 1); 3178 } 3179 3180 /* 3181 * try to push data from one node into the next node left in the 3182 * tree. 3183 * 3184 * returns 0 if some ptrs were pushed left, < 0 if there was some horrible 3185 * error, and > 0 if there was no room in the left hand block. 3186 */ 3187 static int push_node_left(struct btrfs_trans_handle *trans, 3188 struct btrfs_root *root, struct extent_buffer *dst, 3189 struct extent_buffer *src, int empty) 3190 { 3191 int push_items = 0; 3192 int src_nritems; 3193 int dst_nritems; 3194 int ret = 0; 3195 3196 src_nritems = btrfs_header_nritems(src); 3197 dst_nritems = btrfs_header_nritems(dst); 3198 push_items = BTRFS_NODEPTRS_PER_BLOCK(root) - dst_nritems; 3199 WARN_ON(btrfs_header_generation(src) != trans->transid); 3200 WARN_ON(btrfs_header_generation(dst) != trans->transid); 3201 3202 if (!empty && src_nritems <= 8) 3203 return 1; 3204 3205 if (push_items <= 0) 3206 return 1; 3207 3208 if (empty) { 3209 push_items = min(src_nritems, push_items); 3210 if (push_items < src_nritems) { 3211 /* leave at least 8 pointers in the node if 3212 * we aren't going to empty it 3213 */ 3214 if (src_nritems - push_items < 8) { 3215 if (push_items <= 8) 3216 return 1; 3217 push_items -= 8; 3218 } 3219 } 3220 } else 3221 push_items = min(src_nritems - 8, push_items); 3222 3223 ret = tree_mod_log_eb_copy(root->fs_info, dst, src, dst_nritems, 0, 3224 push_items); 3225 if (ret) { 3226 btrfs_abort_transaction(trans, ret); 3227 return ret; 3228 } 3229 copy_extent_buffer(dst, src, 3230 btrfs_node_key_ptr_offset(dst_nritems), 3231 btrfs_node_key_ptr_offset(0), 3232 push_items * sizeof(struct btrfs_key_ptr)); 3233 3234 if (push_items < src_nritems) { 3235 /* 3236 * don't call tree_mod_log_eb_move here, key removal was already 3237 * fully logged by tree_mod_log_eb_copy above. 3238 */ 3239 memmove_extent_buffer(src, btrfs_node_key_ptr_offset(0), 3240 btrfs_node_key_ptr_offset(push_items), 3241 (src_nritems - push_items) * 3242 sizeof(struct btrfs_key_ptr)); 3243 } 3244 btrfs_set_header_nritems(src, src_nritems - push_items); 3245 btrfs_set_header_nritems(dst, dst_nritems + push_items); 3246 btrfs_mark_buffer_dirty(src); 3247 btrfs_mark_buffer_dirty(dst); 3248 3249 return ret; 3250 } 3251 3252 /* 3253 * try to push data from one node into the next node right in the 3254 * tree. 3255 * 3256 * returns 0 if some ptrs were pushed, < 0 if there was some horrible 3257 * error, and > 0 if there was no room in the right hand block. 3258 * 3259 * this will only push up to 1/2 the contents of the left node over 3260 */ 3261 static int balance_node_right(struct btrfs_trans_handle *trans, 3262 struct btrfs_root *root, 3263 struct extent_buffer *dst, 3264 struct extent_buffer *src) 3265 { 3266 int push_items = 0; 3267 int max_push; 3268 int src_nritems; 3269 int dst_nritems; 3270 int ret = 0; 3271 3272 WARN_ON(btrfs_header_generation(src) != trans->transid); 3273 WARN_ON(btrfs_header_generation(dst) != trans->transid); 3274 3275 src_nritems = btrfs_header_nritems(src); 3276 dst_nritems = btrfs_header_nritems(dst); 3277 push_items = BTRFS_NODEPTRS_PER_BLOCK(root) - dst_nritems; 3278 if (push_items <= 0) 3279 return 1; 3280 3281 if (src_nritems < 4) 3282 return 1; 3283 3284 max_push = src_nritems / 2 + 1; 3285 /* don't try to empty the node */ 3286 if (max_push >= src_nritems) 3287 return 1; 3288 3289 if (max_push < push_items) 3290 push_items = max_push; 3291 3292 tree_mod_log_eb_move(root->fs_info, dst, push_items, 0, dst_nritems); 3293 memmove_extent_buffer(dst, btrfs_node_key_ptr_offset(push_items), 3294 btrfs_node_key_ptr_offset(0), 3295 (dst_nritems) * 3296 sizeof(struct btrfs_key_ptr)); 3297 3298 ret = tree_mod_log_eb_copy(root->fs_info, dst, src, 0, 3299 src_nritems - push_items, push_items); 3300 if (ret) { 3301 btrfs_abort_transaction(trans, ret); 3302 return ret; 3303 } 3304 copy_extent_buffer(dst, src, 3305 btrfs_node_key_ptr_offset(0), 3306 btrfs_node_key_ptr_offset(src_nritems - push_items), 3307 push_items * sizeof(struct btrfs_key_ptr)); 3308 3309 btrfs_set_header_nritems(src, src_nritems - push_items); 3310 btrfs_set_header_nritems(dst, dst_nritems + push_items); 3311 3312 btrfs_mark_buffer_dirty(src); 3313 btrfs_mark_buffer_dirty(dst); 3314 3315 return ret; 3316 } 3317 3318 /* 3319 * helper function to insert a new root level in the tree. 3320 * A new node is allocated, and a single item is inserted to 3321 * point to the existing root 3322 * 3323 * returns zero on success or < 0 on failure. 3324 */ 3325 static noinline int insert_new_root(struct btrfs_trans_handle *trans, 3326 struct btrfs_root *root, 3327 struct btrfs_path *path, int level) 3328 { 3329 u64 lower_gen; 3330 struct extent_buffer *lower; 3331 struct extent_buffer *c; 3332 struct extent_buffer *old; 3333 struct btrfs_disk_key lower_key; 3334 3335 BUG_ON(path->nodes[level]); 3336 BUG_ON(path->nodes[level-1] != root->node); 3337 3338 lower = path->nodes[level-1]; 3339 if (level == 1) 3340 btrfs_item_key(lower, &lower_key, 0); 3341 else 3342 btrfs_node_key(lower, &lower_key, 0); 3343 3344 c = btrfs_alloc_tree_block(trans, root, 0, root->root_key.objectid, 3345 &lower_key, level, root->node->start, 0); 3346 if (IS_ERR(c)) 3347 return PTR_ERR(c); 3348 3349 root_add_used(root, root->nodesize); 3350 3351 memzero_extent_buffer(c, 0, sizeof(struct btrfs_header)); 3352 btrfs_set_header_nritems(c, 1); 3353 btrfs_set_header_level(c, level); 3354 btrfs_set_header_bytenr(c, c->start); 3355 btrfs_set_header_generation(c, trans->transid); 3356 btrfs_set_header_backref_rev(c, BTRFS_MIXED_BACKREF_REV); 3357 btrfs_set_header_owner(c, root->root_key.objectid); 3358 3359 write_extent_buffer_fsid(c, root->fs_info->fsid); 3360 write_extent_buffer_chunk_tree_uuid(c, root->fs_info->chunk_tree_uuid); 3361 3362 btrfs_set_node_key(c, &lower_key, 0); 3363 btrfs_set_node_blockptr(c, 0, lower->start); 3364 lower_gen = btrfs_header_generation(lower); 3365 WARN_ON(lower_gen != trans->transid); 3366 3367 btrfs_set_node_ptr_generation(c, 0, lower_gen); 3368 3369 btrfs_mark_buffer_dirty(c); 3370 3371 old = root->node; 3372 tree_mod_log_set_root_pointer(root, c, 0); 3373 rcu_assign_pointer(root->node, c); 3374 3375 /* the super has an extra ref to root->node */ 3376 free_extent_buffer(old); 3377 3378 add_root_to_dirty_list(root); 3379 extent_buffer_get(c); 3380 path->nodes[level] = c; 3381 path->locks[level] = BTRFS_WRITE_LOCK_BLOCKING; 3382 path->slots[level] = 0; 3383 return 0; 3384 } 3385 3386 /* 3387 * worker function to insert a single pointer in a node. 3388 * the node should have enough room for the pointer already 3389 * 3390 * slot and level indicate where you want the key to go, and 3391 * blocknr is the block the key points to. 3392 */ 3393 static void insert_ptr(struct btrfs_trans_handle *trans, 3394 struct btrfs_root *root, struct btrfs_path *path, 3395 struct btrfs_disk_key *key, u64 bytenr, 3396 int slot, int level) 3397 { 3398 struct extent_buffer *lower; 3399 int nritems; 3400 int ret; 3401 3402 BUG_ON(!path->nodes[level]); 3403 btrfs_assert_tree_locked(path->nodes[level]); 3404 lower = path->nodes[level]; 3405 nritems = btrfs_header_nritems(lower); 3406 BUG_ON(slot > nritems); 3407 BUG_ON(nritems == BTRFS_NODEPTRS_PER_BLOCK(root)); 3408 if (slot != nritems) { 3409 if (level) 3410 tree_mod_log_eb_move(root->fs_info, lower, slot + 1, 3411 slot, nritems - slot); 3412 memmove_extent_buffer(lower, 3413 btrfs_node_key_ptr_offset(slot + 1), 3414 btrfs_node_key_ptr_offset(slot), 3415 (nritems - slot) * sizeof(struct btrfs_key_ptr)); 3416 } 3417 if (level) { 3418 ret = tree_mod_log_insert_key(root->fs_info, lower, slot, 3419 MOD_LOG_KEY_ADD, GFP_NOFS); 3420 BUG_ON(ret < 0); 3421 } 3422 btrfs_set_node_key(lower, key, slot); 3423 btrfs_set_node_blockptr(lower, slot, bytenr); 3424 WARN_ON(trans->transid == 0); 3425 btrfs_set_node_ptr_generation(lower, slot, trans->transid); 3426 btrfs_set_header_nritems(lower, nritems + 1); 3427 btrfs_mark_buffer_dirty(lower); 3428 } 3429 3430 /* 3431 * split the node at the specified level in path in two. 3432 * The path is corrected to point to the appropriate node after the split 3433 * 3434 * Before splitting this tries to make some room in the node by pushing 3435 * left and right, if either one works, it returns right away. 3436 * 3437 * returns 0 on success and < 0 on failure 3438 */ 3439 static noinline int split_node(struct btrfs_trans_handle *trans, 3440 struct btrfs_root *root, 3441 struct btrfs_path *path, int level) 3442 { 3443 struct extent_buffer *c; 3444 struct extent_buffer *split; 3445 struct btrfs_disk_key disk_key; 3446 int mid; 3447 int ret; 3448 u32 c_nritems; 3449 3450 c = path->nodes[level]; 3451 WARN_ON(btrfs_header_generation(c) != trans->transid); 3452 if (c == root->node) { 3453 /* 3454 * trying to split the root, lets make a new one 3455 * 3456 * tree mod log: We don't log_removal old root in 3457 * insert_new_root, because that root buffer will be kept as a 3458 * normal node. We are going to log removal of half of the 3459 * elements below with tree_mod_log_eb_copy. We're holding a 3460 * tree lock on the buffer, which is why we cannot race with 3461 * other tree_mod_log users. 