1 // SPDX-License-Identifier: GPL-2.0 2 /* 3 * Copyright (C) 2011 Fujitsu. All rights reserved. 4 * Written by Miao Xie <miaox@cn.fujitsu.com> 5 */ 6 7 #include <linux/slab.h> 8 #include <linux/iversion.h> 9 #include "delayed-inode.h" 10 #include "disk-io.h" 11 #include "transaction.h" 12 #include "ctree.h" 13 #include "qgroup.h" 14 15 #define BTRFS_DELAYED_WRITEBACK 512 16 #define BTRFS_DELAYED_BACKGROUND 128 17 #define BTRFS_DELAYED_BATCH 16 18 19 static struct kmem_cache *delayed_node_cache; 20 21 int __init btrfs_delayed_inode_init(void) 22 { 23 delayed_node_cache = kmem_cache_create("btrfs_delayed_node", 24 sizeof(struct btrfs_delayed_node), 25 0, 26 SLAB_MEM_SPREAD, 27 NULL); 28 if (!delayed_node_cache) 29 return -ENOMEM; 30 return 0; 31 } 32 33 void __cold btrfs_delayed_inode_exit(void) 34 { 35 kmem_cache_destroy(delayed_node_cache); 36 } 37 38 static inline void btrfs_init_delayed_node( 39 struct btrfs_delayed_node *delayed_node, 40 struct btrfs_root *root, u64 inode_id) 41 { 42 delayed_node->root = root; 43 delayed_node->inode_id = inode_id; 44 refcount_set(&delayed_node->refs, 0); 45 delayed_node->ins_root = RB_ROOT_CACHED; 46 delayed_node->del_root = RB_ROOT_CACHED; 47 mutex_init(&delayed_node->mutex); 48 INIT_LIST_HEAD(&delayed_node->n_list); 49 INIT_LIST_HEAD(&delayed_node->p_list); 50 } 51 52 static inline int btrfs_is_continuous_delayed_item( 53 struct btrfs_delayed_item *item1, 54 struct btrfs_delayed_item *item2) 55 { 56 if (item1->key.type == BTRFS_DIR_INDEX_KEY && 57 item1->key.objectid == item2->key.objectid && 58 item1->key.type == item2->key.type && 59 item1->key.offset + 1 == item2->key.offset) 60 return 1; 61 return 0; 62 } 63 64 static struct btrfs_delayed_node *btrfs_get_delayed_node( 65 struct btrfs_inode *btrfs_inode) 66 { 67 struct btrfs_root *root = btrfs_inode->root; 68 u64 ino = btrfs_ino(btrfs_inode); 69 struct btrfs_delayed_node *node; 70 71 node = READ_ONCE(btrfs_inode->delayed_node); 72 if (node) { 73 refcount_inc(&node->refs); 74 return node; 75 } 76 77 spin_lock(&root->inode_lock); 78 node = radix_tree_lookup(&root->delayed_nodes_tree, ino); 79 80 if (node) { 81 if (btrfs_inode->delayed_node) { 82 refcount_inc(&node->refs); /* can be accessed */ 83 BUG_ON(btrfs_inode->delayed_node != node); 84 spin_unlock(&root->inode_lock); 85 return node; 86 } 87 88 /* 89 * It's possible that we're racing into the middle of removing 90 * this node from the radix tree. In this case, the refcount 91 * was zero and it should never go back to one. Just return 92 * NULL like it was never in the radix at all; our release 93 * function is in the process of removing it. 94 * 95 * Some implementations of refcount_inc refuse to bump the 96 * refcount once it has hit zero. If we don't do this dance 97 * here, refcount_inc() may decide to just WARN_ONCE() instead 98 * of actually bumping the refcount. 99 * 100 * If this node is properly in the radix, we want to bump the 101 * refcount twice, once for the inode and once for this get 102 * operation. 103 */ 104 if (refcount_inc_not_zero(&node->refs)) { 105 refcount_inc(&node->refs); 106 btrfs_inode->delayed_node = node; 107 } else { 108 node = NULL; 109 } 110 111 spin_unlock(&root->inode_lock); 112 return node; 113 } 114 spin_unlock(&root->inode_lock); 115 116 return NULL; 117 } 118 119 /* Will return either the node or PTR_ERR(-ENOMEM) */ 120 static struct btrfs_delayed_node *btrfs_get_or_create_delayed_node( 121 struct btrfs_inode *btrfs_inode) 122 { 123 struct btrfs_delayed_node *node; 124 struct btrfs_root *root = btrfs_inode->root; 125 u64 ino = btrfs_ino(btrfs_inode); 126 int ret; 127 128 again: 129 node = btrfs_get_delayed_node(btrfs_inode); 130 if (node) 131 return node; 132 133 node = kmem_cache_zalloc(delayed_node_cache, GFP_NOFS); 134 if (!node) 135 return ERR_PTR(-ENOMEM); 136 btrfs_init_delayed_node(node, root, ino); 137 138 /* cached in the btrfs inode and can be accessed */ 139 refcount_set(&node->refs, 2); 140 141 ret = radix_tree_preload(GFP_NOFS); 142 if (ret) { 143 kmem_cache_free(delayed_node_cache, node); 144 return ERR_PTR(ret); 145 } 146 147 spin_lock(&root->inode_lock); 148 ret = radix_tree_insert(&root->delayed_nodes_tree, ino, node); 149 if (ret == -EEXIST) { 150 spin_unlock(&root->inode_lock); 151 kmem_cache_free(delayed_node_cache, node); 152 radix_tree_preload_end(); 153 goto again; 154 } 155 btrfs_inode->delayed_node = node; 156 spin_unlock(&root->inode_lock); 157 radix_tree_preload_end(); 158 159 return node; 160 } 161 162 /* 163 * Call it when holding delayed_node->mutex 164 * 165 * If mod = 1, add this node into the prepared list. 166 */ 167 static void btrfs_queue_delayed_node(struct btrfs_delayed_root *root, 168 struct btrfs_delayed_node *node, 169 int mod) 170 { 171 spin_lock(&root->lock); 172 if (test_bit(BTRFS_DELAYED_NODE_IN_LIST, &node->flags)) { 173 if (!list_empty(&node->p_list)) 174 list_move_tail(&node->p_list, &root->prepare_list); 175 else if (mod) 176 list_add_tail(&node->p_list, &root->prepare_list); 177 } else { 178 list_add_tail(&node->n_list, &root->node_list); 179 list_add_tail(&node->p_list, &root->prepare_list); 180 refcount_inc(&node->refs); /* inserted into list */ 181 root->nodes++; 182 set_bit(BTRFS_DELAYED_NODE_IN_LIST, &node->flags); 183 } 184 spin_unlock(&root->lock); 185 } 186 187 /* Call it when holding delayed_node->mutex */ 188 static void btrfs_dequeue_delayed_node(struct btrfs_delayed_root *root, 189 struct btrfs_delayed_node *node) 190 { 191 spin_lock(&root->lock); 192 if (test_bit(BTRFS_DELAYED_NODE_IN_LIST, &node->flags)) { 193 root->nodes--; 194 refcount_dec(&node->refs); /* not in the list */ 195 list_del_init(&node->n_list); 196 if (!list_empty(&node->p_list)) 197 list_del_init(&node->p_list); 198 clear_bit(BTRFS_DELAYED_NODE_IN_LIST, &node->flags); 199 } 200 spin_unlock(&root->lock); 201 } 202 203 static struct btrfs_delayed_node *btrfs_first_delayed_node( 204 struct btrfs_delayed_root *delayed_root) 205 { 206 struct list_head *p; 207 struct btrfs_delayed_node *node = NULL; 208 209 spin_lock(&delayed_root->lock); 210 if (list_empty(&delayed_root->node_list)) 211 goto out; 212 213 p = delayed_root->node_list.next; 214 node = list_entry(p, struct btrfs_delayed_node, n_list); 215 refcount_inc(&node->refs); 216 out: 217 spin_unlock(&delayed_root->lock); 218 219 return node; 220 } 221 222 static struct btrfs_delayed_node *btrfs_next_delayed_node( 223 struct btrfs_delayed_node *node) 224 { 225 struct btrfs_delayed_root *delayed_root; 226 struct list_head *p; 227 struct btrfs_delayed_node *next = NULL; 228 229 delayed_root = node->root->fs_info->delayed_root; 230 spin_lock(&delayed_root->lock); 231 if (!test_bit(BTRFS_DELAYED_NODE_IN_LIST, &node->flags)) { 232 /* not in the list */ 233 if (list_empty(&delayed_root->node_list)) 234 goto out; 235 p = delayed_root->node_list.next; 236 } else if (list_is_last(&node->n_list, &delayed_root->node_list)) 237 goto out; 238 else 239 p = node->n_list.next; 240 241 next = list_entry(p, struct btrfs_delayed_node, n_list); 242 refcount_inc(&next->refs); 243 out: 244 spin_unlock(&delayed_root->lock); 245 246 return next; 247 } 248 249 static void __btrfs_release_delayed_node( 250 struct