fs/btrfs/delayed-inode.c

   1 /*
   2  * Copyright (C) 2011 Fujitsu.  All rights reserved.
   3  * Written by Miao Xie <miaox@cn.fujitsu.com>
   4  *
   5  * This program is free software; you can redistribute it and/or
   6  * modify it under the terms of the GNU General Public
   7  * License v2 as published by the Free Software Foundation.
   8  *
   9  * This program is distributed in the hope that it will be useful,
  10  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  11  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  12  * General Public License for more details.
  13  *
  14  * You should have received a copy of the GNU General Public
  15  * License along with this program; if not, write to the
  16  * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  17  * Boston, MA 021110-1307, USA.
  18  */
  19
  20 #include <linux/slab.h>
  21 #include "delayed-inode.h"
  22 #include "disk-io.h"
  23 #include "transaction.h"
  24
  25 #define BTRFS_DELAYED_WRITEBACK         512
  26 #define BTRFS_DELAYED_BACKGROUND        128
  27 #define BTRFS_DELAYED_BATCH             16
  28
  29 static struct kmem_cache *delayed_node_cache;
  30
  31 int __init btrfs_delayed_inode_init(void)
  32 {
  33         delayed_node_cache = kmem_cache_create("btrfs_delayed_node",
  34                                         sizeof(struct btrfs_delayed_node),
  35                                         0,
  36                                         SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD,
  37                                         NULL);
  38         if (!delayed_node_cache)
  39                 return -ENOMEM;
  40         return 0;
  41 }
  42
  43 void btrfs_delayed_inode_exit(void)
  44 {
  45         if (delayed_node_cache)
  46                 kmem_cache_destroy(delayed_node_cache);
  47 }
  48
  49 static inline void btrfs_init_delayed_node(
  50                                 struct btrfs_delayed_node *delayed_node,
  51                                 struct btrfs_root *root, u64 inode_id)
  52 {
  53         delayed_node->root = root;
  54         delayed_node->inode_id = inode_id;
  55         atomic_set(&delayed_node->refs, 0);
  56         delayed_node->count = 0;
  57         delayed_node->in_list = 0;
  58         delayed_node->inode_dirty = 0;
  59         delayed_node->ins_root = RB_ROOT;
  60         delayed_node->del_root = RB_ROOT;
  61         mutex_init(&delayed_node->mutex);
  62         delayed_node->index_cnt = 0;
  63         INIT_LIST_HEAD(&delayed_node->n_list);
  64         INIT_LIST_HEAD(&delayed_node->p_list);
  65         delayed_node->bytes_reserved = 0;
  66         memset(&delayed_node->inode_item, 0, sizeof(delayed_node->inode_item));
  67 }
  68
  69 static inline int btrfs_is_continuous_delayed_item(
  70                                         struct btrfs_delayed_item *item1,
  71                                         struct btrfs_delayed_item *item2)
  72 {
  73         if (item1->key.type == BTRFS_DIR_INDEX_KEY &&
  74             item1->key.objectid == item2->key.objectid &&
  75             item1->key.type == item2->key.type &&
  76             item1->key.offset + 1 == item2->key.offset)
  77                 return 1;
  78         return 0;
  79 }
  80
  81 static inline struct btrfs_delayed_root *btrfs_get_delayed_root(
  82                                                         struct btrfs_root *root)
  83 {
  84         return root->fs_info->delayed_root;
  85 }
  86
  87 static struct btrfs_delayed_node *btrfs_get_delayed_node(struct inode *inode)
  88 {
  89         struct btrfs_inode *btrfs_inode = BTRFS_I(inode);
  90         struct btrfs_root *root = btrfs_inode->root;
  91         u64 ino = btrfs_ino(inode);
  92         struct btrfs_delayed_node *node;
  93
  94         node = ACCESS_ONCE(btrfs_inode->delayed_node);
  95         if (node) {
  96                 atomic_inc(&node->refs);
  97                 return node;
  98         }
  99
 100         spin_lock(&root->inode_lock);
 101         node = radix_tree_lookup(&root->delayed_nodes_tree, ino);
 102         if (node) {
 103                 if (btrfs_inode->delayed_node) {
 104                         atomic_inc(&node->refs);        /* can be accessed */
 105                         BUG_ON(btrfs_inode->delayed_node != node);
 106                         spin_unlock(&root->inode_lock);
 107                         return node;
 108                 }
 109                 btrfs_inode->delayed_node = node;
 110                 atomic_inc(&node->refs);        /* can be accessed */
 111                 atomic_inc(&node->refs);        /* cached in the inode */
 112                 spin_unlock(&root->inode_lock);
 113                 return node;
 114         }
 115         spin_unlock(&root->inode_lock);
 116
 117         return NULL;
 118 }
 119
 120 /* Will return either the node or PTR_ERR(-ENOMEM) */
 121 static struct btrfs_delayed_node *btrfs_get_or_create_delayed_node(
 122                                                         struct inode *inode)
 123 {
 124         struct btrfs_delayed_node *node;
 125         struct btrfs_inode *btrfs_inode = BTRFS_I(inode);
 126         struct btrfs_root *root = btrfs_inode->root;
 127         u64 ino = btrfs_ino(inode);
 128         int ret;
 129
 130 again:
 131         node = btrfs_get_delayed_node(inode);
 132         if (node)
 133                 return node;
 134
 135         node = kmem_cache_alloc(delayed_node_cache, GFP_NOFS);
 136         if (!node)
 137                 return ERR_PTR(-ENOMEM);
 138         btrfs_init_delayed_node(node, root, ino);
 139
 140         atomic_inc(&node->refs);        /* cached in the btrfs inode */
 141         atomic_inc(&node->refs);        /* can be accessed */
 142
 143         ret = radix_tree_preload(GFP_NOFS & ~__GFP_HIGHMEM);
 144         if (ret) {
 145                 kmem_cache_free(delayed_node_cache, node);
 146                 return ERR_PTR(ret);
 147         }
 148
 149         spin_lock(&root->inode_lock);
 150         ret = radix_tree_insert(&root->delayed_nodes_tree, ino, node);
 151         if (ret == -EEXIST) {
 152                 kmem_cache_free(delayed_node_cache, node);
 153                 spin_unlock(&root->inode_lock);
 154                 radix_tree_preload_end();
 155                 goto again;
 156         }
 157         btrfs_inode->delayed_node = node;
 158         spin_unlock(&root->inode_lock);
 159         radix_tree_preload_end();
 160
 161         return node;
 162 }
 163
 164 /*
 165  * Call it when holding delayed_node->mutex
 166  *
 167  * If mod = 1, add this node into the prepared list.