3462 */ 3463 ret = insert_new_root(trans, root, path, level + 1); 3464 if (ret) 3465 return ret; 3466 } else { 3467 ret = push_nodes_for_insert(trans, root, path, level); 3468 c = path->nodes[level]; 3469 if (!ret && btrfs_header_nritems(c) < 3470 BTRFS_NODEPTRS_PER_BLOCK(root) - 3) 3471 return 0; 3472 if (ret < 0) 3473 return ret; 3474 } 3475 3476 c_nritems = btrfs_header_nritems(c); 3477 mid = (c_nritems + 1) / 2; 3478 btrfs_node_key(c, &disk_key, mid); 3479 3480 split = btrfs_alloc_tree_block(trans, root, 0, root->root_key.objectid, 3481 &disk_key, level, c->start, 0); 3482 if (IS_ERR(split)) 3483 return PTR_ERR(split); 3484 3485 root_add_used(root, root->nodesize); 3486 3487 memzero_extent_buffer(split, 0, sizeof(struct btrfs_header)); 3488 btrfs_set_header_level(split, btrfs_header_level(c)); 3489 btrfs_set_header_bytenr(split, split->start); 3490 btrfs_set_header_generation(split, trans->transid); 3491 btrfs_set_header_backref_rev(split, BTRFS_MIXED_BACKREF_REV); 3492 btrfs_set_header_owner(split, root->root_key.objectid); 3493 write_extent_buffer_fsid(split, root->fs_info->fsid); 3494 write_extent_buffer_chunk_tree_uuid(split, 3495 root->fs_info->chunk_tree_uuid); 3496 3497 ret = tree_mod_log_eb_copy(root->fs_info, split, c, 0, 3498 mid, c_nritems - mid); 3499 if (ret) { 3500 btrfs_abort_transaction(trans, ret); 3501 return ret; 3502 } 3503 copy_extent_buffer(split, c, 3504 btrfs_node_key_ptr_offset(0), 3505 btrfs_node_key_ptr_offset(mid), 3506 (c_nritems - mid) * sizeof(struct btrfs_key_ptr)); 3507 btrfs_set_header_nritems(split, c_nritems - mid); 3508 btrfs_set_header_nritems(c, mid); 3509 ret = 0; 3510 3511 btrfs_mark_buffer_dirty(c); 3512 btrfs_mark_buffer_dirty(split); 3513 3514 insert_ptr(trans, root, path, &disk_key, split->start, 3515 path->slots[level + 1] + 1, level + 1); 3516 3517 if (path->slots[level] >= mid) { 3518 path->slots[level] -= mid; 3519 btrfs_tree_unlock(c); 3520 free_extent_buffer(c); 3521 path->nodes[level] = split; 3522 path->slots[level + 1] += 1; 3523 } else { 3524 btrfs_tree_unlock(split); 3525 free_extent_buffer(split); 3526 } 3527 return ret; 3528 } 3529 3530 /* 3531 * how many bytes are required to store the items in a leaf. start 3532 * and nr indicate which items in the leaf to check. This totals up the 3533 * space used both by the item structs and the item data 3534 */ 3535 static int leaf_space_used(struct extent_buffer *l, int start, int nr) 3536 { 3537 struct btrfs_item *start_item; 3538 struct btrfs_item *end_item; 3539 struct btrfs_map_token token; 3540 int data_len; 3541 int nritems = btrfs_header_nritems(l); 3542 int end = min(nritems, start + nr) - 1; 3543 3544 if (!nr) 3545 return 0; 3546 btrfs_init_map_token(&token); 3547 start_item = btrfs_item_nr(start); 3548 end_item = btrfs_item_nr(end); 3549 data_len = btrfs_token_item_offset(l, start_item, &token) + 3550 btrfs_token_item_size(l, start_item, &token); 3551 data_len = data_len - btrfs_token_item_offset(l, end_item, &token); 3552 data_len += sizeof(struct btrfs_item) * nr; 3553 WARN_ON(data_len < 0); 3554 return data_len; 3555 } 3556 3557 /* 3558 * The space between the end of the leaf items and 3559 * the start of the leaf data. IOW, how much room 3560 * the leaf has left for both items and data 3561 */ 3562 noinline int btrfs_leaf_free_space(struct btrfs_root *root, 3563 struct extent_buffer *leaf) 3564 { 3565 int nritems = btrfs_header_nritems(leaf); 3566 int ret; 3567 ret = BTRFS_LEAF_DATA_SIZE(root) - leaf_space_used(leaf, 0, nritems); 3568 if (ret < 0) { 3569 btrfs_crit(root->fs_info, 3570 "leaf free space ret %d, leaf data size %lu, used %d nritems %d", 3571 ret, (unsigned long) BTRFS_LEAF_DATA_SIZE(root), 3572 leaf_space_used(leaf, 0, nritems), nritems); 3573 } 3574 return ret; 3575 } 3576 3577 /* 3578 * min slot controls the lowest index we're willing to push to the 3579 * right. We'll push up to and including min_slot, but no lower 3580 */ 3581 static noinline int __push_leaf_right(struct btrfs_trans_handle *trans, 3582 struct btrfs_root *root, 3583 struct btrfs_path *path, 3584 int data_size, int empty, 3585 struct extent_buffer *right, 3586 int free_space, u32 left_nritems, 3587 u32 min_slot) 3588 { 3589 struct extent_buffer *left = path->nodes[0]; 3590 struct extent_buffer *upper = path->nodes[1]; 3591 struct btrfs_map_token token; 3592 struct btrfs_disk_key disk_key; 3593 int slot; 3594 u32 i; 3595 int push_space = 0; 3596 int push_items = 0; 3597 struct btrfs_item *item; 3598 u32 nr; 3599 u32 right_nritems; 3600 u32 data_end; 3601 u32 this_item_size; 3602 3603 btrfs_init_map_token(&token); 3604 3605 if (empty) 3606 nr = 0; 3607 else 3608 nr = max_t(u32, 1, min_slot); 3609 3610 if (path->slots[0] >= left_nritems) 3611 push_space += data_size; 3612 3613 slot = path->slots[1]; 3614 i = left_nritems - 1; 3615 while (i >= nr) { 3616 item = btrfs_item_nr(i); 3617 3618 if (!empty && push_items > 0) { 3619 if (path->slots[0] > i) 3620 break; 3621 if (path->slots[0] == i) { 3622 int space = btrfs_leaf_free_space(root, left); 3623 if (space + push_space * 2 > free_space) 3624 break; 3625 } 3626 } 3627 3628 if (path->slots[0] == i) 3629 push_space += data_size; 3630 3631 this_item_size = btrfs_item_size(left, item); 3632 if (this_item_size + sizeof(*item) + push_space > free_space) 3633 break; 3634 3635 push_items++; 3636 push_space += this_item_size + sizeof(*item); 3637 if (i == 0) 3638 break; 3639 i--; 3640 } 3641 3642 if (push_items == 0) 3643 goto out_unlock; 3644 3645 WARN_ON(!empty && push_items == left_nritems); 3646 3647 /* push left to right */ 3648 right_nritems = btrfs_header_nritems(right); 3649 3650 push_space = btrfs_item_end_nr(left, left_nritems - push_items); 3651 push_space -= leaf_data_end(root, left); 3652 3653 /* make room in the right data area */ 3654 data_end = leaf_data_end(root, right); 3655 memmove_extent_buffer(right, 3656 btrfs_leaf_data(right) + data_end - push_space, 3657 btrfs_leaf_data(right) + data_end, 3658 BTRFS_LEAF_DATA_SIZE(root) - data_end); 3659 3660 /* copy from the left data area */ 3661 copy_extent_buffer(right, left, btrfs_leaf_data(right) + 3662 BTRFS_LEAF_DATA_SIZE(root) - push_space, 3663 btrfs_leaf_data(left) + leaf_data_end(root, left), 3664 push_space); 3665 3666 memmove_extent_buffer(right, btrfs_item_nr_offset(push_items), 3667 btrfs_item_nr_offset(0), 3668 right_nritems * sizeof(struct btrfs_item)); 3669 3670 /* copy the items from left to right */ 3671 copy_extent_buffer(right, left, btrfs_item_nr_offset(0), 3672 btrfs_item_nr_offset(left_nritems - push_items), 3673 push_items * sizeof(struct btrfs_item)); 3674 3675 /* update the item pointers */ 3676 right_nritems += push_items; 3677 btrfs_set_header_nritems(right, right_nritems); 3678 push_space = BTRFS_LEAF_DATA_SIZE(root); 3679 for (i = 0; i < right_nritems; i++) { 3680 item = btrfs_item_nr(i); 3681 push_space -= btrfs_token_item_size(right, item, &token); 3682 btrfs_set_token_item_offset(right, item, push_space, &token); 3683 } 3684 3685 left_nritems -= push_items; 3686 btrfs_set_header_nritems(left, left_nritems); 3687 3688 if (left_nritems) 3689 btrfs_mark_buffer_dirty(left); 3690 else 3691 clean_tree_block(trans, root->fs_info, left); 3692 3693 btrfs_mark_buffer_dirty(right); 3694 3695 btrfs_item_key(right, &disk_key, 0); 3696 btrfs_set_node_key(upper, &disk_key, slot + 1); 3697 btrfs_mark_buffer_dirty(upper); 3698 3699 /* then fixup the leaf pointer in the path */ 3700 if (path->slots[0] >= left_nritems) { 3701 path->slots[0] -= left_nritems; 3702 if (btrfs_header_nritems(path->nodes[0]) == 0) 3703 clean_tree_block(trans, root->fs_info, path->nodes[0]); 3704 btrfs_tree_unlock(path->nodes[0]); 3705 free_extent_buffer(path->nodes[0]); 3706 path->nodes[0] = right; 3707 path->slots[1] += 1; 3708 } else { 3709 btrfs_tree_unlock(right); 3710 free_extent_buffer(right); 3711 } 3712 return 0; 3713 3714 out_unlock: 3715 btrfs_tree_unlock(right); 3716 free_extent_buffer(right); 3717 return 1; 3718 } 3719 3720 /* 3721 * push some data in the path leaf to the right, trying to free up at 3722 * least data_size bytes. returns zero if the push worked, nonzero otherwise 3723 * 3724 * returns 1 if the push failed because the other node didn't have enough 3725 * room, 0 if everything worked out and < 0 if there were major errors. 3726 * 3727 * this will push starting from min_slot to the end of the leaf. It won't 3728 * push any slot lower than min_slot 3729 */ 3730 static int push_leaf_right(struct btrfs_trans_handle *trans, struct btrfs_root 3731 *root, struct btrfs_path *path, 3732 int min_data_size, int data_size, 3733 int empty, u32 min_slot) 3734 { 3735 struct extent_buffer *left = path->nodes[0]; 3736 struct extent_buffer *right; 3737 struct extent_buffer *upper; 3738 int slot; 3739 int free_space; 3740 u32 left_nritems; 3741 int ret; 3742 3743 if (!path->nodes[1]) 3744 return 1; 3745 3746 slot = path->slots[1]; 3747 upper = path->nodes[1]; 3748 if (slot >= btrfs_header_nritems(upper) - 1) 3749 return 1; 3750 3751 btrfs_assert_tree_locked(path->nodes[1]); 3752 3753 right = read_node_slot(root, upper, slot + 1); 3754 /* 3755 * slot + 1 is not valid or we fail to read the right node, 3756 * no big deal, just return. 3757 */ 3758 if (IS_ERR(right)) 3759 return 1; 3760 3761 btrfs_tree_lock(right); 3762 btrfs_set_lock_blocking(right); 3763 3764 free_space = btrfs_leaf_free_space(root, right); 3765 if (free_space < data_size) 3766 goto out_unlock; 3767 3768 /* cow and double check */ 3769 ret = btrfs_cow_block(trans, root, right, upper, 3770 slot + 1, &right); 3771 if (ret) 3772 goto out_unlock; 3773 3774 free_space = btrfs_leaf_free_space(root, right); 3775 if (free_space < data_size) 3776 goto out_unlock; 3777 3778 left_nritems = btrfs_header_nritems(left); 3779 if (left_nritems == 0) 3780 goto out_unlock; 3781 3782 if (path->slots[0] == left_nritems && !empty) { 3783 /* Key greater than all keys in the leaf, right neighbor has 3784 * enough room for it and we're not emptying our leaf to delete 3785 * it, therefore use right neighbor to insert the new item and 3786 * no need to touch/dirty our left leaft. */ 3787 btrfs_tree_unlock(left); 3788 free_extent_buffer(left); 3789 path->nodes[0] = right; 3790 path->slots[0] = 0; 3791 path->slots[1]++; 3792 return 0; 3793 } 3794 3795 return __push_leaf_right(trans, root, path, min_data_size, empty, 3796 right, free_space, left_nritems, min_slot); 3797 out_unlock: 3798 btrfs_tree_unlock(right); 3799 free_extent_buffer(right); 3800 return 1; 3801 } 3802 3803 /* 3804 * push some data in the path leaf to the left, trying to free up at 3805 * least data_size bytes. returns zero if the push worked, nonzero otherwise 3806 * 3807 * max_slot can put a limit on how far into the leaf we'll push items. The 3808 * item at 'max_slot' won't be touched. Use (u32)-1 to make us do all the 3809 * items 3810 */ 3811 static noinline int __push_leaf_left(struct btrfs_trans_handle *trans, 3812 struct btrfs_root *root, 3813 struct btrfs_path *path, int data_size, 3814 int empty, struct extent_buffer *left, 3815 int free_space, u32 right_nritems, 3816 u32 max_slot) 3817 { 3818 struct btrfs_disk_key disk_key; 3819 struct extent_buffer *right = path->nodes[0]; 3820 int i; 3821 int push_space = 0; 3822 int push_items = 0; 3823 