btrfs_delayed_node *delayed_node, 251 int mod) 252 { 253 struct btrfs_delayed_root *delayed_root; 254 255 if (!delayed_node) 256 return; 257 258 delayed_root = delayed_node->root->fs_info->delayed_root; 259 260 mutex_lock(&delayed_node->mutex); 261 if (delayed_node->count) 262 btrfs_queue_delayed_node(delayed_root, delayed_node, mod); 263 else 264 btrfs_dequeue_delayed_node(delayed_root, delayed_node); 265 mutex_unlock(&delayed_node->mutex); 266 267 if (refcount_dec_and_test(&delayed_node->refs)) { 268 struct btrfs_root *root = delayed_node->root; 269 270 spin_lock(&root->inode_lock); 271 /* 272 * Once our refcount goes to zero, nobody is allowed to bump it 273 * back up. We can delete it now. 274 */ 275 ASSERT(refcount_read(&delayed_node->refs) == 0); 276 radix_tree_delete(&root->delayed_nodes_tree, 277 delayed_node->inode_id); 278 spin_unlock(&root->inode_lock); 279 kmem_cache_free(delayed_node_cache, delayed_node); 280 } 281 } 282 283 static inline void btrfs_release_delayed_node(struct btrfs_delayed_node *node) 284 { 285 __btrfs_release_delayed_node(node, 0); 286 } 287 288 static struct btrfs_delayed_node *btrfs_first_prepared_delayed_node( 289 struct btrfs_delayed_root *delayed_root) 290 { 291 struct list_head *p; 292 struct btrfs_delayed_node *node = NULL; 293 294 spin_lock(&delayed_root->lock); 295 if (list_empty(&delayed_root->prepare_list)) 296 goto out; 297 298 p = delayed_root->prepare_list.next; 299 list_del_init(p); 300 node = list_entry(p, struct btrfs_delayed_node, p_list); 301 refcount_inc(&node->refs); 302 out: 303 spin_unlock(&delayed_root->lock); 304 305 return node; 306 } 307 308 static inline void btrfs_release_prepared_delayed_node( 309 struct btrfs_delayed_node *node) 310 { 311 __btrfs_release_delayed_node(node, 1); 312 } 313 314 static struct btrfs_delayed_item *btrfs_alloc_delayed_item(u32 data_len) 315 { 316 struct btrfs_delayed_item *item; 317 item = kmalloc(sizeof(*item) + data_len, GFP_NOFS); 318 if (item) { 319 item->data_len = data_len; 320 item->ins_or_del = 0; 321 item->bytes_reserved = 0; 322 item->delayed_node = NULL; 323 refcount_set(&item->refs, 1); 324 } 325 return item; 326 } 327 328 /* 329 * __btrfs_lookup_delayed_item - look up the delayed item by key 330 * @delayed_node: pointer to the delayed node 331 * @key: the key to look up 332 * @prev: used to store the prev item if the right item isn't found 333 * @next: used to store the next item if the right item isn't found 334 * 335 * Note: if we don't find the right item, we will return the prev item and 336 * the next item. 337 */ 338 static struct btrfs_delayed_item *__btrfs_lookup_delayed_item( 339 struct rb_root *root, 340 struct btrfs_key *key, 341 struct btrfs_delayed_item **prev, 342 struct btrfs_delayed_item **next) 343 { 344 struct rb_node *node, *prev_node = NULL; 345 struct btrfs_delayed_item *delayed_item = NULL; 346 int ret = 0; 347 348 node = root->rb_node; 349 350 while (node) { 351 delayed_item = rb_entry(node, struct btrfs_delayed_item, 352 rb_node); 353 prev_node = node; 354 ret = btrfs_comp_cpu_keys(&delayed_item->key, key); 355 if (ret < 0) 356 node = node->rb_right; 357 else if (ret > 0) 358 node = node->rb_left; 359 else 360 return delayed_item; 361 } 362 363 if (prev) { 364 if (!prev_node) 365 *prev = NULL; 366 else if (ret < 0) 367 *prev = delayed_item; 368 else if ((node = rb_prev(prev_node)) != NULL) { 369 *prev = rb_entry(node, struct btrfs_delayed_item, 370 rb_node); 371 } else 372 *prev = NULL; 373 } 374 375 if (next) { 376 if (!prev_node) 377 *next = NULL; 378 else if (ret > 0) 379 *next = delayed_item; 380 else if ((node = rb_next(prev_node)) != NULL) { 381 *next = rb_entry(node, struct btrfs_delayed_item, 382 rb_node); 383 } else 384 *next = NULL; 385 } 386 return NULL; 387 } 388 389 static struct btrfs_delayed_item *__btrfs_lookup_delayed_insertion_item( 390 struct btrfs_delayed_node *delayed_node, 391 struct btrfs_key *key) 392 { 393 return __btrfs_lookup_delayed_item(&delayed_node->ins_root.rb_root, key, 394 NULL, NULL); 395 } 396 397 static int __btrfs_add_delayed_item(struct btrfs_delayed_node *delayed_node, 398 struct btrfs_delayed_item *ins, 399 int action) 400 { 401 struct rb_node **p, *node; 402 struct rb_node *parent_node = NULL; 403 struct rb_root_cached *root; 404 struct btrfs_delayed_item *item; 405 int cmp; 406 bool leftmost = true; 407 408 if (action == BTRFS_DELAYED_INSERTION_ITEM) 409 root = &delayed_node->ins_root; 410 else if (action == BTRFS_DELAYED_DELETION_ITEM) 411 root = &delayed_node->del_root; 412 else 413 BUG(); 414 p = &root->rb_root.rb_node; 415 node = &ins->rb_node; 416 417 while (*p) { 418 parent_node = *p; 419 item = rb_entry(parent_node, struct btrfs_delayed_item, 420 rb_node); 421 422 cmp = btrfs_comp_cpu_keys(&item->key, &ins->key); 423 if (cmp < 0) { 424 p = &(*p)->rb_right; 425 leftmost = false; 426 } else if (cmp > 0) { 427 p = &(*p)->rb_left; 428 } else { 429 return -EEXIST; 430 } 431 } 432 433 rb_link_node(node, parent_node, p); 434 rb_insert_color_cached(node, root, leftmost); 435 ins->delayed_node = delayed_node; 436 ins->ins_or_del = action; 437 438 if (ins->key.type == BTRFS_DIR_INDEX_KEY && 439 action == BTRFS_DELAYED_INSERTION_ITEM && 440 ins->key.offset >= delayed_node->index_cnt) 441 delayed_node->index_cnt = ins->key.offset + 1; 442 443 delayed_node->count++; 444 atomic_inc(&delayed_node->root->fs_info->delayed_root->items); 445 return 0; 446 } 447 448 static int __btrfs_add_delayed_insertion_item(struct btrfs_delayed_node *node, 449 struct btrfs_delayed_item *item) 450 { 451 return __btrfs_add_delayed_item(node, item, 452 BTRFS_DELAYED_INSERTION_ITEM); 453 } 454 455 static int __btrfs_add_delayed_deletion_item(struct btrfs_delayed_node *node, 456 struct btrfs_delayed_item *item) 457 { 458 return __btrfs_add_delayed_item(node, item, 459 BTRFS_DELAYED_DELETION_ITEM); 460 } 461 462 static void finish_one_item(struct btrfs_delayed_root *delayed_root) 463 { 464 int seq = atomic_inc_return(&delayed_root->items_seq); 465 466 /* atomic_dec_return implies a barrier */ 467 if ((atomic_dec_return(&delayed_root->items) < 468 BTRFS_DELAYED_BACKGROUND || seq % BTRFS_DELAYED_BATCH == 0)) 469 cond_wake_up_nomb(&delayed_root->wait); 470 } 471 472 static void __btrfs_remove_delayed_item(struct btrfs_delayed_item *delayed_item) 473 { 474 struct rb_root_cached *root; 475 struct btrfs_delayed_root *delayed_root; 476 477 delayed_root = delayed_item->delayed_node->root->fs_info->delayed_root; 478 479 BUG_ON(!delayed_root); 480 BUG_ON(delayed_item->ins_or_del != BTRFS_DELAYED_DELETION_ITEM && 481 delayed_item->ins_or_del != BTRFS_DELAYED_INSERTION_ITEM); 482 483 if (delayed_item->ins_or_del == BTRFS_DELAYED_INSERTION_ITEM) 484 root = &delayed_item->delayed_node->ins_root; 485 else 486 root = &delayed_item->delayed_node->del_root; 487 488 rb_erase_cached(&delayed_item->rb_node, root); 489 delayed_item->delayed_node->count--; 490 491 finish_one_item(delayed_root); 492 } 493 494 static void btrfs_release_delayed_item(struct btrfs_delayed_item *item) 495 { 496 if (item) { 497 __btrfs_remove_delayed_item(item); 498 if (refcount_dec_and_test(&item->refs)) 499 kfree(item); 500 } 501 } 502 503 static struct btrfs_delayed_item *__btrfs_first_delayed_insertion_item( 504 struct btrfs_delayed_node *delayed_node) 505 { 506 struct rb_node *p; 507 struct btrfs_delayed_item *item = NULL; 508 509 p = rb_first_cached(&delayed_node->ins_root); 510 if (p) 511 item = rb_entry(p, struct btrfs_delayed_item, rb_node); 512 513 return item; 514 } 515 516 static struct btrfs_delayed_item *__btrfs_first_delayed_deletion_item( 517 struct btrfs_delayed_node *delayed_node) 518 { 519 struct rb_node *p; 520 struct btrfs_delayed_item *item = NULL; 521 522 p = rb_first_cached(&delayed_node->del_root); 523 if (p) 524 item = rb_entry(p, struct btrfs_delayed_item, rb_node); 525 526 return item; 527 } 528 529 static struct btrfs_delayed_item *__btrfs_next_delayed_item( 530 struct btrfs_delayed_item *item) 531 { 532 struct rb_node *p; 533 struct btrfs_delayed_item *next = NULL; 534 535 p = rb_next(&item->rb_node); 536 if (p) 537 next = rb_entry(p, struct btrfs_delayed_item, rb_node); 538 539 return next; 540 } 541 542 static int btrfs_delayed_item_reserve_metadata(struct btrfs_trans_handle *trans, 543 struct btrfs_root *root, 544 struct btrfs_delayed_item *item) 545 { 546 struct btrfs_block_rsv *src_rsv; 547 struct btrfs_block_rsv *dst_rsv; 548 struct btrfs_fs_info *fs_info = root->fs_info; 549 u64 num_bytes; 550 int ret; 551 552 if (!trans->bytes_reserved) 553 return 0; 554 555 src_rsv = trans->block_rsv; 556 dst_rsv = &fs_info->delayed_block_rsv; 557 558 num_bytes = btrfs_calc_trans_metadata_size(fs_info, 1); 559 560 /* 561 * Here we migrate space rsv from transaction rsv, since have already 562 * reserved space when starting a transaction. So no need to reserve 563 * qgroup space here. 564 */ 565 ret = btrfs_block_rsv_migrate(src_rsv, dst_rsv, num_bytes, true); 566 if (!ret) { 567 trace_btrfs_space_reservation(fs_info, "delayed_item", 568 item->key.objectid, 569 num_bytes, 1); 570 item->bytes_reserved = num_bytes; 571 } 572 573 return ret; 574 } 575 576 static void btrfs_delayed_item_release_metadata(struct btrfs_root *root, 577 struct btrfs_delayed_item *item) 578 { 579 struct btrfs_block_rsv *rsv; 580 struct btrfs_fs_info *fs_info = root->fs_info; 581 582 if (!item->bytes_reserved) 583 return; 584 585 rsv = &fs_info->delayed_block_rsv; 586 /* 587 * Check btrfs_delayed_item_reserve_metadata() to see why we don't need 588 * to release/reserve qgroup space. 589 */ 590 trace_btrfs_space_reservation(fs_info, "delayed_item", 591 item->key.objectid, item->bytes_reserved, 592 0); 593 btrfs_block_rsv_release(fs_info, rsv, 594 item->bytes_reserved); 595 } 596 597 static int btrfs_delayed_inode_reserve_metadata( 598 struct btrfs_trans_handle *trans, 599 struct btrfs_root *root, 600 struct btrfs_inode *inode, 601 struct btrfs_delayed_node *node) 602 { 603 struct btrfs_fs_info *fs_info = root->fs_info; 604 struct btrfs_block_rsv *src_rsv; 605 struct btrfs_block_rsv *dst_rsv; 606 u64 num_bytes; 607 int ret; 608 609 src_rsv = trans->block_rsv; 610 dst_rsv = &fs_info->delayed_block_rsv; 611 612 num_bytes = btrfs_calc_trans_metadata_size(fs_info, 1); 613 614 /* 615 * btrfs_dirty_inode will update the inode under btrfs_join_transaction 616 * which doesn't reserve space for speed. This is a problem since we 617 * still need to reserve space for this update, so try to reserve the 618 * space. 619 * 620 * Now if src_rsv == delalloc_block_rsv we'll let it just steal since 621 * we always reserve enough to update the inode item. 622 */ 623 if (!src_rsv || (!trans->bytes_reserved && 624 src_rsv->type != BTRFS_BLOCK_RSV_DELALLOC)) { 625 ret = btrfs_qgroup_reserve_meta_prealloc(root, 626 fs_info->nodesize, true); 627 if (ret < 0) 628 return ret; 629 ret = btrfs_block_rsv_add(root, dst_rsv, num_bytes, 630 BTRFS_RESERVE_NO_FLUSH); 631 /* 632 * Since we're under a transaction reserve_metadata_bytes could 633 * try to commit the transaction which will make it return 634 * EAGAIN to make us stop the transaction we have, so return 635 * ENOSPC instead so that btrfs_dirty_inode knows what to do. 636 */ 637 if (ret == -EAGAIN) { 638 ret = -ENOSPC; 639 btrfs_qgroup_free_meta_prealloc(root, num_bytes); 640 } 641 if (!ret) { 642 node->bytes_reserved = num_bytes; 643 trace_btrfs_space_reservation(fs_info, 644 "delayed_inode", 645 btrfs_ino(inode), 646 num_bytes, 1); 647 } else { 648 btrfs_qgroup_free_meta_prealloc(root, fs_info->nodesize); 649 } 650 return ret; 651 } 652 653 ret = btrfs_block_rsv_migrate(src_rsv, dst_rsv, num_bytes, true); 654 if (!ret) { 655 trace_btrfs_space_reservation(fs_info, "delayed_inode", 656 btrfs_ino(inode), num_bytes, 1); 657 node->bytes_reserved = num_bytes; 658 } 659 660 return ret; 661 } 662 663 static void btrfs_delayed_inode_release_metadata(struct btrfs_fs_info *fs_info, 664 struct btrfs_delayed_node *node, 665 bool qgroup_free) 666 { 667 struct btrfs_block_rsv *rsv; 668 669 if (!node->bytes_reserved) 670 return; 671 672 rsv = &fs_info->delayed_block_rsv; 673 trace_btrfs_space_reservation(fs_info, "delayed_inode", 674 node->inode_id, node->bytes_reserved, 0); 675 btrfs_block_rsv_release(fs_info, rsv, 676 node->bytes_reserved); 677 if (qgroup_free) 678 btrfs_qgroup_free_meta_prealloc(node->root, 679 node->bytes_reserved); 680 else 681 btrfs_qgroup_convert_reserved_meta(node->root, 682 node->bytes_reserved); 683 node->bytes_reserved = 0; 684 } 685 686 /* 687 * This helper will insert some continuous items into the same leaf according 688 * to the free space of the leaf. 689 */ 690 static int btrfs_batch_insert_items(struct btrfs_root *root, 691 struct btrfs_path *path, 692 struct btrfs_delayed_item *item) 693 { 694 struct btrfs_delayed_item *curr, *next; 695 int free_space; 696 int total_data_size = 0, total_size = 0; 697 struct extent_buffer *leaf; 698 char *data_ptr; 699 struct btrfs_key *keys; 700 u32 *data_size; 701 struct list_head head; 702 int slot; 703 int nitems; 704 int i; 705 int ret = 0; 706 707 BUG_ON(!path->nodes[0]); 708 709 leaf = path->nodes[0]; 710 free_space = btrfs_leaf_free_space(leaf); 711 INIT_LIST_HEAD(&head); 712 713 next = item; 714 nitems = 0; 715 716 /* 717 * count the number of the continuous items that we can insert in batch 718 */ 719 while (total_size + next->data_len + sizeof(struct btrfs_item) <= 720 free_space) { 721 total_data_size += next->data_len; 722 total_size += next->data_len + sizeof(struct btrfs_item); 723 list_add_tail(&next->tree_list, &head); 724 nitems++; 725 726 curr = next; 727 next = __btrfs_next_delayed_item(curr); 728 if (!next) 729 break; 730 731 if (!btrfs_is_continuous_delayed_item(curr, next)) 732 break; 733 } 734 735 if (!nitems) { 736 ret = 0; 737 goto out; 738 } 739 740 /* 741 * we need allocate some memory space, but it might cause the task 742 * to sleep, so we set all locked nodes in the path to blocking locks 743 * first. 