 168  */
 169 static void btrfs_queue_delayed_node(struct btrfs_delayed_root *root,
 170                                      struct btrfs_delayed_node *node,
 171                                      int mod)
 172 {
 173         spin_lock(&root->lock);
 174         if (node->in_list) {
 175                 if (!list_empty(&node->p_list))
 176                         list_move_tail(&node->p_list, &root->prepare_list);
 177                 else if (mod)
 178                         list_add_tail(&node->p_list, &root->prepare_list);
 179         } else {
 180                 list_add_tail(&node->n_list, &root->node_list);
 181                 list_add_tail(&node->p_list, &root->prepare_list);
 182                 atomic_inc(&node->refs);        /* inserted into list */
 183                 root->nodes++;
 184                 node->in_list = 1;
 185         }
 186         spin_unlock(&root->lock);
 187 }
 188
 189 /* Call it when holding delayed_node->mutex */
 190 static void btrfs_dequeue_delayed_node(struct btrfs_delayed_root *root,
 191                                        struct btrfs_delayed_node *node)
 192 {
 193         spin_lock(&root->lock);
 194         if (node->in_list) {
 195                 root->nodes--;
 196                 atomic_dec(&node->refs);        /* not in the list */
 197                 list_del_init(&node->n_list);
 198                 if (!list_empty(&node->p_list))
 199                         list_del_init(&node->p_list);
 200                 node->in_list = 0;
 201         }
 202         spin_unlock(&root->lock);
 203 }
 204
 205 struct btrfs_delayed_node *btrfs_first_delayed_node(
 206                         struct btrfs_delayed_root *delayed_root)
 207 {
 208         struct list_head *p;
 209         struct btrfs_delayed_node *node = NULL;
 210
 211         spin_lock(&delayed_root->lock);
 212         if (list_empty(&delayed_root->node_list))
 213                 goto out;
 214
 215         p = delayed_root->node_list.next;
 216         node = list_entry(p, struct btrfs_delayed_node, n_list);
 217         atomic_inc(&node->refs);
 218 out:
 219         spin_unlock(&delayed_root->lock);
 220
 221         return node;
 222 }
 223
 224 struct btrfs_delayed_node *btrfs_next_delayed_node(
 225                                                 struct btrfs_delayed_node *node)
 226 {
 227         struct btrfs_delayed_root *delayed_root;
 228         struct list_head *p;
 229         struct btrfs_delayed_node *next = NULL;
 230
 231         delayed_root = node->root->fs_info->delayed_root;
 232         spin_lock(&delayed_root->lock);
 233         if (!node->in_list) {   /* not in the list */
 234                 if (list_empty(&delayed_root->node_list))
 235                         goto out;
 236                 p = delayed_root->node_list.next;
 237         } else if (list_is_last(&node->n_list, &delayed_root->node_list))
 238                 goto out;
 239         else
 240                 p = node->n_list.next;
 241
 242         next = list_entry(p, struct btrfs_delayed_node, n_list);
 243         atomic_inc(&next->refs);
 244 out:
 245         spin_unlock(&delayed_root->lock);
 246
 247         return next;
 248 }
 249
 250 static void __btrfs_release_delayed_node(
 251                                 struct btrfs_delayed_node *delayed_node,
 252                                 int mod)
 253 {
 254         struct btrfs_delayed_root *delayed_root;
 255
 256         if (!delayed_node)
 257                 return;
 258
 259         delayed_root = delayed_node->root->fs_info->delayed_root;
 260
 261         mutex_lock(&delayed_node->mutex);
 262         if (delayed_node->count)
 263                 btrfs_queue_delayed_node(delayed_root, delayed_node, mod);
 264         else
 265                 btrfs_dequeue_delayed_node(delayed_root, delayed_node);
 266         mutex_unlock(&delayed_node->mutex);
 267
 268         if (atomic_dec_and_test(&delayed_node->refs)) {
 269                 struct btrfs_root *root = delayed_node->root;
 270                 spin_lock(&root->inode_lock);
 271                 if (atomic_read(&delayed_node->refs) == 0) {
 272                         radix_tree_delete(&root->delayed_nodes_tree,
 273                                           delayed_node->inode_id);
 274                         kmem_cache_free(delayed_node_cache, delayed_node);
 275                 }
 276                 spin_unlock(&root->inode_lock);
 277         }
 278 }
 279
 280 static inline void btrfs_release_delayed_node(struct btrfs_delayed_node *node)
 281 {
 282         __btrfs_release_delayed_node(node, 0);
 283 }
 284
 285 struct btrfs_delayed_node *btrfs_first_prepared_delayed_node(
 286                                         struct btrfs_delayed_root *delayed_root)
 287 {
 288         struct list_head *p;
 289         struct btrfs_delayed_node *node = NULL;
 290
 291         spin_lock(&delayed_root->lock);
 292         if (list_empty(&delayed_root->prepare_list))
 293                 goto out;
 294
 295         p = delayed_root->prepare_list.next;
 296         list_del_init(p);
 297         node = list_entry(p, struct btrfs_delayed_node, p_list);
 298         atomic_inc(&node->refs);
 299 out:
 300         spin_unlock(&delayed_root->lock);
 301
 302         return node;
 303 }
 304
 305 static inline void btrfs_release_prepared_delayed_node(
 306                                         struct btrfs_delayed_node *node)
 307 {
 308         __btrfs_release_delayed_node(node, 1);
 309 }
 310
 311 struct btrfs_delayed_item *btrfs_alloc_delayed_item(u32 data_len)
 312 {
 313         struct btrfs_delayed_item *item;
 314         item = kmalloc(sizeof(*item) + data_len, GFP_NOFS);
 315         if (item) {
 316                 item->data_len = data_len;
 317                 item->ins_or_del = 0;
 318                 item->bytes_reserved = 0;
 319                 item->delayed_node = NULL;
 320                 atomic_set(&item->refs, 1);
 321         }
 322         return item;
 323 }
 324
 325 /*
 326  * __btrfs_lookup_delayed_item - look up the delayed item by key
 327  * @delayed_node: pointer to the delayed node
 328  * @key:          the key to look up
 329  * @prev:         used to store the prev item if the right item isn't found
 330  * @next:         used to store the next item if the right item isn't found
 331  *
 332  * Note: if we don't find the right item, we will return the prev item and
 333  * the next item.