struct btrfs_item *item; 3824 u32 old_left_nritems; 3825 u32 nr; 3826 int ret = 0; 3827 u32 this_item_size; 3828 u32 old_left_item_size; 3829 struct btrfs_map_token token; 3830 3831 btrfs_init_map_token(&token); 3832 3833 if (empty) 3834 nr = min(right_nritems, max_slot); 3835 else 3836 nr = min(right_nritems - 1, max_slot); 3837 3838 for (i = 0; i < nr; i++) { 3839 item = btrfs_item_nr(i); 3840 3841 if (!empty && push_items > 0) { 3842 if (path->slots[0] < i) 3843 break; 3844 if (path->slots[0] == i) { 3845 int space = btrfs_leaf_free_space(root, right); 3846 if (space + push_space * 2 > free_space) 3847 break; 3848 } 3849 } 3850 3851 if (path->slots[0] == i) 3852 push_space += data_size; 3853 3854 this_item_size = btrfs_item_size(right, item); 3855 if (this_item_size + sizeof(*item) + push_space > free_space) 3856 break; 3857 3858 push_items++; 3859 push_space += this_item_size + sizeof(*item); 3860 } 3861 3862 if (push_items == 0) { 3863 ret = 1; 3864 goto out; 3865 } 3866 WARN_ON(!empty && push_items == btrfs_header_nritems(right)); 3867 3868 /* push data from right to left */ 3869 copy_extent_buffer(left, right, 3870 btrfs_item_nr_offset(btrfs_header_nritems(left)), 3871 btrfs_item_nr_offset(0), 3872 push_items * sizeof(struct btrfs_item)); 3873 3874 push_space = BTRFS_LEAF_DATA_SIZE(root) - 3875 btrfs_item_offset_nr(right, push_items - 1); 3876 3877 copy_extent_buffer(left, right, btrfs_leaf_data(left) + 3878 leaf_data_end(root, left) - push_space, 3879 btrfs_leaf_data(right) + 3880 btrfs_item_offset_nr(right, push_items - 1), 3881 push_space); 3882 old_left_nritems = btrfs_header_nritems(left); 3883 BUG_ON(old_left_nritems <= 0); 3884 3885 old_left_item_size = btrfs_item_offset_nr(left, old_left_nritems - 1); 3886 for (i = old_left_nritems; i < old_left_nritems + push_items; i++) { 3887 u32 ioff; 3888 3889 item = btrfs_item_nr(i); 3890 3891 ioff = btrfs_token_item_offset(left, item, &token); 3892 btrfs_set_token_item_offset(left, item, 3893 ioff - (BTRFS_LEAF_DATA_SIZE(root) - old_left_item_size), 3894 &token); 3895 } 3896 btrfs_set_header_nritems(left, old_left_nritems + push_items); 3897 3898 /* fixup right node */ 3899 if (push_items > right_nritems) 3900 WARN(1, KERN_CRIT "push items %d nr %u\n", push_items, 3901 right_nritems); 3902 3903 if (push_items < right_nritems) { 3904 push_space = btrfs_item_offset_nr(right, push_items - 1) - 3905 leaf_data_end(root, right); 3906 memmove_extent_buffer(right, btrfs_leaf_data(right) + 3907 BTRFS_LEAF_DATA_SIZE(root) - push_space, 3908 btrfs_leaf_data(right) + 3909 leaf_data_end(root, right), push_space); 3910 3911 memmove_extent_buffer(right, btrfs_item_nr_offset(0), 3912 btrfs_item_nr_offset(push_items), 3913 (btrfs_header_nritems(right) - push_items) * 3914 sizeof(struct btrfs_item)); 3915 } 3916 right_nritems -= push_items; 3917 btrfs_set_header_nritems(right, right_nritems); 3918 push_space = BTRFS_LEAF_DATA_SIZE(root); 3919 for (i = 0; i < right_nritems; i++) { 3920 item = btrfs_item_nr(i); 3921 3922 push_space = push_space - btrfs_token_item_size(right, 3923 item, &token); 3924 btrfs_set_token_item_offset(right, item, push_space, &token); 3925 } 3926 3927 btrfs_mark_buffer_dirty(left); 3928 if (right_nritems) 3929 btrfs_mark_buffer_dirty(right); 3930 else 3931 clean_tree_block(trans, root->fs_info, right); 3932 3933 btrfs_item_key(right, &disk_key, 0); 3934 fixup_low_keys(root->fs_info, path, &disk_key, 1); 3935 3936 /* then fixup the leaf pointer in the path */ 3937 if (path->slots[0] < push_items) { 3938 path->slots[0] += old_left_nritems; 3939 btrfs_tree_unlock(path->nodes[0]); 3940 free_extent_buffer(path->nodes[0]); 3941 path->nodes[0] = left; 3942 path->slots[1] -= 1; 3943 } else { 3944 btrfs_tree_unlock(left); 3945 free_extent_buffer(left); 3946 path->slots[0] -= push_items; 3947 } 3948 BUG_ON(path->slots[0] < 0); 3949 return ret; 3950 out: 3951 btrfs_tree_unlock(left); 3952 free_extent_buffer(left); 3953 return ret; 3954 } 3955 3956 /* 3957 * push some data in the path leaf to the left, trying to free up at 3958 * least data_size bytes. returns zero if the push worked, nonzero otherwise 3959 * 3960 * max_slot can put a limit on how far into the leaf we'll push items. The 3961 * item at 'max_slot' won't be touched. Use (u32)-1 to make us push all the 3962 * items 3963 */ 3964 static int push_leaf_left(struct btrfs_trans_handle *trans, struct btrfs_root 3965 *root, struct btrfs_path *path, int min_data_size, 3966 int data_size, int empty, u32 max_slot) 3967 { 3968 struct extent_buffer *right = path->nodes[0]; 3969 struct extent_buffer *left; 3970 int slot; 3971 int free_space; 3972 u32 right_nritems; 3973 int ret = 0; 3974 3975 slot = path->slots[1]; 3976 if (slot == 0) 3977 return 1; 3978 if (!path->nodes[1]) 3979 return 1; 3980 3981 right_nritems = btrfs_header_nritems(right); 3982 if (right_nritems == 0) 3983 return 1; 3984 3985 btrfs_assert_tree_locked(path->nodes[1]); 3986 3987 left = read_node_slot(root, path->nodes[1], slot - 1); 3988 /* 3989 * slot - 1 is not valid or we fail to read the left node, 3990 * no big deal, just return. 3991 */ 3992 if (IS_ERR(left)) 3993 return 1; 3994 3995 btrfs_tree_lock(left); 3996 btrfs_set_lock_blocking(left); 3997 3998 free_space = btrfs_leaf_free_space(root, left); 3999 if (free_space < data_size) { 4000 ret = 1; 4001 goto out; 4002 } 4003 4004 /* cow and double check */ 4005 ret = btrfs_cow_block(trans, root, left, 4006 path->nodes[1], slot - 1, &left); 4007 if (ret) { 4008 /* we hit -ENOSPC, but it isn't fatal here */ 4009 if (ret == -ENOSPC) 4010 ret = 1; 4011 goto out; 4012 } 4013 4014 free_space = btrfs_leaf_free_space(root, left); 4015 if (free_space < data_size) { 4016 ret = 1; 4017 goto out; 4018 } 4019 4020 return __push_leaf_left(trans, root, path, min_data_size, 4021 empty, left, free_space, right_nritems, 4022 max_slot); 4023 out: 4024 btrfs_tree_unlock(left); 4025 free_extent_buffer(left); 4026 return ret; 4027 } 4028 4029 /* 4030 * split the path's leaf in two, making sure there is at least data_size 4031 * available for the resulting leaf level of the path. 4032 */ 4033 static noinline void copy_for_split(struct btrfs_trans_handle *trans, 4034 struct btrfs_root *root, 4035 struct btrfs_path *path, 4036 struct extent_buffer *l, 4037 struct extent_buffer *right, 4038 int slot, int mid, int nritems) 4039 { 4040 int data_copy_size; 4041 int rt_data_off; 4042 int i; 4043 struct btrfs_disk_key disk_key; 4044 struct btrfs_map_token token; 4045 4046 btrfs_init_map_token(&token); 4047 4048 nritems = nritems - mid; 4049 btrfs_set_header_nritems(right, nritems); 4050 data_copy_size = btrfs_item_end_nr(l, mid) - leaf_data_end(root, l); 4051 4052 copy_extent_buffer(right, l, btrfs_item_nr_offset(0), 4053 btrfs_item_nr_offset(mid), 4054 nritems * sizeof(struct btrfs_item)); 4055 4056 copy_extent_buffer(right, l, 4057 btrfs_leaf_data(right) + BTRFS_LEAF_DATA_SIZE(root) - 4058 data_copy_size, btrfs_leaf_data(l) + 4059 leaf_data_end(root, l), data_copy_size); 4060 4061 rt_data_off = BTRFS_LEAF_DATA_SIZE(root) - 4062 btrfs_item_end_nr(l, mid); 4063 4064 for (i = 0; i < nritems; i++) { 4065 struct btrfs_item *item = btrfs_item_nr(i); 4066 u32 ioff; 4067 4068 ioff = btrfs_token_item_offset(right, item, &token); 4069 btrfs_set_token_item_offset(right, item, 4070 ioff + rt_data_off, &token); 4071 } 4072 4073 btrfs_set_header_nritems(l, mid); 4074 btrfs_item_key(right, &disk_key, 0); 4075 insert_ptr(trans, root, path, &disk_key, right->start, 4076 path->slots[1] + 1, 1); 4077 4078 btrfs_mark_buffer_dirty(right); 4079 btrfs_mark_buffer_dirty(l); 4080 BUG_ON(path->slots[0] != slot); 4081 4082 if (mid <= slot) { 4083 btrfs_tree_unlock(path->nodes[0]); 4084 free_extent_buffer(path->nodes[0]); 4085 path->nodes[0] = right; 4086 path->slots[0] -= mid; 4087 path->slots[1] += 1; 4088 } else { 4089 btrfs_tree_unlock(right); 4090 free_extent_buffer(right); 4091 } 4092 4093 BUG_ON(path->slots[0] < 0); 4094 } 4095 4096 /* 4097 * double splits happen when we need to insert a big item in the middle 4098 * of a leaf. A double split can leave us with 3 mostly empty leaves: 4099 * leaf: [ slots 0 - N] [ our target ] [ N + 1 - total in leaf ] 4100 * A B C 4101 * 4102 * We avoid this by trying to push the items on either side of our target 4103 * into the adjacent leaves. If all goes well we can avoid the double split 4104 * completely. 4105 */ 4106 static noinline int push_for_double_split(struct btrfs_trans_handle *trans, 4107 struct btrfs_root *root, 4108 struct btrfs_path *path, 4109 int data_size) 4110 { 4111 int ret; 4112 int progress = 0; 4113 int slot; 4114 u32 nritems; 4115 int space_needed = data_size; 4116 4117 slot = path->slots[0]; 4118 if (slot < btrfs_header_nritems(path->nodes[0])) 4119 space_needed -= btrfs_leaf_free_space(root, path->nodes[0]); 4120 4121 /* 4122 * try to push all the items after our slot into the 4123 * right leaf 4124 */ 4125 ret = push_leaf_right(trans, root, path, 1, space_needed, 0, slot); 4126 if (ret < 0) 4127 return ret; 4128 4129 if (ret == 0) 4130 progress++; 4131 4132 nritems = btrfs_header_nritems(path->nodes[0]); 4133 /* 4134 * our goal is to get our slot at the start or end of a leaf. If 4135 * we've done so we're done 4136 */ 4137 if (path->slots[0] == 0 || path->slots[0] == nritems) 4138 return 0; 4139 4140 if (btrfs_leaf_free_space(root, path->nodes[0]) >= data_size) 4141 return 0; 4142 4143 /* try to push all the items before our slot into the next leaf */ 4144 slot = path->slots[0]; 4145 ret = push_leaf_left(trans, root, path, 1, space_needed, 0, slot); 4146 if (ret < 0) 4147 return ret; 4148 4149 if (ret == 0) 4150 progress++; 4151 4152 if (progress) 4153 return 0; 4154 return 1; 4155 } 4156 4157 /* 4158 * split the path's leaf in two, making sure there is at least data_size 4159 * available for the resulting leaf level of the path. 4160 * 4161 * returns 0 if all went well and < 0 on failure. 