744 */ 745 btrfs_set_path_blocking(path); 746 747 keys = kmalloc_array(nitems, sizeof(struct btrfs_key), GFP_NOFS); 748 if (!keys) { 749 ret = -ENOMEM; 750 goto out; 751 } 752 753 data_size = kmalloc_array(nitems, sizeof(u32), GFP_NOFS); 754 if (!data_size) { 755 ret = -ENOMEM; 756 goto error; 757 } 758 759 /* get keys of all the delayed items */ 760 i = 0; 761 list_for_each_entry(next, &head, tree_list) { 762 keys[i] = next->key; 763 data_size[i] = next->data_len; 764 i++; 765 } 766 767 /* insert the keys of the items */ 768 setup_items_for_insert(root, path, keys, data_size, 769 total_data_size, total_size, nitems); 770 771 /* insert the dir index items */ 772 slot = path->slots[0]; 773 list_for_each_entry_safe(curr, next, &head, tree_list) { 774 data_ptr = btrfs_item_ptr(leaf, slot, char); 775 write_extent_buffer(leaf, &curr->data, 776 (unsigned long)data_ptr, 777 curr->data_len); 778 slot++; 779 780 btrfs_delayed_item_release_metadata(root, curr); 781 782 list_del(&curr->tree_list); 783 btrfs_release_delayed_item(curr); 784 } 785 786 error: 787 kfree(data_size); 788 kfree(keys); 789 out: 790 return ret; 791 } 792 793 /* 794 * This helper can just do simple insertion that needn't extend item for new 795 * data, such as directory name index insertion, inode insertion. 796 */ 797 static int btrfs_insert_delayed_item(struct btrfs_trans_handle *trans, 798 struct btrfs_root *root, 799 struct btrfs_path *path, 800 struct btrfs_delayed_item *delayed_item) 801 { 802 struct extent_buffer *leaf; 803 char *ptr; 804 int ret; 805 806 ret = btrfs_insert_empty_item(trans, root, path, &delayed_item->key, 807 delayed_item->data_len); 808 if (ret < 0 && ret != -EEXIST) 809 return ret; 810 811 leaf = path->nodes[0]; 812 813 ptr = btrfs_item_ptr(leaf, path->slots[0], char); 814 815 write_extent_buffer(leaf, delayed_item->data, (unsigned long)ptr, 816 delayed_item->data_len); 817 btrfs_mark_buffer_dirty(leaf); 818 819 btrfs_delayed_item_release_metadata(root, delayed_item); 820 return 0; 821 } 822 823 /* 824 * we insert an item first, then if there are some continuous items, we try 825 * to insert those items into the same leaf. 826 */ 827 static int btrfs_insert_delayed_items(struct btrfs_trans_handle *trans, 828 struct btrfs_path *path, 829 struct btrfs_root *root, 830 struct btrfs_delayed_node *node) 831 { 832 struct btrfs_delayed_item *curr, *prev; 833 int ret = 0; 834 835 do_again: 836 mutex_lock(&node->mutex); 837 curr = __btrfs_first_delayed_insertion_item(node); 838 if (!curr) 839 goto insert_end; 840 841 ret = btrfs_insert_delayed_item(trans, root, path, curr); 842 if (ret < 0) { 843 btrfs_release_path(path); 844 goto insert_end; 845 } 846 847 prev = curr; 848 curr = __btrfs_next_delayed_item(prev); 849 if (curr && btrfs_is_continuous_delayed_item(prev, curr)) { 850 /* insert the continuous items into the same leaf */ 851 path->slots[0]++; 852 btrfs_batch_insert_items(root, path, curr); 853 } 854 btrfs_release_delayed_item(prev); 855 btrfs_mark_buffer_dirty(path->nodes[0]); 856 857 btrfs_release_path(path); 858 mutex_unlock(&node->mutex); 859 goto do_again; 860 861 insert_end: 862 mutex_unlock(&node->mutex); 863 return ret; 864 } 865 866 static int btrfs_batch_delete_items(struct btrfs_trans_handle *trans, 867 struct btrfs_root *root, 868 struct btrfs_path *path, 869 struct btrfs_delayed_item *item) 870 { 871 struct btrfs_delayed_item *curr, *next; 872 struct extent_buffer *leaf; 873 struct btrfs_key key; 874 struct list_head head; 875 int nitems, i, last_item; 876 int ret = 0; 877 878 BUG_ON(!path->nodes[0]); 879 880 leaf = path->nodes[0]; 881 882 i = path->slots[0]; 883 last_item = btrfs_header_nritems(leaf) - 1; 884 if (i > last_item) 885 return -ENOENT; /* FIXME: Is errno suitable? */ 886 887 next = item; 888 INIT_LIST_HEAD(&head); 889 btrfs_item_key_to_cpu(leaf, &key, i); 890 nitems = 0; 891 /* 892 * count the number of the dir index items that we can delete in batch 893 */ 894 while (btrfs_comp_cpu_keys(&next->key, &key) == 0) { 895 list_add_tail(&next->tree_list, &head); 896 nitems++; 897 898 curr = next; 899 next = __btrfs_next_delayed_item(curr); 900 if (!next) 901 break; 902 903 if (!btrfs_is_continuous_delayed_item(curr, next)) 904 break; 905 906 i++; 907 if (i > last_item) 908 break; 909 btrfs_item_key_to_cpu(leaf, &key, i); 910 } 911 912 if (!nitems) 913 return 0; 914 915 ret = btrfs_del_items(trans, root, path, path->slots[0], nitems); 916 if (ret) 917 goto out; 918 919 list_for_each_entry_safe(curr, next, &head, tree_list) { 920 btrfs_delayed_item_release_metadata(root, curr); 921 list_del(&curr->tree_list); 922 btrfs_release_delayed_item(curr); 923 } 924 925 out: 926 return ret; 927 } 928 929 static int btrfs_delete_delayed_items(struct btrfs_trans_handle *trans, 930 struct btrfs_path *path, 931 struct btrfs_root *root, 932 struct btrfs_delayed_node *node) 933 { 934 struct btrfs_delayed_item *curr, *prev; 935 int ret = 0; 936 937 do_again: 938 mutex_lock(&node->mutex); 939 curr = __btrfs_first_delayed_deletion_item(node); 940 if (!curr) 941 goto delete_fail; 942 943 ret = btrfs_search_slot(trans, root, &curr->key, path, -1, 1); 944 if (ret < 0) 945 goto delete_fail; 946 else if (ret > 0) { 947 /* 948 * can't find the item which the node points to, so this node 949 * is invalid, just drop it. 950 */ 951 prev = curr; 952 curr = __btrfs_next_delayed_item(prev); 953 btrfs_release_delayed_item(prev); 954 ret = 0; 955 btrfs_release_path(path); 956 if (curr) { 957 mutex_unlock(&node->mutex); 958 goto do_again; 959 } else 960 goto delete_fail; 961 } 962 963 btrfs_batch_delete_items(trans, root, path, curr); 964 btrfs_release_path(path); 965 mutex_unlock(&node->mutex); 966 goto do_again; 967 968 delete_fail: 969 btrfs_release_path(path); 970 mutex_unlock(&node->mutex); 971 return ret; 972 } 973 974 static void btrfs_release_delayed_inode(struct btrfs_delayed_node *delayed_node) 975 { 976 struct btrfs_delayed_root *delayed_root; 977 978 if (delayed_node && 979 test_bit(BTRFS_DELAYED_NODE_INODE_DIRTY, &delayed_node->flags)) { 980 BUG_ON(!delayed_node->root); 981 clear_bit(BTRFS_DELAYED_NODE_INODE_DIRTY, &delayed_node->flags); 982 delayed_node->count--; 983 984 delayed_root = delayed_node->root->fs_info->delayed_root; 985 finish_one_item(delayed_root); 986 } 987 } 988 989 static void btrfs_release_delayed_iref(struct btrfs_delayed_node *delayed_node) 990 { 991 struct btrfs_delayed_root *delayed_root; 992 993 ASSERT(delayed_node->root); 994 clear_bit(BTRFS_DELAYED_NODE_DEL_IREF, &delayed_node->flags); 995 delayed_node->count--; 996 997 delayed_root = delayed_node->root->fs_info->delayed_root; 998 finish_one_item(delayed_root); 999 } 1000 1001 static int __btrfs_update_delayed_inode(struct btrfs_trans_handle *trans, 1002 struct btrfs_root *root, 1003 struct btrfs_path *path, 1004 struct btrfs_delayed_node *node) 1005 { 1006 struct btrfs_fs_info *fs_info = root->fs_info; 1007 struct btrfs_key key; 1008 struct btrfs_inode_item *inode_item; 1009 struct extent_buffer *leaf; 1010 int mod; 1011 int ret; 1012 1013 key.objectid = node->inode_id; 1014 key.type = BTRFS_INODE_ITEM_KEY; 1015 key.offset = 0; 1016 1017 if (test_bit(BTRFS_DELAYED_NODE_DEL_IREF, &node->flags)) 1018 mod = -1; 1019 else 1020 mod = 1; 1021 1022 ret = btrfs_lookup_inode(trans, root, path, &key, mod); 1023 if (ret > 0) { 1024 btrfs_release_path(path); 1025 return -ENOENT; 1026 } else if (ret < 0) { 1027 return ret; 1028 } 1029 1030 leaf = path->nodes[0]; 1031 inode_item = btrfs_item_ptr(leaf, path->slots[0], 1032 struct btrfs_inode_item); 1033 write_extent_buffer(leaf, &node->inode_item, (unsigned long)inode_item, 1034 sizeof(struct btrfs_inode_item)); 1035 btrfs_mark_buffer_dirty(leaf); 1036 1037 if (!test_bit(BTRFS_DELAYED_NODE_DEL_IREF, &node->flags)) 1038 goto no_iref; 1039 1040 path->slots[0]++; 1041 if (path->slots[0] >= btrfs_header_nritems(leaf)) 1042 goto search; 1043 again: 1044 btrfs_item_key_to_cpu(leaf, &key, path->slots[0]); 1045 if (key.objectid != node->inode_id) 1046 goto out; 1047 1048 if (key.type != BTRFS_INODE_REF_KEY && 1049 key.type != BTRFS_INODE_EXTREF_KEY) 1050 goto out; 1051 1052 /* 1053 * Delayed iref deletion is for the inode who has only one link, 1054 * so there is only one iref. The case that several irefs are 1055 * in the same item doesn't exist. 