 334  */
 335 static struct btrfs_delayed_item *__btrfs_lookup_delayed_item(
 336                                 struct rb_root *root,
 337                                 struct btrfs_key *key,
 338                                 struct btrfs_delayed_item **prev,
 339                                 struct btrfs_delayed_item **next)
 340 {
 341         struct rb_node *node, *prev_node = NULL;
 342         struct btrfs_delayed_item *delayed_item = NULL;
 343         int ret = 0;
 344
 345         node = root->rb_node;
 346
 347         while (node) {
 348                 delayed_item = rb_entry(node, struct btrfs_delayed_item,
 349                                         rb_node);
 350                 prev_node = node;
 351                 ret = btrfs_comp_cpu_keys(&delayed_item->key, key);
 352                 if (ret < 0)
 353                         node = node->rb_right;
 354                 else if (ret > 0)
 355                         node = node->rb_left;
 356                 else
 357                         return delayed_item;
 358         }
 359
 360         if (prev) {
 361                 if (!prev_node)
 362                         *prev = NULL;
 363                 else if (ret < 0)
 364                         *prev = delayed_item;
 365                 else if ((node = rb_prev(prev_node)) != NULL) {
 366                         *prev = rb_entry(node, struct btrfs_delayed_item,
 367                                          rb_node);
 368                 } else
 369                         *prev = NULL;
 370         }
 371
 372         if (next) {
 373                 if (!prev_node)
 374                         *next = NULL;
 375                 else if (ret > 0)
 376                         *next = delayed_item;
 377                 else if ((node = rb_next(prev_node)) != NULL) {
 378                         *next = rb_entry(node, struct btrfs_delayed_item,
 379                                          rb_node);
 380                 } else
 381                         *next = NULL;
 382         }
 383         return NULL;
 384 }
 385
 386 struct btrfs_delayed_item *__btrfs_lookup_delayed_insertion_item(
 387                                         struct btrfs_delayed_node *delayed_node,
 388                                         struct btrfs_key *key)
 389 {
 390         struct btrfs_delayed_item *item;
 391
 392         item = __btrfs_lookup_delayed_item(&delayed_node->ins_root, key,
 393                                            NULL, NULL);
 394         return item;
 395 }
 396
 397 struct btrfs_delayed_item *__btrfs_lookup_delayed_deletion_item(
 398                                         struct btrfs_delayed_node *delayed_node,
 399                                         struct btrfs_key *key)
 400 {
 401         struct btrfs_delayed_item *item;
 402
 403         item = __btrfs_lookup_delayed_item(&delayed_node->del_root, key,
 404                                            NULL, NULL);
 405         return item;
 406 }
 407
 408 struct btrfs_delayed_item *__btrfs_search_delayed_insertion_item(
 409                                         struct btrfs_delayed_node *delayed_node,
 410                                         struct btrfs_key *key)
 411 {
 412         struct btrfs_delayed_item *item, *next;
 413
 414         item = __btrfs_lookup_delayed_item(&delayed_node->ins_root, key,
 415                                            NULL, &next);
 416         if (!item)
 417                 item = next;
 418
 419         return item;
 420 }
 421
 422 struct btrfs_delayed_item *__btrfs_search_delayed_deletion_item(
 423                                         struct btrfs_delayed_node *delayed_node,
 424                                         struct btrfs_key *key)
 425 {
 426         struct btrfs_delayed_item *item, *next;
 427
 428         item = __btrfs_lookup_delayed_item(&delayed_node->del_root, key,
 429                                            NULL, &next);
 430         if (!item)
 431                 item = next;
 432
 433         return item;
 434 }
 435
 436 static int __btrfs_add_delayed_item(struct btrfs_delayed_node *delayed_node,
 437                                     struct btrfs_delayed_item *ins,
 438                                     int action)
 439 {
 440         struct rb_node **p, *node;
 441         struct rb_node *parent_node = NULL;
 442         struct rb_root *root;
 443         struct btrfs_delayed_item *item;
 444         int cmp;
 445
 446         if (action == BTRFS_DELAYED_INSERTION_ITEM)
 447                 root = &delayed_node->ins_root;
 448         else if (action == BTRFS_DELAYED_DELETION_ITEM)
 449                 root = &delayed_node->del_root;
 450         else
 451                 BUG();
 452         p = &root->rb_node;
 453         node = &ins->rb_node;
 454
 455         while (*p) {
 456                 parent_node = *p;
 457                 item = rb_entry(parent_node, struct btrfs_delayed_item,
 458                                  rb_node);
 459
 460                 cmp = btrfs_comp_cpu_keys(&item->key, &ins->key);
 461                 if (cmp < 0)
 462                         p = &(*p)->rb_right;
 463                 else if (cmp > 0)
 464                         p = &(*p)->rb_left;
 465                 else
 466                         return -EEXIST;
 467         }
 468
 469         rb_link_node(node, parent_node, p);
 470         rb_insert_color(node, root);
 471         ins->delayed_node = delayed_node;
 472         ins->ins_or_del = action;
 473
 474         if (ins->key.type == BTRFS_DIR_INDEX_KEY &&
 475             action == BTRFS_DELAYED_INSERTION_ITEM &&
 476             ins->key.offset >= delayed_node->index_cnt)
 477                         delayed_node->index_cnt = ins->key.offset + 1;
 478
 479         delayed_node->count++;
 480         atomic_inc(&delayed_node->root->fs_info->delayed_root->items);
 481         return 0;
 482 }
 483
 484 static int __btrfs_add_delayed_insertion_item(struct btrfs_delayed_node *node,
 485                                               struct btrfs_delayed_item *item)
 486 {
 487         return __btrfs_add_delayed_item(node, item,
 488                                         BTRFS_DELAYED_INSERTION_ITEM);
 489 }
 490
 491 static int __btrfs_add_delayed_deletion_item(struct btrfs_delayed_node *node,
 492                                              struct btrfs_delayed_item *item)
 493 {
 494         return __btrfs_add_delayed_item(node, item,
 495                                         BTRFS_DELAYED_DELETION_ITEM);
 496 }
 497
 498 static void finish_one_item(struct btrfs_delayed_root *delayed_root)
 499 {
 500         int seq = atomic_inc_return(&delayed_root->items_seq);