4162 */ 4163 static noinline int split_leaf(struct btrfs_trans_handle *trans, 4164 struct btrfs_root *root, 4165 struct btrfs_key *ins_key, 4166 struct btrfs_path *path, int data_size, 4167 int extend) 4168 { 4169 struct btrfs_disk_key disk_key; 4170 struct extent_buffer *l; 4171 u32 nritems; 4172 int mid; 4173 int slot; 4174 struct extent_buffer *right; 4175 struct btrfs_fs_info *fs_info = root->fs_info; 4176 int ret = 0; 4177 int wret; 4178 int split; 4179 int num_doubles = 0; 4180 int tried_avoid_double = 0; 4181 4182 l = path->nodes[0]; 4183 slot = path->slots[0]; 4184 if (extend && data_size + btrfs_item_size_nr(l, slot) + 4185 sizeof(struct btrfs_item) > BTRFS_LEAF_DATA_SIZE(root)) 4186 return -EOVERFLOW; 4187 4188 /* first try to make some room by pushing left and right */ 4189 if (data_size && path->nodes[1]) { 4190 int space_needed = data_size; 4191 4192 if (slot < btrfs_header_nritems(l)) 4193 space_needed -= btrfs_leaf_free_space(root, l); 4194 4195 wret = push_leaf_right(trans, root, path, space_needed, 4196 space_needed, 0, 0); 4197 if (wret < 0) 4198 return wret; 4199 if (wret) { 4200 wret = push_leaf_left(trans, root, path, space_needed, 4201 space_needed, 0, (u32)-1); 4202 if (wret < 0) 4203 return wret; 4204 } 4205 l = path->nodes[0]; 4206 4207 /* did the pushes work? */ 4208 if (btrfs_leaf_free_space(root, l) >= data_size) 4209 return 0; 4210 } 4211 4212 if (!path->nodes[1]) { 4213 ret = insert_new_root(trans, root, path, 1); 4214 if (ret) 4215 return ret; 4216 } 4217 again: 4218 split = 1; 4219 l = path->nodes[0]; 4220 slot = path->slots[0]; 4221 nritems = btrfs_header_nritems(l); 4222 mid = (nritems + 1) / 2; 4223 4224 if (mid <= slot) { 4225 if (nritems == 1 || 4226 leaf_space_used(l, mid, nritems - mid) + data_size > 4227 BTRFS_LEAF_DATA_SIZE(root)) { 4228 if (slot >= nritems) { 4229 split = 0; 4230 } else { 4231 mid = slot; 4232 if (mid != nritems && 4233 leaf_space_used(l, mid, nritems - mid) + 4234 data_size > BTRFS_LEAF_DATA_SIZE(root)) { 4235 if (data_size && !tried_avoid_double) 4236 goto push_for_double; 4237 split = 2; 4238 } 4239 } 4240 } 4241 } else { 4242 if (leaf_space_used(l, 0, mid) + data_size > 4243 BTRFS_LEAF_DATA_SIZE(root)) { 4244 if (!extend && data_size && slot == 0) { 4245 split = 0; 4246 } else if ((extend || !data_size) && slot == 0) { 4247 mid = 1; 4248 } else { 4249 mid = slot; 4250 if (mid != nritems && 4251 leaf_space_used(l, mid, nritems - mid) + 4252 data_size > BTRFS_LEAF_DATA_SIZE(root)) { 4253 if (data_size && !tried_avoid_double) 4254 goto push_for_double; 4255 split = 2; 4256 } 4257 } 4258 } 4259 } 4260 4261 if (split == 0) 4262 btrfs_cpu_key_to_disk(&disk_key, ins_key); 4263 else 4264 btrfs_item_key(l, &disk_key, mid); 4265 4266 right = btrfs_alloc_tree_block(trans, root, 0, root->root_key.objectid, 4267 &disk_key, 0, l->start, 0); 4268 if (IS_ERR(right)) 4269 return PTR_ERR(right); 4270 4271 root_add_used(root, root->nodesize); 4272 4273 memzero_extent_buffer(right, 0, sizeof(struct btrfs_header)); 4274 btrfs_set_header_bytenr(right, right->start); 4275 btrfs_set_header_generation(right, trans->transid); 4276 btrfs_set_header_backref_rev(right, BTRFS_MIXED_BACKREF_REV); 4277 btrfs_set_header_owner(right, root->root_key.objectid); 4278 btrfs_set_header_level(right, 0); 4279 write_extent_buffer_fsid(right, fs_info->fsid); 4280 write_extent_buffer_chunk_tree_uuid(right, fs_info->chunk_tree_uuid); 4281 4282 if (split == 0) { 4283 if (mid <= slot) { 4284 btrfs_set_header_nritems(right, 0); 4285 insert_ptr(trans, root, path, &disk_key, right->start, 4286 path->slots[1] + 1, 1); 4287 btrfs_tree_unlock(path->nodes[0]); 4288 free_extent_buffer(path->nodes[0]); 4289 path->nodes[0] = right; 4290 path->slots[0] = 0; 4291 path->slots[1] += 1; 4292 } else { 4293 btrfs_set_header_nritems(right, 0); 4294 insert_ptr(trans, root, path, &disk_key, right->start, 4295 path->slots[1], 1); 4296 btrfs_tree_unlock(path->nodes[0]); 4297 free_extent_buffer(path->nodes[0]); 4298 path->nodes[0] = right; 4299 path->slots[0] = 0; 4300 if (path->slots[1] == 0) 4301 fixup_low_keys(fs_info, path, &disk_key, 1); 4302 } 4303 /* 4304 * We create a new leaf 'right' for the required ins_len and 4305 * we'll do btrfs_mark_buffer_dirty() on this leaf after copying 4306 * the content of ins_len to 'right'. 4307 */ 4308 return ret; 4309 } 4310 4311 copy_for_split(trans, root, path, l, right, slot, mid, nritems); 4312 4313 if (split == 2) { 4314 BUG_ON(num_doubles != 0); 4315 num_doubles++; 4316 goto again; 4317 } 4318 4319 return 0; 4320 4321 push_for_double: 4322 push_for_double_split(trans, root, path, data_size); 4323 tried_avoid_double = 1; 4324 if (btrfs_leaf_free_space(root, path->nodes[0]) >= data_size) 4325 return 0; 4326 goto again; 4327 } 4328 4329 static noinline int setup_leaf_for_split(struct btrfs_trans_handle *trans, 4330 struct btrfs_root *root, 4331 struct btrfs_path *path, int ins_len) 4332 { 4333 struct btrfs_key key; 4334 struct extent_buffer *leaf; 4335 struct btrfs_file_extent_item *fi; 4336 u64 extent_len = 0; 4337 u32 item_size; 4338 int ret; 4339 4340 leaf = path->nodes[0]; 4341 btrfs_item_key_to_cpu(leaf, &key, path->slots[0]); 4342 4343 BUG_ON(key.type != BTRFS_EXTENT_DATA_KEY && 4344 key.type != BTRFS_EXTENT_CSUM_KEY); 4345 4346 if (btrfs_leaf_free_space(root, leaf) >= ins_len) 4347 return 0; 4348 4349 item_size = btrfs_item_size_nr(leaf, path->slots[0]); 4350 if (key.type == BTRFS_EXTENT_DATA_KEY) { 4351 fi = btrfs_item_ptr(leaf, path->slots[0], 4352 struct btrfs_file_extent_item); 4353 extent_len = btrfs_file_extent_num_bytes(leaf, fi); 4354 } 4355 btrfs_release_path(path); 4356 4357 path->keep_locks = 1; 4358 path->search_for_split = 1; 4359 ret = btrfs_search_slot(trans, root, &key, path, 0, 1); 4360 path->search_for_split = 0; 4361 if (ret > 0) 4362 ret = -EAGAIN; 4363 if (ret < 0) 4364 goto err; 4365 4366 ret = -EAGAIN; 4367 leaf = path->nodes[0]; 4368 /* if our item isn't there, return now */ 4369 if (item_size != btrfs_item_size_nr(leaf, path->slots[0])) 4370 goto err; 4371 4372 /* the leaf has changed, it now has room. return now */ 4373 if (btrfs_leaf_free_space(root, path->nodes[0]) >= ins_len) 4374 goto err; 4375 4376 if (key.type == BTRFS_EXTENT_DATA_KEY) { 4377 fi = btrfs_item_ptr(leaf, path->slots[0], 4378 struct btrfs_file_extent_item); 4379 if (extent_len != btrfs_file_extent_num_bytes(leaf, fi)) 4380 goto err; 4381 } 4382 4383 btrfs_set_path_blocking(path); 4384 ret = split_leaf(trans, root, &key, path, ins_len, 1); 4385 if (ret) 4386 goto err; 4387 4388 path->keep_locks = 0; 4389 btrfs_unlock_up_safe(path, 1); 4390 return 0; 4391 err: 4392 path->keep_locks = 0; 4393 return ret; 4394 } 4395 4396 static noinline int split_item(struct btrfs_trans_handle *trans, 4397 struct btrfs_root *root, 4398 struct btrfs_path *path, 4399 struct btrfs_key *new_key, 4400 unsigned long split_offset) 4401 { 4402 struct extent_buffer *leaf; 4403 struct btrfs_item *item; 4404 struct btrfs_item *new_item; 4405 int slot; 4406 char *buf; 4407 u32 nritems; 4408 u32 item_size; 4409 u32 orig_offset; 4410 struct btrfs_disk_key disk_key; 4411 4412 leaf = path->nodes[0]; 4413 BUG_ON(btrfs_leaf_free_space(root, leaf) < sizeof(struct btrfs_item)); 4414 4415 btrfs_set_path_blocking(path); 4416 4417 item = btrfs_item_nr(path->slots[0]); 4418 orig_offset = btrfs_item_offset(leaf, item); 4419 item_size = btrfs_item_size(leaf, item); 4420 4421 buf = kmalloc(item_size, GFP_NOFS); 4422 if (!buf) 4423 return -ENOMEM; 4424 4425 read_extent_buffer(leaf, buf, btrfs_item_ptr_offset(leaf, 4426 path->slots[0]), item_size); 4427 4428 slot = path->slots[0] + 1; 4429 nritems = btrfs_header_nritems(leaf); 4430 if (slot != nritems) { 4431 /* shift the items */ 4432 memmove_extent_buffer(leaf, btrfs_item_nr_offset(slot + 1), 4433 btrfs_item_nr_offset(slot), 4434 (nritems - slot) * sizeof(struct btrfs_item)); 4435 } 4436 4437 btrfs_cpu_key_to_disk(&disk_key, new_key); 4438 btrfs_set_item_key(leaf, &disk_key, slot); 4439 4440 new_item = btrfs_item_nr(slot); 4441 4442 btrfs_set_item_offset(leaf, new_item, orig_offset); 4443 btrfs_set_item_size(leaf, new_item, item_size - split_offset); 4444 4445 btrfs_set_item_offset(leaf, item, 4446 orig_offset + item_size - split_offset); 4447 btrfs_set_item_size(leaf, item, split_offset); 4448 4449 btrfs_set_header_nritems(leaf, nritems + 1); 4450 4451 /* write the data for the start of the original item */ 4452 write_extent_buffer(leaf, buf, 4453 btrfs_item_ptr_offset(leaf, path->slots[0]), 4454 split_offset); 4455 4456 /* write the data for the new item */ 4457 write_extent_buffer(leaf, buf + split_offset, 4458 btrfs_item_ptr_offset(leaf, slot), 4459 item_size - split_offset); 4460 btrfs_mark_buffer_dirty(leaf); 4461 4462 BUG_ON(btrfs_leaf_free_space(root, leaf) < 0); 4463 kfree(buf); 4464 return 0; 4465 } 4466 4467 /* 4468 * This function splits a single item into two items, 4469 * giving 'new_key' to the new item and splitting the 4470 * old one at split_offset (from the start of the item). 4471 * 4472 * The path may be released by this operation. After 4473 * the split, the path is pointing to the old item. The 4474 * new item is going to be in the same node as the old one. 4475 * 4476 * Note, the item being split must be smaller enough to live alone on 4477 * a tree block with room for one extra struct btrfs_item 4478 * 4479 * This allows us to split the item in place, keeping a lock on the 4480 * leaf the entire time. 4481 */ 4482 int btrfs_split_item(struct btrfs_trans_handle *trans, 4483 struct btrfs_root *root, 4484 struct btrfs_path *path, 4485 struct btrfs_key *new_key, 4486 unsigned long split_offset) 4487 { 4488 int ret; 4489 ret = setup_leaf_for_split(trans, root, path, 4490 sizeof(struct btrfs_item)); 4491 if (ret) 4492 return ret; 4493 4494 ret = split_item(trans, root, path, new_key, split_offset); 4495 return ret; 4496 } 4497 4498 /* 4499 * This function duplicate a item, giving 'new_key' to the new item. 4500 * It guarantees both items live in the same tree leaf and the new item 4501 * is contiguous with the original item. 4502 * 4503 * This allows us to split file extent in place, keeping a lock on the 4504 * leaf the entire time. 4505 */ 4506 int btrfs_duplicate_item(struct btrfs_trans_handle *trans, 4507 struct btrfs_root *root, 4508 struct btrfs_path *path, 4509 struct btrfs_key *new_key) 4510 { 4511 struct extent_buffer *leaf; 4512 int ret; 4513 u32 item_size; 4514 4515 leaf = path->nodes[0]; 4516 item_size = btrfs_item_size_nr(leaf, path->slots[0]); 4517 ret = setup_leaf_for_split(trans, root, path, 4518 item_size + sizeof(struct btrfs_item)); 4519 if (ret) 4520 return ret; 4521 4522 path->slots[0]++; 4523 setup_items_for_insert(root, path, new_key, &item_size, 4524 item_size, item_size + 4525 sizeof(struct btrfs_item), 1); 4526 leaf = path->nodes[0]; 4527 memcpy_extent_buffer(leaf, 4528 btrfs_item_ptr_offset(leaf, path->slots[0]), 4529 btrfs_item_ptr_offset(leaf, path->slots[0] - 1), 4530 item_size); 4531 return 0; 4532 } 4533 4534 /* 4535 * make the item pointed to by the path smaller. new_size indicates 4536 * how small to make it, and from_end tells us if we just chop bytes 4537 * off the end of the item or if we shift the item to chop bytes off 4538 * the front. 