1056 */ 1057 btrfs_del_item(trans, root, path); 1058 out: 1059 btrfs_release_delayed_iref(node); 1060 no_iref: 1061 btrfs_release_path(path); 1062 err_out: 1063 btrfs_delayed_inode_release_metadata(fs_info, node, (ret < 0)); 1064 btrfs_release_delayed_inode(node); 1065 1066 return ret; 1067 1068 search: 1069 btrfs_release_path(path); 1070 1071 key.type = BTRFS_INODE_EXTREF_KEY; 1072 key.offset = -1; 1073 ret = btrfs_search_slot(trans, root, &key, path, -1, 1); 1074 if (ret < 0) 1075 goto err_out; 1076 ASSERT(ret); 1077 1078 ret = 0; 1079 leaf = path->nodes[0]; 1080 path->slots[0]--; 1081 goto again; 1082 } 1083 1084 static inline int btrfs_update_delayed_inode(struct btrfs_trans_handle *trans, 1085 struct btrfs_root *root, 1086 struct btrfs_path *path, 1087 struct btrfs_delayed_node *node) 1088 { 1089 int ret; 1090 1091 mutex_lock(&node->mutex); 1092 if (!test_bit(BTRFS_DELAYED_NODE_INODE_DIRTY, &node->flags)) { 1093 mutex_unlock(&node->mutex); 1094 return 0; 1095 } 1096 1097 ret = __btrfs_update_delayed_inode(trans, root, path, node); 1098 mutex_unlock(&node->mutex); 1099 return ret; 1100 } 1101 1102 static inline int 1103 __btrfs_commit_inode_delayed_items(struct btrfs_trans_handle *trans, 1104 struct btrfs_path *path, 1105 struct btrfs_delayed_node *node) 1106 { 1107 int ret; 1108 1109 ret = btrfs_insert_delayed_items(trans, path, node->root, node); 1110 if (ret) 1111 return ret; 1112 1113 ret = btrfs_delete_delayed_items(trans, path, node->root, node); 1114 if (ret) 1115 return ret; 1116 1117 ret = btrfs_update_delayed_inode(trans, node->root, path, node); 1118 return ret; 1119 } 1120 1121 /* 1122 * Called when committing the transaction. 1123 * Returns 0 on success. 1124 * Returns < 0 on error and returns with an aborted transaction with any 1125 * outstanding delayed items cleaned up. 1126 */ 1127 static int __btrfs_run_delayed_items(struct btrfs_trans_handle *trans, int nr) 1128 { 1129 struct btrfs_fs_info *fs_info = trans->fs_info; 1130 struct btrfs_delayed_root *delayed_root; 1131 struct btrfs_delayed_node *curr_node, *prev_node; 1132 struct btrfs_path *path; 1133 struct btrfs_block_rsv *block_rsv; 1134 int ret = 0; 1135 bool count = (nr > 0); 1136 1137 if (trans->aborted) 1138 return -EIO; 1139 1140 path = btrfs_alloc_path(); 1141 if (!path) 1142 return -ENOMEM; 1143 path->leave_spinning = 1; 1144 1145 block_rsv = trans->block_rsv; 1146 trans->block_rsv = &fs_info->delayed_block_rsv; 1147 1148 delayed_root = fs_info->delayed_root; 1149 1150 curr_node = btrfs_first_delayed_node(delayed_root); 1151 while (curr_node && (!count || (count && nr--))) { 1152 ret = __btrfs_commit_inode_delayed_items(trans, path, 1153 curr_node); 1154 if (ret) { 1155 btrfs_release_delayed_node(curr_node); 1156 curr_node = NULL; 1157 btrfs_abort_transaction(trans, ret); 1158 break; 1159 } 1160 1161 prev_node = curr_node; 1162 curr_node = btrfs_next_delayed_node(curr_node); 1163 btrfs_release_delayed_node(prev_node); 1164 } 1165 1166 if (curr_node) 1167 btrfs_release_delayed_node(curr_node); 1168 btrfs_free_path(path); 1169 trans->block_rsv = block_rsv; 1170 1171 return ret; 1172 } 1173 1174 int btrfs_run_delayed_items(struct btrfs_trans_handle *trans) 1175 { 1176 return __btrfs_run_delayed_items(trans, -1); 1177 } 1178 1179 int btrfs_run_delayed_items_nr(struct btrfs_trans_handle *trans, int nr) 1180 { 1181 return __btrfs_run_delayed_items(trans, nr); 1182 } 1183 1184 int btrfs_commit_inode_delayed_items(struct btrfs_trans_handle *trans, 1185 struct btrfs_inode *inode) 1186 { 1187 struct btrfs_delayed_node *delayed_node = btrfs_get_delayed_node(inode); 1188 struct btrfs_path *path; 1189 struct btrfs_block_rsv *block_rsv; 1190 int ret; 1191 1192 if (!delayed_node) 1193 return 0; 1194 1195 mutex_lock(&delayed_node->mutex); 1196 if (!delayed_node->count) { 1197 mutex_unlock(&delayed_node->mutex); 1198 btrfs_release_delayed_node(delayed_node); 1199 return 0; 1200 } 1201 mutex_unlock(&delayed_node->mutex); 1202 1203 path = btrfs_alloc_path(); 1204 if (!path) { 1205 btrfs_release_delayed_node(delayed_node); 1206 return -ENOMEM; 1207 } 1208 path->leave_spinning = 1; 1209 1210 block_rsv = trans->block_rsv; 1211 trans->block_rsv = &delayed_node->root->fs_info->delayed_block_rsv; 1212 1213 ret = __btrfs_commit_inode_delayed_items(trans, path, delayed_node); 1214 1215 btrfs_release_delayed_node(delayed_node); 1216 btrfs_free_path(path); 1217 trans->block_rsv = block_rsv; 1218 1219 return ret; 1220 } 1221 1222 int btrfs_commit_inode_delayed_inode(struct btrfs_inode *inode) 1223 { 1224 struct btrfs_fs_info *fs_info = inode->root->fs_info; 1225 struct btrfs_trans_handle *trans; 1226 struct btrfs_delayed_node *delayed_node = btrfs_get_delayed_node(inode); 1227 struct btrfs_path *path; 1228 struct btrfs_block_rsv *block_rsv; 1229 int ret; 1230 1231 if (!delayed_node) 1232 return 0; 1233 1234 mutex_lock(&delayed_node->mutex); 1235 if (!test_bit(BTRFS_DELAYED_NODE_INODE_DIRTY, &delayed_node->flags)) { 1236 mutex_unlock(&delayed_node->mutex); 1237 btrfs_release_delayed_node(delayed_node); 1238 return 0; 1239 } 1240 mutex_unlock(&delayed_node->mutex); 1241 1242 trans = btrfs_join_transaction(delayed_node->root); 1243 if (IS_ERR(trans)) { 1244 ret = PTR_ERR(trans); 1245 goto out; 1246 } 1247 1248 path = btrfs_alloc_path(); 1249 if (!path) { 1250 ret = -ENOMEM; 1251 goto trans_out; 1252 } 1253 path->leave_spinning = 1; 1254 1255 block_rsv = trans->block_rsv; 1256 trans->block_rsv = &fs_info->delayed_block_rsv; 1257 1258 mutex_lock(&delayed_node->mutex); 1259 if (test_bit(BTRFS_DELAYED_NODE_INODE_DIRTY, &delayed_node->flags)) 1260 ret = __btrfs_update_delayed_inode(trans, delayed_node->root, 1261 path, delayed_node); 1262 else 1263 ret = 0; 1264 mutex_unlock(&delayed_node->mutex); 1265 1266 btrfs_free_path(path); 1267 trans->block_rsv = block_rsv; 1268 trans_out: 1269 btrfs_end_transaction(trans); 1270 btrfs_btree_balance_dirty(fs_info); 1271 out: 1272 btrfs_release_delayed_node(delayed_node); 1273 1274 return ret; 1275 } 1276 1277 void btrfs_remove_delayed_node(struct btrfs_inode *inode) 1278 { 1279 struct btrfs_delayed_node *delayed_node; 1280 1281 delayed_node = READ_ONCE(inode->delayed_node); 1282 if (!delayed_node) 1283 return; 1284 1285 inode->delayed_node = NULL; 1286 btrfs_release_delayed_node(delayed_node); 1287 } 1288 1289 struct btrfs_async_delayed_work { 1290 struct btrfs_delayed_root *delayed_root; 1291 int nr; 1292 struct btrfs_work work; 1293 }; 1294 1295 static void btrfs_async_run_delayed_root(struct