 501         if ((atomic_dec_return(&delayed_root->items) <
 502             BTRFS_DELAYED_BACKGROUND || seq % BTRFS_DELAYED_BATCH == 0) &&
 503             waitqueue_active(&delayed_root->wait))
 504                 wake_up(&delayed_root->wait);
 505 }
 506
 507 static void __btrfs_remove_delayed_item(struct btrfs_delayed_item *delayed_item)
 508 {
 509         struct rb_root *root;
 510         struct btrfs_delayed_root *delayed_root;
 511
 512         delayed_root = delayed_item->delayed_node->root->fs_info->delayed_root;
 513
 514         BUG_ON(!delayed_root);
 515         BUG_ON(delayed_item->ins_or_del != BTRFS_DELAYED_DELETION_ITEM &&
 516                delayed_item->ins_or_del != BTRFS_DELAYED_INSERTION_ITEM);
 517
 518         if (delayed_item->ins_or_del == BTRFS_DELAYED_INSERTION_ITEM)
 519                 root = &delayed_item->delayed_node->ins_root;
 520         else
 521                 root = &delayed_item->delayed_node->del_root;
 522
 523         rb_erase(&delayed_item->rb_node, root);
 524         delayed_item->delayed_node->count--;
 525
 526         finish_one_item(delayed_root);
 527 }
 528
 529 static void btrfs_release_delayed_item(struct btrfs_delayed_item *item)
 530 {
 531         if (item) {
 532                 __btrfs_remove_delayed_item(item);
 533                 if (atomic_dec_and_test(&item->refs))
 534                         kfree(item);
 535         }
 536 }
 537
 538 struct btrfs_delayed_item *__btrfs_first_delayed_insertion_item(
 539                                         struct btrfs_delayed_node *delayed_node)
 540 {
 541         struct rb_node *p;
 542         struct btrfs_delayed_item *item = NULL;
 543
 544         p = rb_first(&delayed_node->ins_root);
 545         if (p)
 546                 item = rb_entry(p, struct btrfs_delayed_item, rb_node);
 547
 548         return item;
 549 }
 550
 551 struct btrfs_delayed_item *__btrfs_first_delayed_deletion_item(
 552                                         struct btrfs_delayed_node *delayed_node)
 553 {
 554         struct rb_node *p;
 555         struct btrfs_delayed_item *item = NULL;
 556
 557         p = rb_first(&delayed_node->del_root);
 558         if (p)
 559                 item = rb_entry(p, struct btrfs_delayed_item, rb_node);
 560
 561         return item;
 562 }
 563
 564 struct btrfs_delayed_item *__btrfs_next_delayed_item(
 565                                                 struct btrfs_delayed_item *item)
 566 {
 567         struct rb_node *p;
 568         struct btrfs_delayed_item *next = NULL;
 569
 570         p = rb_next(&item->rb_node);
 571         if (p)
 572                 next = rb_entry(p, struct btrfs_delayed_item, rb_node);
 573
 574         return next;
 575 }
 576
 577 static inline struct btrfs_root *btrfs_get_fs_root(struct btrfs_root *root,
 578                                                    u64 root_id)
 579 {
 580         struct btrfs_key root_key;
 581
 582         if (root->objectid == root_id)
 583                 return root;
 584
 585         root_key.objectid = root_id;
 586         root_key.type = BTRFS_ROOT_ITEM_KEY;
 587         root_key.offset = (u64)-1;
 588         return btrfs_read_fs_root_no_name(root->fs_info, &root_key);
 589 }
 590
 591 static int btrfs_delayed_item_reserve_metadata(struct btrfs_trans_handle *trans,
 592                                                struct btrfs_root *root,
 593                                                struct btrfs_delayed_item *item)
 594 {
 595         struct btrfs_block_rsv *src_rsv;
 596         struct btrfs_block_rsv *dst_rsv;
 597         u64 num_bytes;
 598         int ret;
 599
 600         if (!trans->bytes_reserved)
 601                 return 0;
 602
 603         src_rsv = trans->block_rsv;
 604         dst_rsv = &root->fs_info->delayed_block_rsv;
 605
 606         num_bytes = btrfs_calc_trans_metadata_size(root, 1);
 607         ret = btrfs_block_rsv_migrate(src_rsv, dst_rsv, num_bytes);
 608         if (!ret) {
 609                 trace_btrfs_space_reservation(root->fs_info, "delayed_item",
 610                                               item->key.objectid,
 611                                               num_bytes, 1);
 612                 item->bytes_reserved = num_bytes;
 613         }
 614
 615         return ret;
 616 }
 617
 618 static void btrfs_delayed_item_release_metadata(struct btrfs_root *root,
 619                                                 struct btrfs_delayed_item *item)
 620 {
 621         struct btrfs_block_rsv *rsv;
 622
 623         if (!item->bytes_reserved)
 624                 return;
 625
 626         rsv = &root->fs_info->delayed_block_rsv;
 627         trace_btrfs_space_reservation(root->fs_info, "delayed_item",
 628                                       item->key.objectid, item->bytes_reserved,
 629                                       0);
 630         btrfs_block_rsv_release(root, rsv,
 631                                 item->bytes_reserved);
 632 }
 633
 634 static int btrfs_delayed_inode_reserve_metadata(
 635                                         struct btrfs_trans_handle *trans,
 636                                         struct btrfs_root *root,
 637                                         struct inode *inode,
 638                                         struct btrfs_delayed_node *node)
 639 {
 640         struct btrfs_block_rsv *src_rsv;
 641         struct btrfs_block_rsv *dst_rsv;
 642         u64 num_bytes;
 643         int ret;
 644         bool release = false;
 645
 646         src_rsv = trans->block_rsv;
 647         dst_rsv = &root->fs_info->delayed_block_rsv;
 648
 649         num_bytes = btrfs_calc_trans_metadata_size(root, 1);
 650
 651         /*
 652          * btrfs_dirty_inode will update the inode under btrfs_join_transaction
 653          * which doesn't reserve space for speed.  This is a problem since we
 654          * still need to reserve space for this update, so try to reserve the
 655          * space.
 656          *
 657          * Now if src_rsv == delalloc_block_rsv we'll let it just steal since
 658          * we're accounted for.
 659          */
 660         if (!src_rsv || (!trans->bytes_reserved &&
 661                          src_rsv->type != BTRFS_BLOCK_RSV_DELALLOC)) {
 662                 ret = btrfs_block_rsv_add(root, dst_rsv, num_bytes,
 663                                           BTRFS_RESERVE_NO_FLUSH);
 664                 /*
 665                  * Since we're under a transaction reserve_metadata_bytes could
 666                  * try to commit the transaction which will make it return
 667                  * EAGAIN to make us stop the transaction we have, so return
 668                  * ENOSPC instead so that btrfs_dirty_inode knows what to do.