4539 */ 4540 void btrfs_truncate_item(struct btrfs_root *root, struct btrfs_path *path, 4541 u32 new_size, int from_end) 4542 { 4543 int slot; 4544 struct extent_buffer *leaf; 4545 struct btrfs_item *item; 4546 u32 nritems; 4547 unsigned int data_end; 4548 unsigned int old_data_start; 4549 unsigned int old_size; 4550 unsigned int size_diff; 4551 int i; 4552 struct btrfs_map_token token; 4553 4554 btrfs_init_map_token(&token); 4555 4556 leaf = path->nodes[0]; 4557 slot = path->slots[0]; 4558 4559 old_size = btrfs_item_size_nr(leaf, slot); 4560 if (old_size == new_size) 4561 return; 4562 4563 nritems = btrfs_header_nritems(leaf); 4564 data_end = leaf_data_end(root, leaf); 4565 4566 old_data_start = btrfs_item_offset_nr(leaf, slot); 4567 4568 size_diff = old_size - new_size; 4569 4570 BUG_ON(slot < 0); 4571 BUG_ON(slot >= nritems); 4572 4573 /* 4574 * item0..itemN ... dataN.offset..dataN.size .. data0.size 4575 */ 4576 /* first correct the data pointers */ 4577 for (i = slot; i < nritems; i++) { 4578 u32 ioff; 4579 item = btrfs_item_nr(i); 4580 4581 ioff = btrfs_token_item_offset(leaf, item, &token); 4582 btrfs_set_token_item_offset(leaf, item, 4583 ioff + size_diff, &token); 4584 } 4585 4586 /* shift the data */ 4587 if (from_end) { 4588 memmove_extent_buffer(leaf, btrfs_leaf_data(leaf) + 4589 data_end + size_diff, btrfs_leaf_data(leaf) + 4590 data_end, old_data_start + new_size - data_end); 4591 } else { 4592 struct btrfs_disk_key disk_key; 4593 u64 offset; 4594 4595 btrfs_item_key(leaf, &disk_key, slot); 4596 4597 if (btrfs_disk_key_type(&disk_key) == BTRFS_EXTENT_DATA_KEY) { 4598 unsigned long ptr; 4599 struct btrfs_file_extent_item *fi; 4600 4601 fi = btrfs_item_ptr(leaf, slot, 4602 struct btrfs_file_extent_item); 4603 fi = (struct btrfs_file_extent_item *)( 4604 (unsigned long)fi - size_diff); 4605 4606 if (btrfs_file_extent_type(leaf, fi) == 4607 BTRFS_FILE_EXTENT_INLINE) { 4608 ptr = btrfs_item_ptr_offset(leaf, slot); 4609 memmove_extent_buffer(leaf, ptr, 4610 (unsigned long)fi, 4611 BTRFS_FILE_EXTENT_INLINE_DATA_START); 4612 } 4613 } 4614 4615 memmove_extent_buffer(leaf, btrfs_leaf_data(leaf) + 4616 data_end + size_diff, btrfs_leaf_data(leaf) + 4617 data_end, old_data_start - data_end); 4618 4619 offset = btrfs_disk_key_offset(&disk_key); 4620 btrfs_set_disk_key_offset(&disk_key, offset + size_diff); 4621 btrfs_set_item_key(leaf, &disk_key, slot); 4622 if (slot == 0) 4623 fixup_low_keys(root->fs_info, path, &disk_key, 1); 4624 } 4625 4626 item = btrfs_item_nr(slot); 4627 btrfs_set_item_size(leaf, item, new_size); 4628 btrfs_mark_buffer_dirty(leaf); 4629 4630 if (btrfs_leaf_free_space(root, leaf) < 0) { 4631 btrfs_print_leaf(root, leaf); 4632 BUG(); 4633 } 4634 } 4635 4636 /* 4637 * make the item pointed to by the path bigger, data_size is the added size. 4638 */ 4639 void btrfs_extend_item(struct btrfs_root *root, struct btrfs_path *path, 4640 u32 data_size) 4641 { 4642 int slot; 4643 struct extent_buffer *leaf; 4644 struct btrfs_item *item; 4645 u32 nritems; 4646 unsigned int data_end; 4647 unsigned int old_data; 4648 unsigned int old_size; 4649 int i; 4650 struct btrfs_map_token token; 4651 4652 btrfs_init_map_token(&token); 4653 4654 leaf = path->nodes[0]; 4655 4656 nritems = btrfs_header_nritems(leaf); 4657 data_end = leaf_data_end(root, leaf); 4658 4659 if (btrfs_leaf_free_space(root, leaf) < data_size) { 4660 btrfs_print_leaf(root, leaf); 4661 BUG(); 4662 } 4663 slot = path->slots[0]; 4664 old_data = btrfs_item_end_nr(leaf, slot); 4665 4666 BUG_ON(slot < 0); 4667 if (slot >= nritems) { 4668 btrfs_print_leaf(root, leaf); 4669 btrfs_crit(root->fs_info, "slot %d too large, nritems %d", 4670 slot, nritems); 4671 BUG_ON(1); 4672 } 4673 4674 /* 4675 * item0..itemN ... dataN.offset..dataN.size .. data0.size 4676 */ 4677 /* first correct the data pointers */ 4678 for (i = slot; i < nritems; i++) { 4679 u32 ioff; 4680 item = btrfs_item_nr(i); 4681 4682 ioff = btrfs_token_item_offset(leaf, item, &token); 4683 btrfs_set_token_item_offset(leaf, item, 4684 ioff - data_size, &token); 4685 } 4686 4687 /* shift the data */ 4688 memmove_extent_buffer(leaf, btrfs_leaf_data(leaf) + 4689 data_end - data_size, btrfs_leaf_data(leaf) + 4690 data_end, old_data - data_end); 4691 4692 data_end = old_data; 4693 old_size = btrfs_item_size_nr(leaf, slot); 4694 item = btrfs_item_nr(slot); 4695 btrfs_set_item_size(leaf, item, old_size + data_size); 4696 btrfs_mark_buffer_dirty(leaf); 4697 4698 if (btrfs_leaf_free_space(root, leaf) < 0) { 4699 btrfs_print_leaf(root, leaf); 4700 BUG(); 4701 } 4702 } 4703 4704 /* 4705 * this is a helper for btrfs_insert_empty_items, the main goal here is 4706 * to save stack depth by doing the bulk of the work in a function 4707 * that doesn't call btrfs_search_slot 4708 */ 4709 void setup_items_for_insert(struct btrfs_root *root, struct btrfs_path *path, 4710 struct btrfs_key *cpu_key, u32 *data_size, 4711 u32 total_data, u32 total_size, int nr) 4712 { 4713 struct btrfs_item *item; 4714 int i; 4715 u32 nritems; 4716 unsigned int data_end; 4717 struct btrfs_disk_key disk_key; 4718 struct extent_buffer *leaf; 4719 int slot; 4720 struct btrfs_map_token token; 4721 4722 if (path->slots[0] == 0) { 4723 btrfs_cpu_key_to_disk(&disk_key, cpu_key); 4724 fixup_low_keys(root->fs_info, path, &disk_key, 1); 4725 } 4726 btrfs_unlock_up_safe(path, 1); 4727 4728 btrfs_init_map_token(&token); 4729 4730 leaf = path->nodes[0]; 4731 slot = path->slots[0]; 4732 4733 nritems = btrfs_header_nritems(leaf); 4734 data_end = leaf_data_end(root, leaf); 4735 4736 if (btrfs_leaf_free_space(root, leaf) < total_size) { 4737 btrfs_print_leaf(root, leaf); 4738 btrfs_crit(root->fs_info, 4739 "not enough freespace need %u have %d", 4740 total_size, btrfs_leaf_free_space(root, leaf)); 4741 BUG(); 4742 } 4743 4744 if (slot != nritems) { 4745 unsigned int old_data = btrfs_item_end_nr(leaf, slot); 4746 4747 if (old_data < data_end) { 4748 btrfs_print_leaf(root, leaf); 4749 btrfs_crit(root->fs_info, 4750 "slot %d old_data %d data_end %d", 4751 slot, old_data, data_end); 4752 BUG_ON(1); 4753 } 4754 /* 4755 * item0..itemN ... dataN.offset..dataN.size .. data0.size 4756 */ 4757 /* first correct the data pointers */ 4758 for (i = slot; i < nritems; i++) { 4759 u32 ioff; 4760 4761 item = btrfs_item_nr(i); 4762 ioff = btrfs_token_item_offset(leaf, item, &token); 4763 btrfs_set_token_item_offset(leaf, item, 4764 ioff - total_data, &token); 4765 } 4766 /* shift the items */ 4767 memmove_extent_buffer(leaf, btrfs_item_nr_offset(slot + nr), 4768 btrfs_item_nr_offset(slot), 4769 (nritems - slot) * sizeof(struct btrfs_item)); 4770 4771 /* shift the data */ 4772 memmove_extent_buffer(leaf, btrfs_leaf_data(leaf) + 4773 data_end - total_data, btrfs_leaf_data(leaf) + 4774 data_end, old_data - data_end); 4775 data_end = old_data; 4776 } 4777 4778 /* setup the item for the new data */ 4779 for (i = 0; i < nr; i++) { 4780 btrfs_cpu_key_to_disk(&disk_key, cpu_key + i); 4781 btrfs_set_item_key(leaf, &disk_key, slot + i); 4782 item = btrfs_item_nr(slot + i); 4783 btrfs_set_token_item_offset(leaf, item, 4784 data_end - data_size[i], &token); 4785 data_end -= data_size[i]; 4786 btrfs_set_token_item_size(leaf, item, data_size[i], &token); 4787 } 4788 4789 btrfs_set_header_nritems(leaf, nritems + nr); 4790 btrfs_mark_buffer_dirty(leaf); 4791 4792 if (btrfs_leaf_free_space(root, leaf) < 0) { 4793 btrfs_print_leaf(root, leaf); 4794 BUG(); 4795 } 4796 } 4797 4798 /* 4799 * Given a key and some data, insert items into the tree. 4800 * This does all the path init required, making room in the tree if needed. 4801 */ 4802 int btrfs_insert_empty_items(struct btrfs_trans_handle *trans, 4803 struct btrfs_root *root, 4804 struct btrfs_path *path, 4805 struct btrfs_key *cpu_key, u32 *data_size, 4806 int nr) 4807 { 4808 int ret = 0; 4809 int slot; 4810 int i; 4811 u32 total_size = 0; 4812 u32 total_data = 0; 4813 4814 for (i = 0; i < nr; i++) 4815 total_data += data_size[i]; 4816 4817 total_size = total_data + (nr * sizeof(struct btrfs_item)); 4818 ret = btrfs_search_slot(trans, root, cpu_key, path, total_size, 1); 4819 if (ret == 0) 4820 return -EEXIST; 4821 if (ret < 0) 4822 return ret; 4823 4824 slot = path->slots[0]; 4825 BUG_ON(slot < 0); 4826 4827 setup_items_for_insert(root, path, cpu_key, data_size, 4828 total_data, total_size, nr); 4829 return 0; 4830 } 4831 4832 /* 4833 * Given a key and some data, insert an item into the tree. 4834 * This does all the path init required, making room in the tree if needed. 