btrfs_work *work) 1296 { 1297 struct btrfs_async_delayed_work *async_work; 1298 struct btrfs_delayed_root *delayed_root; 1299 struct btrfs_trans_handle *trans; 1300 struct btrfs_path *path; 1301 struct btrfs_delayed_node *delayed_node = NULL; 1302 struct btrfs_root *root; 1303 struct btrfs_block_rsv *block_rsv; 1304 int total_done = 0; 1305 1306 async_work = container_of(work, struct btrfs_async_delayed_work, work); 1307 delayed_root = async_work->delayed_root; 1308 1309 path = btrfs_alloc_path(); 1310 if (!path) 1311 goto out; 1312 1313 do { 1314 if (atomic_read(&delayed_root->items) < 1315 BTRFS_DELAYED_BACKGROUND / 2) 1316 break; 1317 1318 delayed_node = btrfs_first_prepared_delayed_node(delayed_root); 1319 if (!delayed_node) 1320 break; 1321 1322 path->leave_spinning = 1; 1323 root = delayed_node->root; 1324 1325 trans = btrfs_join_transaction(root); 1326 if (IS_ERR(trans)) { 1327 btrfs_release_path(path); 1328 btrfs_release_prepared_delayed_node(delayed_node); 1329 total_done++; 1330 continue; 1331 } 1332 1333 block_rsv = trans->block_rsv; 1334 trans->block_rsv = &root->fs_info->delayed_block_rsv; 1335 1336 __btrfs_commit_inode_delayed_items(trans, path, delayed_node); 1337 1338 trans->block_rsv = block_rsv; 1339 btrfs_end_transaction(trans); 1340 btrfs_btree_balance_dirty_nodelay(root->fs_info); 1341 1342 btrfs_release_path(path); 1343 btrfs_release_prepared_delayed_node(delayed_node); 1344 total_done++; 1345 1346 } while ((async_work->nr == 0 && total_done < BTRFS_DELAYED_WRITEBACK) 1347 || total_done < async_work->nr); 1348 1349 btrfs_free_path(path); 1350 out: 1351 wake_up(&delayed_root->wait); 1352 kfree(async_work); 1353 } 1354 1355 1356 static int btrfs_wq_run_delayed_node(struct btrfs_delayed_root *delayed_root, 1357 struct btrfs_fs_info *fs_info, int nr) 1358 { 1359 struct btrfs_async_delayed_work *async_work; 1360 1361 async_work = kmalloc(sizeof(*async_work), GFP_NOFS); 1362 if (!async_work) 1363 return -ENOMEM; 1364 1365 async_work->delayed_root = delayed_root; 1366 btrfs_init_work(&async_work->work, btrfs_delayed_meta_helper, 1367 btrfs_async_run_delayed_root, NULL, NULL); 1368 async_work->nr = nr; 1369 1370 btrfs_queue_work(fs_info->delayed_workers, &async_work->work); 1371 return 0; 1372 } 1373 1374 void btrfs_assert_delayed_root_empty(struct btrfs_fs_info *fs_info) 1375 { 1376 WARN_ON(btrfs_first_delayed_node(fs_info->delayed_root)); 1377 } 1378 1379 static int could_end_wait(struct btrfs_delayed_root *delayed_root, int seq) 1380 { 1381 int val = atomic_read(&delayed_root->items_seq); 1382 1383 if (val < seq || val >= seq + BTRFS_DELAYED_BATCH) 1384 return 1; 1385 1386 if (atomic_read(&delayed_root->items) < BTRFS_DELAYED_BACKGROUND) 1387 return 1; 1388 1389 return 0; 1390 } 1391 1392 void btrfs_balance_delayed_items(struct btrfs_fs_info *fs_info) 1393 { 1394 struct btrfs_delayed_root *delayed_root = fs_info->delayed_root; 1395 1396 if ((atomic_read(&delayed_root->items) < BTRFS_DELAYED_BACKGROUND) || 1397 btrfs_workqueue_normal_congested(fs_info->delayed_workers)) 1398 return; 1399 1400 if (atomic_read(&delayed_root->items) >= BTRFS_DELAYED_WRITEBACK) { 1401 int seq; 1402 int ret; 1403 1404 seq = atomic_read(&delayed_root->items_seq); 1405 1406 ret = btrfs_wq_run_delayed_node(delayed_root, fs_info, 0); 1407 if (ret) 1408 return; 1409 1410 wait_event_interruptible(delayed_root->wait, 1411 could_end_wait(delayed_root, seq)); 1412 return; 1413 } 1414 1415 btrfs_wq_run_delayed_node(delayed_root, fs_info, BTRFS_DELAYED_BATCH); 1416 } 1417 1418 /* Will return 0 or -ENOMEM */ 1419 int btrfs_insert_delayed_dir_index(struct btrfs_trans_handle *trans, 1420 const char *name, int name_len, 1421 struct btrfs_inode *dir, 1422 struct btrfs_disk_key *disk_key, u8 type, 1423 u64 index) 1424 { 1425 struct btrfs_delayed_node *delayed_node; 1426 struct btrfs_delayed_item *delayed_item; 1427 struct btrfs_dir_item *dir_item; 1428 int ret; 1429 1430 delayed_node = btrfs_get_or_create_delayed_node(dir); 1431 if (IS_ERR(delayed_node)) 1432 return PTR_ERR(delayed_node); 1433 1434 delayed_item = btrfs_alloc_delayed_item(sizeof(*dir_item) + name_len); 1435 if (!delayed_item) { 1436 ret = -ENOMEM; 1437 goto release_node; 1438 } 1439 1440 delayed_item->key.objectid = btrfs_ino(dir); 1441 delayed_item->key.type = BTRFS_DIR_INDEX_KEY; 1442 delayed_item->key.offset = index; 1443 1444 dir_item = (struct btrfs_dir_item *)delayed_item->data; 1445 dir_item->location = *disk_key; 1446 btrfs_set_stack_dir_transid(dir_item, trans->transid); 1447 btrfs_set_stack_dir_data_len(dir_item, 0); 1448 btrfs_set_stack_dir_name_len(dir_item, name_len); 1449 btrfs_set_stack_dir_type(dir_item, type); 1450 memcpy((char *)(dir_item + 1), name, name_len); 1451 1452 ret = btrfs_delayed_item_reserve_metadata(trans, dir->root, delayed_item); 1453 /* 1454 * we have reserved enough space when we start a new transaction, 1455 * so reserving metadata failure is impossible 1456 */ 1457 BUG_ON(ret); 1458 1459 mutex_lock(&delayed_node->mutex); 1460 ret = __btrfs_add_delayed_insertion_item(delayed_node, delayed_item); 1461 if (unlikely(ret)) { 1462 btrfs_err(trans->fs_info, 1463 "err add delayed dir index item(name: %.*s) into the insertion tree of the delayed node(root id: %llu, inode id: %llu, errno: %d)", 1464 name_len, name, delayed_node->root->root_key.objectid, 1465 delayed_node->inode_id, ret); 1466 BUG(); 1467 } 1468 mutex_unlock(&delayed_node->mutex); 1469 1470 release_node: 1471 btrfs_release_delayed_node(delayed_node); 1472 return ret; 1473 } 1474 1475 static int btrfs_delete_delayed_insertion_item(struct btrfs_fs_info *fs_info, 1476 struct btrfs_delayed_node *node, 1477 struct btrfs_key *key) 1478 { 1479 struct btrfs_delayed_item *item; 1480 1481 mutex_lock(&node->mutex); 1482 item = __btrfs_lookup_delayed_insertion_item(node, key); 1483 if (!item) { 1484 mutex_unlock(&node->mutex); 1485 return 1; 1486 } 1487 1488 btrfs_delayed_item_release_metadata(node->root, item); 1489 btrfs_release_delayed_item(item); 1490 mutex_unlock(&node->mutex); 1491 return 0; 1492 } 1493 1494 int btrfs_delete_delayed_dir_index(struct btrfs_trans_handle *trans, 1495 struct btrfs_inode *dir, u64 index) 1496 { 1497 struct btrfs_delayed_node *node; 1498 struct btrfs_delayed_item *item; 1499 struct btrfs_key item_key; 1500 int ret; 1501 1502 node = btrfs_get_or_create_delayed_node(dir); 1503 if (IS_ERR(node)) 1504 return PTR_ERR(node); 1505 1506 item_key.objectid = btrfs_ino(dir); 1507 item_key.type = BTRFS_DIR_INDEX_KEY; 1508 item_key.offset = index; 1509 1510 ret = btrfs_delete_delayed_insertion_item(trans->fs_info, node, 1511 &item_key); 1512 if (!ret) 1513 goto end; 1514 1515 item = btrfs_alloc_delayed_item(0); 1516 if (!item) { 1517 ret = -ENOMEM; 1518 goto end; 1519 } 1520 1521 item->key = item_key; 1522 1523 ret = btrfs_delayed_item_reserve_metadata(trans, dir->root, item); 1524 /* 1525 * we have reserved enough space when we start a new transaction, 1526 * so reserving metadata failure is impossible. 