 669                  */
 670                 if (ret == -EAGAIN)
 671                         ret = -ENOSPC;
 672                 if (!ret) {
 673                         node->bytes_reserved = num_bytes;
 674                         trace_btrfs_space_reservation(root->fs_info,
 675                                                       "delayed_inode",
 676                                                       btrfs_ino(inode),
 677                                                       num_bytes, 1);
 678                 }
 679                 return ret;
 680         } else if (src_rsv->type == BTRFS_BLOCK_RSV_DELALLOC) {
 681                 spin_lock(&BTRFS_I(inode)->lock);
 682                 if (test_and_clear_bit(BTRFS_INODE_DELALLOC_META_RESERVED,
 683                                        &BTRFS_I(inode)->runtime_flags)) {
 684                         spin_unlock(&BTRFS_I(inode)->lock);
 685                         release = true;
 686                         goto migrate;
 687                 }
 688                 spin_unlock(&BTRFS_I(inode)->lock);
 689
 690                 /* Ok we didn't have space pre-reserved.  This shouldn't happen
 691                  * too often but it can happen if we do delalloc to an existing
 692                  * inode which gets dirtied because of the time update, and then
 693                  * isn't touched again until after the transaction commits and
 694                  * then we try to write out the data.  First try to be nice and
 695                  * reserve something strictly for us.  If not be a pain and try
 696                  * to steal from the delalloc block rsv.
 697                  */
 698                 ret = btrfs_block_rsv_add(root, dst_rsv, num_bytes,
 699                                           BTRFS_RESERVE_NO_FLUSH);
 700                 if (!ret)
 701                         goto out;
 702
 703                 ret = btrfs_block_rsv_migrate(src_rsv, dst_rsv, num_bytes);
 704                 if (!ret)
 705                         goto out;
 706
 707                 /*
 708                  * Ok this is a problem, let's just steal from the global rsv
 709                  * since this really shouldn't happen that often.
 710                  */
 711                 WARN_ON(1);
 712                 ret = btrfs_block_rsv_migrate(&root->fs_info->global_block_rsv,
 713                                               dst_rsv, num_bytes);
 714                 goto out;
 715         }
 716
 717 migrate:
 718         ret = btrfs_block_rsv_migrate(src_rsv, dst_rsv, num_bytes);
 719
 720 out:
 721         /*
 722          * Migrate only takes a reservation, it doesn't touch the size of the
 723          * block_rsv.  This is to simplify people who don't normally have things
 724          * migrated from their block rsv.  If they go to release their
 725          * reservation, that will decrease the size as well, so if migrate
 726          * reduced size we'd end up with a negative size.  But for the
 727          * delalloc_meta_reserved stuff we will only know to drop 1 reservation,
 728          * but we could in fact do this reserve/migrate dance several times
 729          * between the time we did the original reservation and we'd clean it
 730          * up.  So to take care of this, release the space for the meta
 731          * reservation here.  I think it may be time for a documentation page on
 732          * how block rsvs. work.
 733          */
 734         if (!ret) {
 735                 trace_btrfs_space_reservation(root->fs_info, "delayed_inode",
 736                                               btrfs_ino(inode), num_bytes, 1);
 737                 node->bytes_reserved = num_bytes;
 738         }
 739
 740         if (release) {
 741                 trace_btrfs_space_reservation(root->fs_info, "delalloc",
 742                                               btrfs_ino(inode), num_bytes, 0);
 743                 btrfs_block_rsv_release(root, src_rsv, num_bytes);
 744         }
 745
 746         return ret;
 747 }
 748
 749 static void btrfs_delayed_inode_release_metadata(struct btrfs_root *root,
 750                                                 struct btrfs_delayed_node *node)
 751 {
 752         struct btrfs_block_rsv *rsv;
 753
 754         if (!node->bytes_reserved)
 755                 return;
 756
 757         rsv = &root->fs_info->delayed_block_rsv;
 758         trace_btrfs_space_reservation(root->fs_info, "delayed_inode",
 759                                       node->inode_id, node->bytes_reserved, 0);
 760         btrfs_block_rsv_release(root, rsv,
 761                                 node->bytes_reserved);
 762         node->bytes_reserved = 0;
 763 }
 764
 765 /*
 766  * This helper will insert some continuous items into the same leaf according
 767  * to the free space of the leaf.
 768  */
 769 static int btrfs_batch_insert_items(struct btrfs_trans_handle *trans,
 770                                 struct btrfs_root *root,
 771                                 struct btrfs_path *path,
 772                                 struct btrfs_delayed_item *item)
 773 {
 774         struct btrfs_delayed_item *curr, *next;
 775         int free_space;
 776         int total_data_size = 0, total_size = 0;
 777         struct extent_buffer *leaf;
 778         char *data_ptr;
 779         struct btrfs_key *keys;
 780         u32 *data_size;
 781         struct list_head head;
 782         int slot;
 783         int nitems;
 784         int i;
 785         int ret = 0;
 786
 787         BUG_ON(!path->nodes[0]);
 788
 789         leaf = path->nodes[0];
 790         free_space = btrfs_leaf_free_space(root, leaf);
 791         INIT_LIST_HEAD(&head);
 792
 793         next = item;
 794         nitems = 0;
 795
 796         /*
 797          * count the number of the continuous items that we can insert in batch
 798          */
 799         while (total_size + next->data_len + sizeof(struct btrfs_item) <=
 800                free_space) {
 801                 total_data_size += next->data_len;
 802                 total_size += next->data_len + sizeof(struct btrfs_item);
 803                 list_add_tail(&next->tree_list, &head);
 804                 nitems++;
 805
 806                 curr = next;
 807                 next = __btrfs_next_delayed_item(curr);
 808                 if (!next)
 809                         break;
 810
 811                 if (!btrfs_is_continuous_delayed_item(curr, next))
 812                         break;
 813         }
 814
 815         if (!nitems) {
 816                 ret = 0;
 817                 goto out;
 818         }
 819
 820         /*
 821          * we need allocate some memory space, but it might cause the task
 822          * to sleep, so we set all locked nodes in the path to blocking locks
 823          * first.