4835 */ 4836 int btrfs_insert_item(struct btrfs_trans_handle *trans, struct btrfs_root 4837 *root, struct btrfs_key *cpu_key, void *data, u32 4838 data_size) 4839 { 4840 int ret = 0; 4841 struct btrfs_path *path; 4842 struct extent_buffer *leaf; 4843 unsigned long ptr; 4844 4845 path = btrfs_alloc_path(); 4846 if (!path) 4847 return -ENOMEM; 4848 ret = btrfs_insert_empty_item(trans, root, path, cpu_key, data_size); 4849 if (!ret) { 4850 leaf = path->nodes[0]; 4851 ptr = btrfs_item_ptr_offset(leaf, path->slots[0]); 4852 write_extent_buffer(leaf, data, ptr, data_size); 4853 btrfs_mark_buffer_dirty(leaf); 4854 } 4855 btrfs_free_path(path); 4856 return ret; 4857 } 4858 4859 /* 4860 * delete the pointer from a given node. 4861 * 4862 * the tree should have been previously balanced so the deletion does not 4863 * empty a node. 4864 */ 4865 static void del_ptr(struct btrfs_root *root, struct btrfs_path *path, 4866 int level, int slot) 4867 { 4868 struct extent_buffer *parent = path->nodes[level]; 4869 u32 nritems; 4870 int ret; 4871 4872 nritems = btrfs_header_nritems(parent); 4873 if (slot != nritems - 1) { 4874 if (level) 4875 tree_mod_log_eb_move(root->fs_info, parent, slot, 4876 slot + 1, nritems - slot - 1); 4877 memmove_extent_buffer(parent, 4878 btrfs_node_key_ptr_offset(slot), 4879 btrfs_node_key_ptr_offset(slot + 1), 4880 sizeof(struct btrfs_key_ptr) * 4881 (nritems - slot - 1)); 4882 } else if (level) { 4883 ret = tree_mod_log_insert_key(root->fs_info, parent, slot, 4884 MOD_LOG_KEY_REMOVE, GFP_NOFS); 4885 BUG_ON(ret < 0); 4886 } 4887 4888 nritems--; 4889 btrfs_set_header_nritems(parent, nritems); 4890 if (nritems == 0 && parent == root->node) { 4891 BUG_ON(btrfs_header_level(root->node) != 1); 4892 /* just turn the root into a leaf and break */ 4893 btrfs_set_header_level(root->node, 0); 4894 } else if (slot == 0) { 4895 struct btrfs_disk_key disk_key; 4896 4897 btrfs_node_key(parent, &disk_key, 0); 4898 fixup_low_keys(root->fs_info, path, &disk_key, level + 1); 4899 } 4900 btrfs_mark_buffer_dirty(parent); 4901 } 4902 4903 /* 4904 * a helper function to delete the leaf pointed to by path->slots[1] and 4905 * path->nodes[1]. 4906 * 4907 * This deletes the pointer in path->nodes[1] and frees the leaf 4908 * block extent. zero is returned if it all worked out, < 0 otherwise. 4909 * 4910 * The path must have already been setup for deleting the leaf, including 4911 * all the proper balancing. path->nodes[1] must be locked. 4912 */ 4913 static noinline void btrfs_del_leaf(struct btrfs_trans_handle *trans, 4914 struct btrfs_root *root, 4915 struct btrfs_path *path, 4916 struct extent_buffer *leaf) 4917 { 4918 WARN_ON(btrfs_header_generation(leaf) != trans->transid); 4919 del_ptr(root, path, 1, path->slots[1]); 4920 4921 /* 4922 * btrfs_free_extent is expensive, we want to make sure we 4923 * aren't holding any locks when we call it 4924 */ 4925 btrfs_unlock_up_safe(path, 0); 4926 4927 root_sub_used(root, leaf->len); 4928 4929 extent_buffer_get(leaf); 4930 btrfs_free_tree_block(trans, root, leaf, 0, 1); 4931 free_extent_buffer_stale(leaf); 4932 } 4933 /* 4934 * delete the item at the leaf level in path. If that empties 4935 * the leaf, remove it from the tree 4936 */ 4937 int btrfs_del_items(struct btrfs_trans_handle *trans, struct btrfs_root *root, 4938 struct btrfs_path *path, int slot, int nr) 4939 { 4940 struct extent_buffer *leaf; 4941 struct btrfs_item *item; 4942 u32 last_off; 4943 u32 dsize = 0; 4944 int ret = 0; 4945 int wret; 4946 int i; 4947 u32 nritems; 4948 struct btrfs_map_token token; 4949 4950 btrfs_init_map_token(&token); 4951 4952 leaf = path->nodes[0]; 4953 last_off = btrfs_item_offset_nr(leaf, slot + nr - 1); 4954 4955 for (i = 0; i < nr; i++) 4956 dsize += btrfs_item_size_nr(leaf, slot + i); 4957 4958 nritems = btrfs_header_nritems(leaf); 4959 4960 if (slot + nr != nritems) { 4961 int data_end = leaf_data_end(root, leaf); 4962 4963 memmove_extent_buffer(leaf, btrfs_leaf_data(leaf) + 4964 data_end + dsize, 4965 btrfs_leaf_data(leaf) + data_end, 4966 last_off - data_end); 4967 4968 for (i = slot + nr; i < nritems; i++) { 4969 u32 ioff; 4970 4971 item = btrfs_item_nr(i); 4972 ioff = btrfs_token_item_offset(leaf, item, &token); 4973 btrfs_set_token_item_offset(leaf, item, 4974 ioff + dsize, &token); 4975 } 4976 4977 memmove_extent_buffer(leaf, btrfs_item_nr_offset(slot), 4978 btrfs_item_nr_offset(slot + nr), 4979 sizeof(struct btrfs_item) * 4980 (nritems - slot - nr)); 4981 } 4982 btrfs_set_header_nritems(leaf, nritems - nr); 4983 nritems -= nr; 4984 4985 /* delete the leaf if we've emptied it */ 4986 if (nritems == 0) { 4987 if (leaf == root->node) { 4988 btrfs_set_header_level(leaf, 0); 4989 } else { 4990 btrfs_set_path_blocking(path); 4991 clean_tree_block(trans, root->fs_info, leaf); 4992 btrfs_del_leaf(trans, root, path, leaf); 4993 } 4994 } else { 4995 int used = leaf_space_used(leaf, 0, nritems); 4996 if (slot == 0) { 4997 struct btrfs_disk_key disk_key; 4998 4999 btrfs_item_key(leaf, &disk_key, 0); 5000 fixup_low_keys(root->fs_info, path, &disk_key, 1); 5001 } 5002 5003 /* delete the leaf if it is mostly empty */ 5004 if (used < BTRFS_LEAF_DATA_SIZE(root) / 3) { 5005 /* push_leaf_left fixes the path. 5006 * make sure the path still points to our leaf 5007 * for possible call to del_ptr below 5008 */ 5009 slot = path->slots[1]; 5010 extent_buffer_get(leaf); 5011 5012 btrfs_set_path_blocking(path); 5013 wret = push_leaf_left(trans, root, path, 1, 1, 5014 1, (u32)-1); 5015 if (wret < 0 && wret != -ENOSPC) 5016 ret = wret; 5017 5018 if (path->nodes[0] == leaf && 5019 btrfs_header_nritems(leaf)) { 5020 wret = push_leaf_right(trans, root, path, 1, 5021 1, 1, 0); 5022 if (wret < 0 && wret != -ENOSPC) 5023 ret = wret; 5024 } 5025 5026 if (btrfs_header_nritems(leaf) == 0) { 5027 path->slots[1] = slot; 5028 btrfs_del_leaf(trans, root, path, leaf); 5029 free_extent_buffer(leaf); 5030 ret = 0; 5031 } else { 5032 /* if we're still in the path, make sure 5033 * we're dirty. Otherwise, one of the 5034 * push_leaf functions must have already 5035 * dirtied this buffer 5036 */ 5037 if (path->nodes[0] == leaf) 5038 btrfs_mark_buffer_dirty(leaf); 5039 free_extent_buffer(leaf); 5040 } 5041 } else { 5042 btrfs_mark_buffer_dirty(leaf); 5043 } 5044 } 5045 return ret; 5046 } 5047 5048 /* 5049 * search the tree again to find a leaf with lesser keys 5050 * returns 0 if it found something or 1 if there are no lesser leaves. 5051 * returns < 0 on io errors. 5052 * 5053 * This may release the path, and so you may lose any locks held at the 5054 * time you call it. 5055 */ 5056 int btrfs_prev_leaf(struct btrfs_root *root, struct btrfs_path *path) 5057 { 5058 struct btrfs_key key; 5059 struct btrfs_disk_key found_key; 5060 int ret; 5061 5062 btrfs_item_key_to_cpu(path->nodes[0], &key, 0); 5063 5064 if (key.offset > 0) { 5065 key.offset--; 5066 } else if (key.type > 0) { 5067 key.type--; 5068 key.offset = (u64)-1; 5069 } else if (key.objectid > 0) { 5070 key.objectid--; 5071 key.type = (u8)-1; 5072 key.offset = (u64)-1; 5073 } else { 5074 return 1; 5075 } 5076 5077 btrfs_release_path(path); 5078 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); 5079 if (ret < 0) 5080 return ret; 5081 btrfs_item_key(path->nodes[0], &found_key, 0); 5082 ret = comp_keys(&found_key, &key); 5083 /* 5084 * We might have had an item with the previous key in the tree right 5085 * before we released our path. And after we released our path, that 5086 * item might have been pushed to the first slot (0) of the leaf we 5087 * were holding due to a tree balance. Alternatively, an item with the 5088 * previous key can exist as the only element of a leaf (big fat item). 5089 * Therefore account for these 2 cases, so that our callers (like 5090 * btrfs_previous_item) don't miss an existing item with a key matching 5091 * the previous key we computed above. 5092 */ 5093 if (ret <= 0) 5094 return 0; 5095 return 1; 5096 } 5097 5098 /* 5099 * A helper function to walk down the tree starting at min_key, and looking 5100 * for nodes or leaves that are have a minimum transaction id. 5101 * This is used by the btree defrag code, and tree logging 5102 * 5103 * This does not cow, but it does stuff the starting key it finds back 5104 * into min_key, so you can call btrfs_search_slot with cow=1 on the 5105 * key and get a writable path. 5106 * 5107 * This does lock as it descends, and path->keep_locks should be set 5108 * to 1 by the caller. 5109 * 5110 * This honors path->lowest_level to prevent descent past a given level 5111 * of the tree. 5112 * 5113 * min_trans indicates the oldest transaction that you are interested 5114 * in walking through. Any nodes or leaves older than min_trans are 5115 * skipped over (without reading them). 5116 * 5117 * returns zero if something useful was found, < 0 on error and 1 if there 5118 * was nothing in the tree that matched the search criteria. 5119 */ 5120 int btrfs_search_forward(struct btrfs_root *root, struct btrfs_key *min_key, 5121 struct btrfs_path *path, 5122 u64 min_trans) 5123 { 5124 struct extent_buffer *cur; 5125 struct btrfs_key found_key; 5126 int slot; 5127 int sret; 5128 u32 nritems; 5129 int level; 5130 int ret = 1; 5131 int keep_locks = path->keep_locks; 5132 5133 path->keep_locks = 1; 5134 again: 5135 cur = btrfs_read_lock_root_node(root); 5136 level = btrfs_header_level(cur); 5137 WARN_ON(path->nodes[level]); 5138 path->nodes[level] = cur; 5139 path->locks[level] = BTRFS_READ_LOCK; 5140 5141 if (btrfs_header_generation(cur) < min_trans) { 5142 ret = 1; 5143 goto out; 5144 } 5145 while (1) { 5146 nritems = btrfs_header_nritems(cur); 5147 level = btrfs_header_level(cur); 5148 sret = bin_search(cur, min_key, level, &slot); 5149 5150 /* at the lowest level, we're done, setup the path and exit */ 5151 if (level == path->lowest_level) { 5152 if (slot >= nritems) 5153 goto find_next_key; 5154 ret = 0; 5155 path->slots[level] = slot; 5156 btrfs_item_key_to_cpu(cur, &found_key, slot); 5157 goto out; 5158 } 5159 if (sret && slot > 0) 5160 slot--; 5161 /* 5162 * check this node pointer against the min_trans parameters. 