1527 */ 1528 BUG_ON(ret); 1529 1530 mutex_lock(&node->mutex); 1531 ret = __btrfs_add_delayed_deletion_item(node, item); 1532 if (unlikely(ret)) { 1533 btrfs_err(trans->fs_info, 1534 "err add delayed dir index item(index: %llu) into the deletion tree of the delayed node(root id: %llu, inode id: %llu, errno: %d)", 1535 index, node->root->root_key.objectid, 1536 node->inode_id, ret); 1537 BUG(); 1538 } 1539 mutex_unlock(&node->mutex); 1540 end: 1541 btrfs_release_delayed_node(node); 1542 return ret; 1543 } 1544 1545 int btrfs_inode_delayed_dir_index_count(struct btrfs_inode *inode) 1546 { 1547 struct btrfs_delayed_node *delayed_node = btrfs_get_delayed_node(inode); 1548 1549 if (!delayed_node) 1550 return -ENOENT; 1551 1552 /* 1553 * Since we have held i_mutex of this directory, it is impossible that 1554 * a new directory index is added into the delayed node and index_cnt 1555 * is updated now. So we needn't lock the delayed node. 1556 */ 1557 if (!delayed_node->index_cnt) { 1558 btrfs_release_delayed_node(delayed_node); 1559 return -EINVAL; 1560 } 1561 1562 inode->index_cnt = delayed_node->index_cnt; 1563 btrfs_release_delayed_node(delayed_node); 1564 return 0; 1565 } 1566 1567 bool btrfs_readdir_get_delayed_items(struct inode *inode, 1568 struct list_head *ins_list, 1569 struct list_head *del_list) 1570 { 1571 struct btrfs_delayed_node *delayed_node; 1572 struct btrfs_delayed_item *item; 1573 1574 delayed_node = btrfs_get_delayed_node(BTRFS_I(inode)); 1575 if (!delayed_node) 1576 return false; 1577 1578 /* 1579 * We can only do one readdir with delayed items at a time because of 1580 * item->readdir_list. 1581 */ 1582 inode_unlock_shared(inode); 1583 inode_lock(inode); 1584 1585 mutex_lock(&delayed_node->mutex); 1586 item = __btrfs_first_delayed_insertion_item(delayed_node); 1587 while (item) { 1588 refcount_inc(&item->refs); 1589 list_add_tail(&item->readdir_list, ins_list); 1590 item = __btrfs_next_delayed_item(item); 1591 } 1592 1593 item = __btrfs_first_delayed_deletion_item(delayed_node); 1594 while (item) { 1595 refcount_inc(&item->refs); 1596 list_add_tail(&item->readdir_list, del_list); 1597 item = __btrfs_next_delayed_item(item); 1598 } 1599 mutex_unlock(&delayed_node->mutex); 1600 /* 1601 * This delayed node is still cached in the btrfs inode, so refs 1602 * must be > 1 now, and we needn't check it is going to be freed 1603 * or not. 1604 * 1605 * Besides that, this function is used to read dir, we do not 1606 * insert/delete delayed items in this period. So we also needn't 1607 * requeue or dequeue this delayed node. 1608 */ 1609 refcount_dec(&delayed_node->refs); 1610 1611 return true; 1612 } 1613 1614 void btrfs_readdir_put_delayed_items(struct inode *inode, 1615 struct list_head *ins_list, 1616 struct list_head *del_list) 1617 { 1618 struct btrfs_delayed_item *curr, *next; 1619 1620 list_for_each_entry_safe(curr, next, ins_list, readdir_list) { 1621 list_del(&curr->readdir_list); 1622 if (refcount_dec_and_test(&curr->refs)) 1623 kfree(curr); 1624 } 1625 1626 list_for_each_entry_safe(curr, next, del_list, readdir_list) { 1627 list_del(&curr->readdir_list); 1628 if (refcount_dec_and_test(&curr->refs)) 1629 kfree(curr); 1630 } 1631 1632 /* 1633 * The VFS is going to do up_read(), so we need to downgrade back to a 1634 * read lock. 1635 */ 1636 downgrade_write(&inode->i_rwsem); 1637 } 1638 1639 int btrfs_should_delete_dir_index(struct list_head *del_list, 1640 u64 index) 1641 { 1642 struct btrfs_delayed_item *curr; 1643 int ret = 0; 1644 1645 list_for_each_entry(curr, del_list, readdir_list) { 1646 if (curr->key.offset > index) 1647 break; 1648 if (curr->key.offset == index) { 1649 ret = 1; 1650 break; 1651 } 1652 } 1653 return ret; 1654 } 1655 1656 /* 1657 * btrfs_readdir_delayed_dir_index - read dir info stored in the delayed tree 1658 * 1659 */ 1660 int btrfs_readdir_delayed_dir_index(struct dir_context *ctx, 1661 struct list_head *ins_list) 1662 { 1663 struct btrfs_dir_item *di; 1664 struct btrfs_delayed_item *curr, *next; 1665 struct btrfs_key location; 1666 char *name; 1667 int name_len; 1668 int over = 0; 1669 unsigned char d_type; 1670 1671 if (list_empty(ins_list)) 1672 return 0; 1673 1674 /* 1675 * Changing the data of the delayed item is impossible. So 1676 * we needn't lock them. And we have held i_mutex of the 1677 * directory, nobody can delete any directory indexes now. 1678 */ 1679 list_for_each_entry_safe(curr, next, ins_list, readdir_list) { 1680 list_del(&curr->readdir_list); 1681 1682 if (curr->key.offset < ctx->pos) { 1683 if (refcount_dec_and_test(&curr->refs)) 1684 kfree(curr); 1685 continue; 1686 } 1687 1688 ctx->pos = curr->key.offset; 1689 1690 di = (struct btrfs_dir_item *)curr->data; 1691 name = (char *)(di + 1); 1692 name_len = btrfs_stack_dir_name_len(di); 1693 1694 d_type = fs_ftype_to_dtype(di->type); 1695 btrfs_disk_key_to_cpu(&location, &di->location); 1696 1697 over = !dir_emit(ctx, name, name_len, 1698 location.objectid, d_type); 1699 1700 if (refcount_dec_and_test(&curr->refs)) 1701 kfree(curr); 1702 1703 if (over) 1704 return 1; 1705 ctx->pos++; 1706 } 1707 return 0; 1708 } 1709 1710 static void fill_stack_inode_item(struct btrfs_trans_handle *trans, 1711 struct btrfs_inode_item *inode_item, 1712 struct inode *inode) 1713 { 1714 btrfs_set_stack_inode_uid(inode_item, i_uid_read(inode)); 1715 btrfs_set_stack_inode_gid(inode_item, i_gid_read(inode)); 1716 btrfs_set_stack_inode_size(inode_item, BTRFS_I(inode)->disk_i_size); 1717 btrfs_set_stack_inode_mode(inode_item, inode->i_mode); 1718 btrfs_set_stack_inode_nlink(inode_item, inode->i_nlink); 1719 btrfs_set_stack_inode_nbytes(inode_item, inode_get_bytes(inode)); 1720 btrfs_set_stack_inode_generation(inode_item, 1721 BTRFS_I(inode)->generation); 1722 btrfs_set_stack_inode_sequence(inode_item, 1723 inode_peek_iversion(inode)); 1724 btrfs_set_stack_inode_transid(inode_item, trans->transid); 1725 btrfs_set_stack_inode_rdev(inode_item, inode->i_rdev); 1726 btrfs_set_stack_inode_flags(inode_item, BTRFS_I(inode)->flags); 1727 btrfs_set_stack_inode_block_group(inode_item, 0); 1728 1729 btrfs_set_stack_timespec_sec(&inode_item->atime, 1730 inode->i_atime.tv_sec); 1731 btrfs_set_stack_timespec_nsec(&inode_item->atime, 1732 inode->i_atime.tv_nsec); 1733 1734 btrfs_set_stack_timespec_sec(&inode_item->mtime, 1735 inode->i_mtime.tv_sec); 1736 btrfs_set_stack_timespec_nsec(&inode_item->mtime, 1737 inode->i_mtime.tv_nsec); 1738 1739 btrfs_set_stack_timespec_sec(&inode_item->ctime, 1740 inode->i_ctime.tv_sec); 1741 btrfs_set_stack_timespec_nsec(&inode_item->ctime, 1742 inode->i_ctime.tv_nsec); 1743 1744 btrfs_set_stack_timespec_sec(&inode_item->otime, 1745 BTRFS_I(inode)->i_otime.tv_sec); 1746 btrfs_set_stack_timespec_nsec(&inode_item->otime, 1747 BTRFS_I(inode)->i_otime.tv_nsec); 1748 } 1749 1750 int btrfs_fill_inode(struct inode *inode, u32 *rdev) 1751 { 1752 struct btrfs_delayed_node *delayed_node; 1753 struct btrfs_inode_item *inode_item; 1754 1755 delayed_node = btrfs_get_delayed_node(BTRFS_I(inode)); 1756 if (!delayed_node) 1757 return -ENOENT; 1758 1759 mutex_lock(&delayed_node->mutex); 1760 if (!test_bit(BTRFS_DELAYED_NODE_INODE_DIRTY, &delayed_node->flags)) { 1761 mutex_unlock(&delayed_node->mutex); 1762 btrfs_release_delayed_node(delayed_node); 1763 return -ENOENT; 1764 } 1765 1766 inode_item = &delayed_node->inode_item; 1767 1768 i_uid_write(inode, btrfs_stack_inode_uid(inode_item)); 