 824          */
 825         btrfs_set_path_blocking(path);
 826
 827         keys = kmalloc(sizeof(struct btrfs_key) * nitems, GFP_NOFS);
 828         if (!keys) {
 829                 ret = -ENOMEM;
 830                 goto out;
 831         }
 832
 833         data_size = kmalloc(sizeof(u32) * nitems, GFP_NOFS);
 834         if (!data_size) {
 835                 ret = -ENOMEM;
 836                 goto error;
 837         }
 838
 839         /* get keys of all the delayed items */
 840         i = 0;
 841         list_for_each_entry(next, &head, tree_list) {
 842                 keys[i] = next->key;
 843                 data_size[i] = next->data_len;
 844                 i++;
 845         }
 846
 847         /* reset all the locked nodes in the patch to spinning locks. */
 848         btrfs_clear_path_blocking(path, NULL, 0);
 849
 850         /* insert the keys of the items */
 851         setup_items_for_insert(trans, root, path, keys, data_size,
 852                                total_data_size, total_size, nitems);
 853
 854         /* insert the dir index items */
 855         slot = path->slots[0];
 856         list_for_each_entry_safe(curr, next, &head, tree_list) {
 857                 data_ptr = btrfs_item_ptr(leaf, slot, char);
 858                 write_extent_buffer(leaf, &curr->data,
 859                                     (unsigned long)data_ptr,
 860                                     curr->data_len);
 861                 slot++;
 862
 863                 btrfs_delayed_item_release_metadata(root, curr);
 864
 865                 list_del(&curr->tree_list);
 866                 btrfs_release_delayed_item(curr);
 867         }
 868
 869 error:
 870         kfree(data_size);
 871         kfree(keys);
 872 out:
 873         return ret;
 874 }
 875
 876 /*
 877  * This helper can just do simple insertion that needn't extend item for new
 878  * data, such as directory name index insertion, inode insertion.
 879  */
 880 static int btrfs_insert_delayed_item(struct btrfs_trans_handle *trans,
 881                                      struct btrfs_root *root,
 882                                      struct btrfs_path *path,
 883                                      struct btrfs_delayed_item *delayed_item)
 884 {
 885         struct extent_buffer *leaf;
 886         char *ptr;
 887         int ret;
 888
 889         ret = btrfs_insert_empty_item(trans, root, path, &delayed_item->key,
 890                                       delayed_item->data_len);
 891         if (ret < 0 && ret != -EEXIST)
 892                 return ret;
 893
 894         leaf = path->nodes[0];
 895
 896         ptr = btrfs_item_ptr(leaf, path->slots[0], char);
 897
 898         write_extent_buffer(leaf, delayed_item->data, (unsigned long)ptr,
 899                             delayed_item->data_len);
 900         btrfs_mark_buffer_dirty(leaf);
 901
 902         btrfs_delayed_item_release_metadata(root, delayed_item);
 903         return 0;
 904 }
 905
 906 /*
 907  * we insert an item first, then if there are some continuous items, we try
 908  * to insert those items into the same leaf.
 909  */
 910 static int btrfs_insert_delayed_items(struct btrfs_trans_handle *trans,
 911                                       struct btrfs_path *path,
 912                                       struct btrfs_root *root,
 913                                       struct btrfs_delayed_node *node)
 914 {
 915         struct btrfs_delayed_item *curr, *prev;
 916         int ret = 0;
 917
 918 do_again:
 919         mutex_lock(&node->mutex);
 920         curr = __btrfs_first_delayed_insertion_item(node);
 921         if (!curr)
 922                 goto insert_end;
 923
 924         ret = btrfs_insert_delayed_item(trans, root, path, curr);
 925         if (ret < 0) {
 926                 btrfs_release_path(path);
 927                 goto insert_end;
 928         }
 929
 930         prev = curr;
 931         curr = __btrfs_next_delayed_item(prev);
 932         if (curr && btrfs_is_continuous_delayed_item(prev, curr)) {
 933                 /* insert the continuous items into the same leaf */
 934                 path->slots[0]++;
 935                 btrfs_batch_insert_items(trans, root, path, curr);
 936         }
 937         btrfs_release_delayed_item(prev);
 938         btrfs_mark_buffer_dirty(path->nodes[0]);
 939
 940         btrfs_release_path(path);
 941         mutex_unlock(&node->mutex);
 942         goto do_again;
 943
 944 insert_end:
 945         mutex_unlock(&node->mutex);
 946         return ret;
 947 }
 948
 949 static int btrfs_batch_delete_items(struct btrfs_trans_handle *trans,
 950                                     struct btrfs_root *root,
 951                                     struct btrfs_path *path,
 952                                     struct btrfs_delayed_item *item)
 953 {
 954         struct btrfs_delayed_item *curr, *next;
 955         struct extent_buffer *leaf;
 956         struct btrfs_key key;
 957         struct list_head head;
 958         int nitems, i, last_item;
 959         int ret = 0;
 960
 961         BUG_ON(!path->nodes[0]);
 962
 963         leaf = path->nodes[0];
 964
 965         i = path->slots[0];
 966         last_item = btrfs_header_nritems(leaf) - 1;
 967         if (i > last_item)
 968                 return -ENOENT; /* FIXME: Is errno suitable? */
 969
 970         next = item;
 971         INIT_LIST_HEAD(&head);
 972         btrfs_item_key_to_cpu(leaf, &key, i);
 973         nitems = 0;
 974         /*
 975          * count the number of the dir index items that we can delete in batch
 976          */
 977         while (btrfs_comp_cpu_keys(&next->key, &key) == 0) {
 978                 list_add_tail(&next->tree_list, &head);
 979                 nitems++;
 980
 981                 curr = next;
 982                 next = __btrfs_next_delayed_item(curr);
 983                 if (!next)
 984                         break;
 985
 986                 if (!btrfs_is_continuous_delayed_item(curr, next))
 987                         break;
 988
 989                 i++;
 990                 if (i > last_item)
 991                         break;
 992                 btrfs_item_key_to_cpu(leaf, &key, i);
 993         }
 994
 995         if (!nitems)
 996                 return 0;
 997
 998         ret = btrfs_del_items(trans, root, path, path->slots[0], nitems);
 999         if (ret)
1000                 goto out;
1001
1002         list_for_each_entry_safe(curr, next, &head, tree_list) {
1003                 btrfs_delayed_item_release_metadata(root, curr);
1004                 list_del(&curr->tree_list);
1005                 btrfs_release_delayed_item(curr);
1006         }
1007
1008 out:
1009         return ret;
1010 }
1011
1012 static int btrfs_delete_delayed_items(struct btrfs_trans_handle *trans,
1013                                       struct btrfs_path *path,
1014                                       struct btrfs_root *root,
1015                                       struct btrfs_delayed_node *node)
1016 {
1017         struct btrfs_delayed_item *curr, *prev;
1018         int ret = 0;
1019
1020 do_again:
1021         mutex_lock(&node->mutex);
1022         curr = __btrfs_first_delayed_deletion_item(node);
1023         if (!curr)
1024                 goto delete_fail;
1025
1026         ret = btrfs_search_slot(trans, root, &curr->key, path, -1, 1);
1027         if (ret < 0)
1028                 goto delete_fail;
1029         else if (ret > 0) {
1030                 /*
1031                  * can't find the item which the node points to, so this node
1032                  * is invalid, just drop it.