5163 * If it is too old, old, skip to the next one. 5164 */ 5165 while (slot < nritems) { 5166 u64 gen; 5167 5168 gen = btrfs_node_ptr_generation(cur, slot); 5169 if (gen < min_trans) { 5170 slot++; 5171 continue; 5172 } 5173 break; 5174 } 5175 find_next_key: 5176 /* 5177 * we didn't find a candidate key in this node, walk forward 5178 * and find another one 5179 */ 5180 if (slot >= nritems) { 5181 path->slots[level] = slot; 5182 btrfs_set_path_blocking(path); 5183 sret = btrfs_find_next_key(root, path, min_key, level, 5184 min_trans); 5185 if (sret == 0) { 5186 btrfs_release_path(path); 5187 goto again; 5188 } else { 5189 goto out; 5190 } 5191 } 5192 /* save our key for returning back */ 5193 btrfs_node_key_to_cpu(cur, &found_key, slot); 5194 path->slots[level] = slot; 5195 if (level == path->lowest_level) { 5196 ret = 0; 5197 goto out; 5198 } 5199 btrfs_set_path_blocking(path); 5200 cur = read_node_slot(root, cur, slot); 5201 if (IS_ERR(cur)) { 5202 ret = PTR_ERR(cur); 5203 goto out; 5204 } 5205 5206 btrfs_tree_read_lock(cur); 5207 5208 path->locks[level - 1] = BTRFS_READ_LOCK; 5209 path->nodes[level - 1] = cur; 5210 unlock_up(path, level, 1, 0, NULL); 5211 btrfs_clear_path_blocking(path, NULL, 0); 5212 } 5213 out: 5214 path->keep_locks = keep_locks; 5215 if (ret == 0) { 5216 btrfs_unlock_up_safe(path, path->lowest_level + 1); 5217 btrfs_set_path_blocking(path); 5218 memcpy(min_key, &found_key, sizeof(found_key)); 5219 } 5220 return ret; 5221 } 5222 5223 static int tree_move_down(struct btrfs_root *root, 5224 struct btrfs_path *path, 5225 int *level, int root_level) 5226 { 5227 struct extent_buffer *eb; 5228 5229 BUG_ON(*level == 0); 5230 eb = read_node_slot(root, path->nodes[*level], path->slots[*level]); 5231 if (IS_ERR(eb)) 5232 return PTR_ERR(eb); 5233 5234 path->nodes[*level - 1] = eb; 5235 path->slots[*level - 1] = 0; 5236 (*level)--; 5237 return 0; 5238 } 5239 5240 static int tree_move_next_or_upnext(struct btrfs_root *root, 5241 struct btrfs_path *path, 5242 int *level, int root_level) 5243 { 5244 int ret = 0; 5245 int nritems; 5246 nritems = btrfs_header_nritems(path->nodes[*level]); 5247 5248 path->slots[*level]++; 5249 5250 while (path->slots[*level] >= nritems) { 5251 if (*level == root_level) 5252 return -1; 5253 5254 /* move upnext */ 5255 path->slots[*level] = 0; 5256 free_extent_buffer(path->nodes[*level]); 5257 path->nodes[*level] = NULL; 5258 (*level)++; 5259 path->slots[*level]++; 5260 5261 nritems = btrfs_header_nritems(path->nodes[*level]); 5262 ret = 1; 5263 } 5264 return ret; 5265 } 5266 5267 /* 5268 * Returns 1 if it had to move up and next. 0 is returned if it moved only next 5269 * or down. 5270 */ 5271 static int tree_advance(struct btrfs_root *root, 5272 struct btrfs_path *path, 5273 int *level, int root_level, 5274 int allow_down, 5275 struct btrfs_key *key) 5276 { 5277 int ret; 5278 5279 if (*level == 0 || !allow_down) { 5280 ret = tree_move_next_or_upnext(root, path, level, root_level); 5281 } else { 5282 ret = tree_move_down(root, path, level, root_level); 5283 } 5284 if (ret >= 0) { 5285 if (*level == 0) 5286 btrfs_item_key_to_cpu(path->nodes[*level], key, 5287 path->slots[*level]); 5288 else 5289 btrfs_node_key_to_cpu(path->nodes[*level], key, 5290 path->slots[*level]); 5291 } 5292 return ret; 5293 } 5294 5295 static int tree_compare_item(struct btrfs_root *left_root, 5296 struct btrfs_path *left_path, 5297 struct btrfs_path *right_path, 5298 char *tmp_buf) 5299 { 5300 int cmp; 5301 int len1, len2; 5302 unsigned long off1, off2; 5303 5304 len1 = btrfs_item_size_nr(left_path->nodes[0], left_path->slots[0]); 5305 len2 = btrfs_item_size_nr(right_path->nodes[0], right_path->slots[0]); 5306 if (len1 != len2) 5307 return 1; 5308 5309 off1 = btrfs_item_ptr_offset(left_path->nodes[0], left_path->slots[0]); 5310 off2 = btrfs_item_ptr_offset(right_path->nodes[0], 5311 right_path->slots[0]); 5312 5313 read_extent_buffer(left_path->nodes[0], tmp_buf, off1, len1); 5314 5315 cmp = memcmp_extent_buffer(right_path->nodes[0], tmp_buf, off2, len1); 5316 if (cmp) 5317 return 1; 5318 return 0; 5319 } 5320 5321 #define ADVANCE 1 5322 #define ADVANCE_ONLY_NEXT -1 5323 5324 /* 5325 * This function compares two trees and calls the provided callback for 5326 * every changed/new/deleted item it finds. 5327 * If shared tree blocks are encountered, whole subtrees are skipped, making 5328 * the compare pretty fast on snapshotted subvolumes. 5329 * 5330 * This currently works on commit roots only. As commit roots are read only, 5331 * we don't do any locking. The commit roots are protected with transactions. 5332 * Transactions are ended and rejoined when a commit is tried in between. 5333 * 5334 * This function checks for modifications done to the trees while comparing. 5335 * If it detects a change, it aborts immediately. 5336 */ 5337 int btrfs_compare_trees(struct btrfs_root *left_root, 5338 struct btrfs_root *right_root, 5339 btrfs_changed_cb_t changed_cb, void *ctx) 5340 { 5341 int ret; 5342 int cmp; 5343 struct btrfs_path *left_path = NULL; 5344 struct btrfs_path *right_path = NULL; 5345 struct btrfs_key left_key; 5346 struct btrfs_key right_key; 5347 char *tmp_buf = NULL; 5348 int left_root_level; 5349 int right_root_level; 5350 int left_level; 5351 int right_level; 5352 int left_end_reached; 5353 int right_end_reached; 5354 int advance_left; 5355 int advance_right; 5356 u64 left_blockptr; 5357 u64 right_blockptr; 5358 u64 left_gen; 5359 u64 right_gen; 5360 5361 left_path = btrfs_alloc_path(); 5362 if (!left_path) { 5363 ret = -ENOMEM; 5364 goto out; 5365 } 5366 right_path = btrfs_alloc_path(); 5367 if (!right_path) { 5368 ret = -ENOMEM; 5369 goto out; 5370 } 5371 5372 tmp_buf = kmalloc(left_root->nodesize, GFP_KERNEL | __GFP_NOWARN); 5373 if (!tmp_buf) { 5374 tmp_buf = vmalloc(left_root->nodesize); 5375 if (!tmp_buf) { 5376 ret = -ENOMEM; 5377 goto out; 5378 } 5379 } 5380 5381 left_path->search_commit_root = 1; 5382 left_path->skip_locking = 1; 5383 right_path->search_commit_root = 1; 5384 right_path->skip_locking = 1; 5385 5386 /* 5387 * Strategy: Go to the first items of both trees. Then do 5388 * 5389 * If both trees are at level 0 5390 * Compare keys of current items 5391 * If left < right treat left item as new, advance left tree 5392 * and repeat 5393 * If left > right treat right item as deleted, advance right tree 5394 * and repeat 5395 * If left == right do deep compare of items, treat as changed if 5396 * needed, advance both trees and repeat 5397 * If both trees are at the same level but not at level 0 5398 * Compare keys of current nodes/leafs 5399 * If left < right advance left tree and repeat 5400 * If left > right advance right tree and repeat 5401 * If left == right compare blockptrs of the next nodes/leafs 5402 * If they match advance both trees but stay at the same level 5403 * and repeat 5404 * If they don't match advance both trees while allowing to go 5405 * deeper and repeat 5406 * If tree levels are different 5407 * Advance the tree that needs it and repeat 5408 * 5409 * Advancing a tree means: 5410 * If we are at level 0, try to go to the next slot. If that's not 5411 * possible, go one level up and repeat. Stop when we found a level 5412 * where we could go to the next slot. We may at this point be on a 5413 * node or a leaf. 5414 * 5415 * If we are not at level 0 and not on shared tree blocks, go one 5416 * level deeper. 5417 * 5418 * If we are not at level 0 and on shared tree blocks, go one slot to 5419 * the right if possible or go up and right. 5420 */ 5421 5422 down_read(&left_root->fs_info->commit_root_sem); 5423 left_level = btrfs_header_level(left_root->commit_root); 5424 left_root_level = left_level; 5425 left_path->nodes[left_level] = left_root->commit_root; 5426 extent_buffer_get(left_path->nodes[left_level]); 5427 5428 right_level = btrfs_header_level(right_root->commit_root); 5429 right_root_level = right_level; 5430 right_path->nodes[right_level] = right_root->commit_root; 5431 extent_buffer_get(right_path->nodes[right_level]); 5432 up_read(&left_root->fs_info->commit_root_sem); 5433 5434 if (left_level == 0) 5435 btrfs_item_key_to_cpu(left_path->nodes[left_level], 5436 &left_key, left_path->slots[left_level]); 5437 else 5438 btrfs_node_key_to_cpu(left_path->nodes[left_level], 5439 &left_key, left_path->slots[left_level]); 5440 if (right_level == 0) 5441 btrfs_item_key_to_cpu(right_path->nodes[right_level], 5442 &right_key, right_path->slots[right_level]); 5443 else 5444 btrfs_node_key_to_cpu(right_path->nodes[right_level], 5445 &right_key, right_path->slots[right_level]); 5446 5447 left_end_reached = right_end_reached = 0; 5448 advance_left = advance_right = 0; 5449 5450 while (1) { 5451 if (advance_left && !left_end_reached) { 5452 ret = tree_advance(left_root, left_path, &left_level, 5453 left_root_level, 5454 advance_left != ADVANCE_ONLY_NEXT, 5455 &left_key); 5456 if (ret == -1) 5457 left_end_reached = ADVANCE; 5458 else if (ret < 0) 5459 goto out; 5460 advance_left = 0; 5461 } 5462 if (advance_right && !right_end_reached) { 5463 ret = tree_advance(right_root, right_path, &right_level, 5464 right_root_level, 5465 advance_right != ADVANCE_ONLY_NEXT, 5466 &right_key); 5467 if (ret == -1) 5468 right_end_reached = ADVANCE; 5469 else if (ret < 0) 5470 goto out; 5471 advance_right = 0; 5472 } 5473 5474 if (left_end_reached && right_end_reached) { 5475 ret = 0; 5476 goto out; 5477 } else if (left_end_reached) { 5478 if (right_level == 0) { 5479 ret = changed_cb(left_root, right_root, 5480 left_path, right_path, 5481 &right_key, 5482 BTRFS_COMPARE_TREE_DELETED, 5483 ctx); 5484 if (ret < 0) 5485 goto out; 5486 } 5487 advance_right = ADVANCE; 5488 continue; 5489 } else if (right_end_reached) { 5490 if (left_level == 0) { 5491 ret = changed_cb(left_root, right_root, 5492 left_path, right_path, 5493 &left_key, 5494 BTRFS_COMPARE_TREE_NEW, 5495 ctx); 5496 if (ret < 0) 5497 goto out; 5498 } 5499 advance_left = ADVANCE; 5500 continue; 5501 } 5502 5503 if (left_level == 0 && right_level == 0) { 5504 cmp = btrfs_comp_cpu_keys(&left_key, &right_key); 5505 if (cmp < 0) { 5506 ret = changed_cb(left_root, right_root, 5507 left_path, right_path, 5508 &left_key, 5509 BTRFS_COMPARE_TREE_NEW, 5510 ctx); 5511 if (ret < 0) 5512 goto out; 5513 advance_left = ADVANCE; 5514 } else if (cmp > 0) { 5515 ret = changed_cb(left_root, right_root, 5516 left_path, right_path, 5517 &right_key, 5518 BTRFS_COMPARE_TREE_DELETED, 5519 ctx); 5520 if (ret < 0) 5521 goto out; 5522 advance_right = ADVANCE; 5523 } else { 5524 enum btrfs_compare_tree_result result; 5525 5526 WARN_ON(!extent_buffer_uptodate(left_path->nodes[0])); 5527 ret = tree_compare_item(left_root, left_path, 5528 right_path, tmp_buf); 5529 if (ret) 5530 result = BTRFS_COMPARE_TREE_CHANGED; 5531 else 5532 result = BTRFS_COMPARE_TREE_SAME; 5533 ret = changed_cb(left_root, right_root, 5534 left_path, right_path, 5535 &left_key, result, ctx); 5536 if (ret < 0) 5537 goto out; 5538 advance_left = ADVANCE; 5539 advance_right = ADVANCE; 5540 } 5541 } else if (left_level == right_level) { 5542 cmp = btrfs_comp_cpu_keys(&left_key, &right_key); 5543 if (cmp < 0) { 5544 advance_left = ADVANCE; 5545 } else if (cmp > 0) { 5546 advance_right = ADVANCE; 5547 } else { 5548 left_blockptr = btrfs_node_blockptr( 5549 left_path->nodes[left_level], 5550 left_path->slots[left_level]); 5551 right_blockptr = btrfs_node_blockptr( 5552 right_path->nodes[right_level], 5553 right_path->slots[right_level]); 5554 left_gen = btrfs_node_ptr_generation( 5555 left_path->nodes[left_level], 5556 left_path->slots[left_level]); 5557 right_gen = btrfs_node_ptr_generation( 5558 right_path->nodes[right_level], 5559 right_path->slots[right_level]); 5560 if (left_blockptr == right_blockptr && 5561 left_gen == right_gen) { 5562 /* 5563 * As we're on a shared block, don't 5564 * allow to go deeper. 