1769 i_gid_write(inode, btrfs_stack_inode_gid(inode_item)); 1770 btrfs_i_size_write(BTRFS_I(inode), btrfs_stack_inode_size(inode_item)); 1771 inode->i_mode = btrfs_stack_inode_mode(inode_item); 1772 set_nlink(inode, btrfs_stack_inode_nlink(inode_item)); 1773 inode_set_bytes(inode, btrfs_stack_inode_nbytes(inode_item)); 1774 BTRFS_I(inode)->generation = btrfs_stack_inode_generation(inode_item); 1775 BTRFS_I(inode)->last_trans = btrfs_stack_inode_transid(inode_item); 1776 1777 inode_set_iversion_queried(inode, 1778 btrfs_stack_inode_sequence(inode_item)); 1779 inode->i_rdev = 0; 1780 *rdev = btrfs_stack_inode_rdev(inode_item); 1781 BTRFS_I(inode)->flags = btrfs_stack_inode_flags(inode_item); 1782 1783 inode->i_atime.tv_sec = btrfs_stack_timespec_sec(&inode_item->atime); 1784 inode->i_atime.tv_nsec = btrfs_stack_timespec_nsec(&inode_item->atime); 1785 1786 inode->i_mtime.tv_sec = btrfs_stack_timespec_sec(&inode_item->mtime); 1787 inode->i_mtime.tv_nsec = btrfs_stack_timespec_nsec(&inode_item->mtime); 1788 1789 inode->i_ctime.tv_sec = btrfs_stack_timespec_sec(&inode_item->ctime); 1790 inode->i_ctime.tv_nsec = btrfs_stack_timespec_nsec(&inode_item->ctime); 1791 1792 BTRFS_I(inode)->i_otime.tv_sec = 1793 btrfs_stack_timespec_sec(&inode_item->otime); 1794 BTRFS_I(inode)->i_otime.tv_nsec = 1795 btrfs_stack_timespec_nsec(&inode_item->otime); 1796 1797 inode->i_generation = BTRFS_I(inode)->generation; 1798 BTRFS_I(inode)->index_cnt = (u64)-1; 1799 1800 mutex_unlock(&delayed_node->mutex); 1801 btrfs_release_delayed_node(delayed_node); 1802 return 0; 1803 } 1804 1805 int btrfs_delayed_update_inode(struct btrfs_trans_handle *trans, 1806 struct btrfs_root *root, struct inode *inode) 1807 { 1808 struct btrfs_delayed_node *delayed_node; 1809 int ret = 0; 1810 1811 delayed_node = btrfs_get_or_create_delayed_node(BTRFS_I(inode)); 1812 if (IS_ERR(delayed_node)) 1813 return PTR_ERR(delayed_node); 1814 1815 mutex_lock(&delayed_node->mutex); 1816 if (test_bit(BTRFS_DELAYED_NODE_INODE_DIRTY, &delayed_node->flags)) { 1817 fill_stack_inode_item(trans, &delayed_node->inode_item, inode); 1818 goto release_node; 1819 } 1820 1821 ret = btrfs_delayed_inode_reserve_metadata(trans, root, BTRFS_I(inode), 1822 delayed_node); 1823 if (ret) 1824 goto release_node; 1825 1826 fill_stack_inode_item(trans, &delayed_node->inode_item, inode); 1827 set_bit(BTRFS_DELAYED_NODE_INODE_DIRTY, &delayed_node->flags); 1828 delayed_node->count++; 1829 atomic_inc(&root->fs_info->delayed_root->items); 1830 release_node: 1831 mutex_unlock(&delayed_node->mutex); 1832 btrfs_release_delayed_node(delayed_node); 1833 return ret; 1834 } 1835 1836 int btrfs_delayed_delete_inode_ref(struct btrfs_inode *inode) 1837 { 1838 struct btrfs_fs_info *fs_info = inode->root->fs_info; 1839 struct btrfs_delayed_node *delayed_node; 1840 1841 /* 1842 * we don't do delayed inode updates during log recovery because it 1843 * leads to enospc problems. This means we also can't do 1844 * delayed inode refs 1845 */ 1846 if (test_bit(BTRFS_FS_LOG_RECOVERING, &fs_info->flags)) 1847 return -EAGAIN; 1848 1849 delayed_node = btrfs_get_or_create_delayed_node(inode); 1850 if (IS_ERR(delayed_node)) 1851 return PTR_ERR(delayed_node); 1852 1853 /* 1854 * We don't reserve space for inode ref deletion is because: 1855 * - We ONLY do async inode ref deletion for the inode who has only 1856 * one link(i_nlink == 1), it means there is only one inode ref. 1857 * And in most case, the inode ref and the inode item are in the 1858 * same leaf, and we will deal with them at the same time. 1859 * Since we are sure we will reserve the space for the inode item, 1860 * it is unnecessary to reserve space for inode ref deletion. 1861 * - If the inode ref and the inode item are not in the same leaf, 1862 * We also needn't worry about enospc problem, because we reserve 1863 * much more space for the inode update than it needs. 1864 * - At the worst, we can steal some space from the global reservation. 1865 * It is very rare. 1866 */ 1867 mutex_lock(&delayed_node->mutex); 1868 if (test_bit(BTRFS_DELAYED_NODE_DEL_IREF, &delayed_node->flags)) 1869 goto release_node; 1870 1871 set_bit(BTRFS_DELAYED_NODE_DEL_IREF, &delayed_node->flags); 1872 delayed_node->count++; 1873 atomic_inc(&fs_info->delayed_root->items); 1874 release_node: 1875 mutex_unlock(&delayed_node->mutex); 1876 btrfs_release_delayed_node(delayed_node); 1877 return 0; 1878 } 1879 1880 static void __btrfs_kill_delayed_node(struct btrfs_delayed_node *delayed_node) 1881 { 1882 struct btrfs_root *root = delayed_node->root; 1883 struct btrfs_fs_info *fs_info = root->fs_info; 1884 struct btrfs_delayed_item *curr_item, *prev_item; 1885 1886 mutex_lock(&delayed_node->mutex); 1887 curr_item = __btrfs_first_delayed_insertion_item(delayed_node); 1888 while (curr_item) { 1889 btrfs_delayed_item_release_metadata(root, curr_item); 1890 prev_item = curr_item; 1891 curr_item = __btrfs_next_delayed_item(prev_item); 1892 btrfs_release_delayed_item(prev_item); 1893 } 1894 1895 curr_item = __btrfs_first_delayed_deletion_item(delayed_node); 1896 while (curr_item) { 1897 btrfs_delayed_item_release_metadata(root, curr_item); 1898 prev_item = curr_item; 1899 curr_item = __btrfs_next_delayed_item(prev_item); 1900 btrfs_release_delayed_item(prev_item); 1901 } 1902 1903 if (test_bit(BTRFS_DELAYED_NODE_DEL_IREF, &delayed_node->flags)) 1904 btrfs_release_delayed_iref(delayed_node); 1905 1906 if (test_bit(BTRFS_DELAYED_NODE_INODE_DIRTY, &delayed_node->flags)) { 1907 btrfs_delayed_inode_release_metadata(fs_info, delayed_node, false); 1908 btrfs_release_delayed_inode(delayed_node); 1909 } 1910 mutex_unlock(&delayed_node->mutex); 1911 } 1912 1913 void btrfs_kill_delayed_inode_items(struct btrfs_inode *inode) 1914 { 1915 struct btrfs_delayed_node *delayed_node; 1916 1917 delayed_node = btrfs_get_delayed_node(inode); 1918 if (!delayed_node) 1919 return; 1920 1921 __btrfs_kill_delayed_node(delayed_node); 1922 btrfs_release_delayed_node(delayed_node); 1923 } 1924 1925 void btrfs_kill_all_delayed_nodes(struct btrfs_root *root) 1926 { 1927 u64 inode_id = 0; 1928 struct btrfs_delayed_node *delayed_nodes[8]; 1929 int i, n; 1930 1931 while (1) { 1932 spin_lock(&root->inode_lock); 1933 n = radix_tree_gang_lookup(&root->delayed_nodes_tree, 1934 (void **)delayed_nodes, inode_id, 1935 ARRAY_SIZE(delayed_nodes)); 1936 if (!n) { 1937 spin_unlock(&root->inode_lock); 1938 break; 1939 } 1940 1941 inode_id = delayed_nodes[n - 1]->inode_id + 1; 1942 1943 for (i = 0; i < n; i++) 1944 refcount_inc(&delayed_nodes[i]->refs); 1945 spin_unlock(&root->inode_lock); 1946 1947 for (i = 0; i < n; i++) { 1948 __btrfs_kill_delayed_node(delayed_nodes[i]); 1949 btrfs_release_delayed_node(delayed_nodes[i]); 1950 } 1951 } 1952 } 1953 1954 void btrfs_destroy_delayed_inodes(struct btrfs_fs_info *fs_info) 1955 { 1956 struct btrfs_delayed_node *curr_node, *prev_node; 1957 1958 curr_node = btrfs_first_delayed_node(fs_info->delayed_root); 1959 while (curr_node) { 1960 __btrfs_kill_delayed_node(curr_node); 1961 1962 prev_node = curr_node; 1963 curr_node = btrfs_next_delayed_node(curr_node); 1964 btrfs_release_delayed_node(prev_node); 1965 } 1966 } 1967 1968