1033                  */
1034                 prev = curr;
1035                 curr = __btrfs_next_delayed_item(prev);
1036                 btrfs_release_delayed_item(prev);
1037                 ret = 0;
1038                 btrfs_release_path(path);
1039                 if (curr) {
1040                         mutex_unlock(&node->mutex);
1041                         goto do_again;
1042                 } else
1043                         goto delete_fail;
1044         }
1045
1046         btrfs_batch_delete_items(trans, root, path, curr);
1047         btrfs_release_path(path);
1048         mutex_unlock(&node->mutex);
1049         goto do_again;
1050
1051 delete_fail:
1052         btrfs_release_path(path);
1053         mutex_unlock(&node->mutex);
1054         return ret;
1055 }
1056
1057 static void btrfs_release_delayed_inode(struct btrfs_delayed_node *delayed_node)
1058 {
1059         struct btrfs_delayed_root *delayed_root;
1060
1061         if (delayed_node && delayed_node->inode_dirty) {
1062                 BUG_ON(!delayed_node->root);
1063                 delayed_node->inode_dirty = 0;
1064                 delayed_node->count--;
1065
1066                 delayed_root = delayed_node->root->fs_info->delayed_root;
1067                 finish_one_item(delayed_root);
1068         }
1069 }
1070
1071 static int __btrfs_update_delayed_inode(struct btrfs_trans_handle *trans,
1072                                         struct btrfs_root *root,
1073                                         struct btrfs_path *path,
1074                                         struct btrfs_delayed_node *node)
1075 {
1076         struct btrfs_key key;
1077         struct btrfs_inode_item *inode_item;
1078         struct extent_buffer *leaf;
1079         int ret;
1080
1081         key.objectid = node->inode_id;
1082         btrfs_set_key_type(&key, BTRFS_INODE_ITEM_KEY);
1083         key.offset = 0;
1084
1085         ret = btrfs_lookup_inode(trans, root, path, &key, 1);
1086         if (ret > 0) {
1087                 btrfs_release_path(path);
1088                 return -ENOENT;
1089         } else if (ret < 0) {
1090                 return ret;
1091         }
1092
1093         btrfs_unlock_up_safe(path, 1);
1094         leaf = path->nodes[0];
1095         inode_item = btrfs_item_ptr(leaf, path->slots[0],
1096                                     struct btrfs_inode_item);
1097         write_extent_buffer(leaf, &node->inode_item, (unsigned long)inode_item,
1098                             sizeof(struct btrfs_inode_item));
1099         btrfs_mark_buffer_dirty(leaf);
1100         btrfs_release_path(path);
1101
1102         btrfs_delayed_inode_release_metadata(root, node);
1103         btrfs_release_delayed_inode(node);
1104
1105         return 0;
1106 }
1107
1108 static inline int btrfs_update_delayed_inode(struct btrfs_trans_handle *trans,
1109                                              struct btrfs_root *root,
1110                                              struct btrfs_path *path,
1111                                              struct btrfs_delayed_node *node)
1112 {
1113         int ret;
1114
1115         mutex_lock(&node->mutex);
1116         if (!node->inode_dirty) {
1117                 mutex_unlock(&node->mutex);
1118                 return 0;
1119         }
1120
1121         ret = __btrfs_update_delayed_inode(trans, root, path, node);
1122         mutex_unlock(&node->mutex);
1123         return ret;
1124 }
1125
1126 static inline int
1127 __btrfs_commit_inode_delayed_items(struct btrfs_trans_handle *trans,
1128                                    struct btrfs_path *path,
1129                                    struct btrfs_delayed_node *node)
1130 {
1131         int ret;
1132
1133         ret = btrfs_insert_delayed_items(trans, path, node->root, node);
1134         if (ret)
1135                 return ret;
1136
1137         ret = btrfs_delete_delayed_items(trans, path, node->root, node);
1138         if (ret)
1139                 return ret;
1140
1141         ret = btrfs_update_delayed_inode(trans, node->root, path, node);
1142         return ret;
1143 }
1144
1145 /*
1146  * Called when committing the transaction.
1147  * Returns 0 on success.
1148  * Returns < 0 on error and returns with an aborted transaction with any
1149  * outstanding delayed items cleaned up.
1150  */
1151 static int __btrfs_run_delayed_items(struct btrfs_trans_handle *trans,
1152                                      struct btrfs_root *root, int nr)
1153 {
1154         struct btrfs_delayed_root *delayed_root;
1155         struct btrfs_delayed_node *curr_node, *prev_node;
1156         struct btrfs_path *path;
1157         struct btrfs_block_rsv *block_rsv;
1158         int ret = 0;
1159         bool count = (nr > 0);
1160
1161         if (trans->aborted)
1162                 return -EIO;
1163
1164         path = btrfs_alloc_path();
1165         if (!path)
1166                 return -ENOMEM;
1167         path->leave_spinning = 1;
1168
1169         block_rsv = trans->block_rsv;
1170         trans->block_rsv = &root->fs_info->delayed_block_rsv;
1171
1172         delayed_root = btrfs_get_delayed_root(root);
1173
1174         curr_node = btrfs_first_delayed_node(delayed_root);
1175         while (curr_node && (!count || (count && nr--))) {
1176                 ret = __btrfs_commit_inode_delayed_items(trans, path,
1177                                                          curr_node);
1178                 if (ret) {
1179                         btrfs_release_delayed_node(curr_node);
1180                         curr_node = NULL;
1181                         btrfs_abort_transaction(trans, root, ret);
1182                         break;
1183                 }
1184
1185                 prev_node = curr_node;
1186                 curr_node = btrfs_next_delayed_node(curr_node);
1187                 btrfs_release_delayed_node(prev_node);
1188         }
1189
1190         if (curr_node)
1191                 btrfs_release_delayed_node(curr_node);
1192         btrfs_free_path(path);
1193         trans->block_rsv = block_rsv;
1194
1195         return ret;
1196 }
1197
1198 int btrfs_run_delayed_items(struct btrfs_trans_handle *trans,
1199                             struct btrfs_root *root)
1200 {
1201         return __btrfs_run_delayed_items(trans, root, -1);
1202 }
1203
1204 int btrfs_run_delayed_items_nr(struct btrfs_trans_handle *trans,
1205                                struct btrfs_root *root, int nr)
1206 {
1207         return __btrfs_run_delayed_items(trans, root, nr);
1208 }
1209