5565 */ 5566 advance_left = ADVANCE_ONLY_NEXT; 5567 advance_right = ADVANCE_ONLY_NEXT; 5568 } else { 5569 advance_left = ADVANCE; 5570 advance_right = ADVANCE; 5571 } 5572 } 5573 } else if (left_level < right_level) { 5574 advance_right = ADVANCE; 5575 } else { 5576 advance_left = ADVANCE; 5577 } 5578 } 5579 5580 out: 5581 btrfs_free_path(left_path); 5582 btrfs_free_path(right_path); 5583 kvfree(tmp_buf); 5584 return ret; 5585 } 5586 5587 /* 5588 * this is similar to btrfs_next_leaf, but does not try to preserve 5589 * and fixup the path. It looks for and returns the next key in the 5590 * tree based on the current path and the min_trans parameters. 5591 * 5592 * 0 is returned if another key is found, < 0 if there are any errors 5593 * and 1 is returned if there are no higher keys in the tree 5594 * 5595 * path->keep_locks should be set to 1 on the search made before 5596 * calling this function. 5597 */ 5598 int btrfs_find_next_key(struct btrfs_root *root, struct btrfs_path *path, 5599 struct btrfs_key *key, int level, u64 min_trans) 5600 { 5601 int slot; 5602 struct extent_buffer *c; 5603 5604 WARN_ON(!path->keep_locks); 5605 while (level < BTRFS_MAX_LEVEL) { 5606 if (!path->nodes[level]) 5607 return 1; 5608 5609 slot = path->slots[level] + 1; 5610 c = path->nodes[level]; 5611 next: 5612 if (slot >= btrfs_header_nritems(c)) { 5613 int ret; 5614 int orig_lowest; 5615 struct btrfs_key cur_key; 5616 if (level + 1 >= BTRFS_MAX_LEVEL || 5617 !path->nodes[level + 1]) 5618 return 1; 5619 5620 if (path->locks[level + 1]) { 5621 level++; 5622 continue; 5623 } 5624 5625 slot = btrfs_header_nritems(c) - 1; 5626 if (level == 0) 5627 btrfs_item_key_to_cpu(c, &cur_key, slot); 5628 else 5629 btrfs_node_key_to_cpu(c, &cur_key, slot); 5630 5631 orig_lowest = path->lowest_level; 5632 btrfs_release_path(path); 5633 path->lowest_level = level; 5634 ret = btrfs_search_slot(NULL, root, &cur_key, path, 5635 0, 0); 5636 path->lowest_level = orig_lowest; 5637 if (ret < 0) 5638 return ret; 5639 5640 c = path->nodes[level]; 5641 slot = path->slots[level]; 5642 if (ret == 0) 5643 slot++; 5644 goto next; 5645 } 5646 5647 if (level == 0) 5648 btrfs_item_key_to_cpu(c, key, slot); 5649 else { 5650 u64 gen = btrfs_node_ptr_generation(c, slot); 5651 5652 if (gen < min_trans) { 5653 slot++; 5654 goto next; 5655 } 5656 btrfs_node_key_to_cpu(c, key, slot); 5657 } 5658 return 0; 5659 } 5660 return 1; 5661 } 5662 5663 /* 5664 * search the tree again to find a leaf with greater keys 5665 * returns 0 if it found something or 1 if there are no greater leaves. 5666 * returns < 0 on io errors. 5667 */ 5668 int btrfs_next_leaf(struct btrfs_root *root, struct btrfs_path *path) 5669 { 5670 return btrfs_next_old_leaf(root, path, 0); 5671 } 5672 5673 int btrfs_next_old_leaf(struct btrfs_root *root, struct btrfs_path *path, 5674 u64 time_seq) 5675 { 5676 int slot; 5677 int level; 5678 struct extent_buffer *c; 5679 struct extent_buffer *next; 5680 struct btrfs_key key; 5681 u32 nritems; 5682 int ret; 5683 int old_spinning = path->leave_spinning; 5684 int next_rw_lock = 0; 5685 5686 nritems = btrfs_header_nritems(path->nodes[0]); 5687 if (nritems == 0) 5688 return 1; 5689 5690 btrfs_item_key_to_cpu(path->nodes[0], &key, nritems - 1); 5691 again: 5692 level = 1; 5693 next = NULL; 5694 next_rw_lock = 0; 5695 btrfs_release_path(path); 5696 5697 path->keep_locks = 1; 5698 path->leave_spinning = 1; 5699 5700 if (time_seq) 5701 ret = btrfs_search_old_slot(root, &key, path, time_seq); 5702 else 5703 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); 5704 path->keep_locks = 0; 5705 5706 if (ret < 0) 5707 return ret; 5708 5709 nritems = btrfs_header_nritems(path->nodes[0]); 5710 /* 5711 * by releasing the path above we dropped all our locks. A balance 5712 * could have added more items next to the key that used to be 5713 * at the very end of the block. So, check again here and 5714 * advance the path if there are now more items available. 5715 */ 5716 if (nritems > 0 && path->slots[0] < nritems - 1) { 5717 if (ret == 0) 5718 path->slots[0]++; 5719 ret = 0; 5720 goto done; 5721 } 5722 /* 5723 * So the above check misses one case: 5724 * - after releasing the path above, someone has removed the item that 5725 * used to be at the very end of the block, and balance between leafs 5726 * gets another one with bigger key.offset to replace it. 5727 * 5728 * This one should be returned as well, or we can get leaf corruption 5729 * later(esp. in __btrfs_drop_extents()). 5730 * 5731 * And a bit more explanation about this check, 5732 * with ret > 0, the key isn't found, the path points to the slot 5733 * where it should be inserted, so the path->slots[0] item must be the 5734 * bigger one. 5735 */ 5736 if (nritems > 0 && ret > 0 && path->slots[0] == nritems - 1) { 5737 ret = 0; 5738 goto done; 5739 } 5740 5741 while (level < BTRFS_MAX_LEVEL) { 5742 if (!path->nodes[level]) { 5743 ret = 1; 5744 goto done; 5745 } 5746 5747 slot = path->slots[level] + 1; 5748 c = path->nodes[level]; 5749 if (slot >= btrfs_header_nritems(c)) { 5750 level++; 5751 if (level == BTRFS_MAX_LEVEL) { 5752 ret = 1; 5753 goto done; 5754 } 5755 continue; 5756 } 5757 5758 if (next) { 5759 btrfs_tree_unlock_rw(next, next_rw_lock); 5760 free_extent_buffer(next); 5761 } 5762 5763 next = c; 5764 next_rw_lock = path->locks[level]; 5765 ret = read_block_for_search(NULL, root, path, &next, level, 5766 slot, &key, 0); 5767 if (ret == -EAGAIN) 5768 goto again; 5769 5770 if (ret < 0) { 5771 btrfs_release_path(path); 5772 goto done; 5773 } 5774 5775 if (!path->skip_locking) { 5776 ret = btrfs_try_tree_read_lock(next); 5777 if (!ret && time_seq) { 5778 /* 5779 * If we don't get the lock, we may be racing 5780 * with push_leaf_left, holding that lock while 5781 * itself waiting for the leaf we've currently 5782 * locked. To solve this situation, we give up 5783 * on our lock and cycle. 5784 */ 5785 free_extent_buffer(next); 5786 btrfs_release_path(path); 5787 cond_resched(); 5788 goto again; 5789 } 5790 if (!ret) { 5791 btrfs_set_path_blocking(path); 5792 btrfs_tree_read_lock(next); 5793 btrfs_clear_path_blocking(path, next, 5794 BTRFS_READ_LOCK); 5795 } 5796 next_rw_lock = BTRFS_READ_LOCK; 5797 } 5798 break; 5799 } 5800 path->slots[level] = slot; 5801 while (1) { 5802 level--; 5803 c = path->nodes[level]; 5804 if (path->locks[level]) 5805 btrfs_tree_unlock_rw(c, path->locks[level]); 5806 5807 free_extent_buffer(c); 5808 path->nodes[level] = next; 5809 path->slots[level] = 0; 5810 if (!path->skip_locking) 5811 path->locks[level] = next_rw_lock; 5812 if (!level) 5813 break; 5814 5815 ret = read_block_for_search(NULL, root, path, &next, level, 5816 0, &key, 0); 5817 if (ret == -EAGAIN) 5818 goto again; 5819 5820 if (ret < 0) { 5821 btrfs_release_path(path); 5822 goto done; 5823 } 5824 5825 if (!path->skip_locking) { 5826 ret = btrfs_try_tree_read_lock(next); 5827 if (!ret) { 5828 btrfs_set_path_blocking(path); 5829 btrfs_tree_read_lock(next); 5830 btrfs_clear_path_blocking(path, next, 5831 BTRFS_READ_LOCK); 5832 } 5833 next_rw_lock = BTRFS_READ_LOCK; 5834 } 5835 } 5836 ret = 0; 5837 done: 5838 unlock_up(path, 0, 1, 0, NULL); 5839 path->leave_spinning = old_spinning; 5840 if (!old_spinning) 5841 btrfs_set_path_blocking(path); 5842 5843 return ret; 5844 } 5845 5846 /* 5847 * this uses btrfs_prev_leaf to walk backwards in the tree, and keeps 5848 * searching until it gets past min_objectid or finds an item of 'type' 5849 * 5850 * returns 0 if something is found, 1 if nothing was found and < 0 on error 5851 */ 5852 int btrfs_previous_item(struct btrfs_root *root, 5853 struct btrfs_path *path, u64 min_objectid, 5854 int type) 5855 { 5856 struct btrfs_key found_key; 5857 struct extent_buffer *leaf; 5858 u32 nritems; 5859 int ret; 5860 5861 while (1) { 5862 if (path->slots[0] == 0) { 5863 btrfs_set_path_blocking(path); 5864 ret = btrfs_prev_leaf(root, path); 5865 if (ret != 0) 5866 return ret; 5867 } else { 5868 path->slots[0]--; 5869 } 5870 leaf = path->nodes[0]; 5871 nritems = btrfs_header_nritems(leaf); 5872 if (nritems == 0) 5873 return 1; 5874 if (path->slots[0] == nritems) 5875 path->slots[0]--; 5876 5877 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]); 5878 if (found_key.objectid < min_objectid) 5879 break; 5880 if (found_key.type == type) 5881 return 0; 5882 if (found_key.objectid == min_objectid && 5883 found_key.type < type) 5884 break; 5885 } 5886 return 1; 5887 } 5888 5889 /* 5890 * search in extent tree to find a previous Metadata/Data extent item with 5891 * min objecitd. 5892 * 5893 * returns 0 if something is found, 1 if nothing was found and < 0 on error 5894 */ 5895 int btrfs_previous_extent_item(struct btrfs_root *root, 5896 struct btrfs_path *path, u64 min_objectid) 5897 { 5898 struct btrfs_key found_key; 5899 struct extent_buffer *leaf; 5900 u32 nritems; 5901 int ret; 5902 5903 while (1) { 5904 if (path->slots[0] == 0) { 5905 btrfs_set_path_blocking(path); 5906 ret = btrfs_prev_leaf(root, path); 5907 if (ret != 0) 5908 return ret; 5909 } else { 5910 path->slots[0]--; 5911 } 5912 leaf = path->nodes[0]; 5913 nritems = btrfs_header_nritems(leaf); 5914 if (nritems == 0) 5915 return 1; 5916 if (path->slots[0] == nritems) 5917 path->slots[0]--; 5918 5919 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]); 5920 if (found_key.objectid < min_objectid) 5921 break; 5922 if (found_key.type == BTRFS_EXTENT_ITEM_KEY || 5923 found_key.type == BTRFS_METADATA_ITEM_KEY) 5924 return 0; 5925 if (found_key.objectid == min_objectid && 5926 found_key.type < BTRFS_EXTENT_ITEM_KEY) 5927 break; 5928 } 5929 return 1; 5930 } 5931