1210 int btrfs_commit_inode_delayed_items(struct btrfs_trans_handle *trans,
1211                                      struct inode *inode)
1212 {
1213         struct btrfs_delayed_node *delayed_node = btrfs_get_delayed_node(inode);
1214         struct btrfs_path *path;
1215         struct btrfs_block_rsv *block_rsv;
1216         int ret;
1217
1218         if (!delayed_node)
1219                 return 0;
1220
1221         mutex_lock(&delayed_node->mutex);
1222         if (!delayed_node->count) {
1223                 mutex_unlock(&delayed_node->mutex);
1224                 btrfs_release_delayed_node(delayed_node);
1225                 return 0;
1226         }
1227         mutex_unlock(&delayed_node->mutex);
1228
1229         path = btrfs_alloc_path();
1230         if (!path)
1231                 return -ENOMEM;
1232         path->leave_spinning = 1;
1233
1234         block_rsv = trans->block_rsv;
1235         trans->block_rsv = &delayed_node->root->fs_info->delayed_block_rsv;
1236
1237         ret = __btrfs_commit_inode_delayed_items(trans, path, delayed_node);
1238
1239         btrfs_release_delayed_node(delayed_node);
1240         btrfs_free_path(path);
1241         trans->block_rsv = block_rsv;
1242
1243         return ret;
1244 }
1245
1246 int btrfs_commit_inode_delayed_inode(struct inode *inode)
1247 {
1248         struct btrfs_trans_handle *trans;
1249         struct btrfs_delayed_node *delayed_node = btrfs_get_delayed_node(inode);
1250         struct btrfs_path *path;
1251         struct btrfs_block_rsv *block_rsv;
1252         int ret;
1253
1254         if (!delayed_node)
1255                 return 0;
1256
1257         mutex_lock(&delayed_node->mutex);
1258         if (!delayed_node->inode_dirty) {
1259                 mutex_unlock(&delayed_node->mutex);
1260                 btrfs_release_delayed_node(delayed_node);
1261                 return 0;
1262         }
1263         mutex_unlock(&delayed_node->mutex);
1264
1265         trans = btrfs_join_transaction(delayed_node->root);
1266         if (IS_ERR(trans)) {
1267                 ret = PTR_ERR(trans);
1268                 goto out;
1269         }
1270
1271         path = btrfs_alloc_path();
1272         if (!path) {
1273                 ret = -ENOMEM;
1274                 goto trans_out;
1275         }
1276         path->leave_spinning = 1;
1277
1278         block_rsv = trans->block_rsv;
1279         trans->block_rsv = &delayed_node->root->fs_info->delayed_block_rsv;
1280
1281         mutex_lock(&delayed_node->mutex);
1282         if (delayed_node->inode_dirty)
1283                 ret = __btrfs_update_delayed_inode(trans, delayed_node->root,
1284                                                    path, delayed_node);
1285         else
1286                 ret = 0;
1287         mutex_unlock(&delayed_node->mutex);
1288
1289         btrfs_free_path(path);
1290         trans->block_rsv = block_rsv;
1291 trans_out:
1292         btrfs_end_transaction(trans, delayed_node->root);
1293         btrfs_btree_balance_dirty(delayed_node->root);
1294 out:
1295         btrfs_release_delayed_node(delayed_node);
1296
1297         return ret;
1298 }
1299
1300 void btrfs_remove_delayed_node(struct inode *inode)
1301 {
1302         struct btrfs_delayed_node *delayed_node;
1303
1304         delayed_node = ACCESS_ONCE(BTRFS_I(inode)->delayed_node);
1305         if (!delayed_node)
1306                 return;
1307
1308         BTRFS_I(inode)->delayed_node = NULL;
1309         btrfs_release_delayed_node(delayed_node);
1310 }
1311
1312 struct btrfs_async_delayed_work {
1313         struct btrfs_delayed_root *delayed_root;
1314         int nr;
1315         struct btrfs_work work;
1316 };
1317
1318 static void btrfs_async_run_delayed_root(struct btrfs_work *work)
1319 {
1320         struct btrfs_async_delayed_work *async_work;
1321         struct btrfs_delayed_root *delayed_root;
1322         struct btrfs_trans_handle *trans;
1323         struct btrfs_path *path;
1324         struct btrfs_delayed_node *delayed_node = NULL;
1325         struct btrfs_root *root;
1326         struct btrfs_block_rsv *block_rsv;
1327         int total_done = 0;
1328
1329         async_work = container_of(work, struct btrfs_async_delayed_work, work);
1330         delayed_root = async_work->delayed_root;
1331
1332         path = btrfs_alloc_path();
1333         if (!path)
1334                 goto out;
1335
1336 again:
1337         if (atomic_read(&delayed_root->items) < BTRFS_DELAYED_BACKGROUND / 2)
1338                 goto free_path;
1339
1340         delayed_node = btrfs_first_prepared_delayed_node(delayed_root);
1341         if (!delayed_node)
1342                 goto free_path;
1343
1344         path->leave_spinning = 1;
1345         root = delayed_node->root;
1346
1347         trans = btrfs_join_transaction(root);
1348         if (IS_ERR(trans))
1349                 goto release_path;
1350
1351         block_rsv = trans->block_rsv;
1352         trans->block_rsv = &root->fs_info->delayed_block_rsv;
1353
1354         __btrfs_commit_inode_delayed_items(trans, path, delayed_node);
1355         /*
1356          * Maybe new delayed items have been inserted, so we need requeue
1357          * the work. Besides that, we must dequeue the empty delayed nodes
1358          * to avoid the race between delayed items balance and the worker.
1359          * The race like this:
1360          *      Task1                           Worker thread
1361          *                                      count == 0, needn't requeue
1362          *                                        also needn't insert the
1363          *                                        delayed node into prepare
1364          *                                        list again.
1365          *      add lots of delayed items
1366          *      queue the delayed node
1367          *        already in the list,
1368          *        and not in the prepare
1369          *        list, it means the delayed
1370          *        node is being dealt with
1371          *        by the worker.
1372          *      do delayed items balance
1373          *        the delayed node is being
1374          *        dealt with by the worker
1375          *        now, just wait.
1376          *                                      the worker goto idle.
1377          * Task1 will sleep until the transaction is commited.
1378          */
1379         mutex_lock(&delayed_node->mutex);
1380         btrfs_dequeue_delayed_node(root->fs_info->delayed_root, delayed_node);
1381         mutex_unlock(&delayed_node->mutex);
1382
1383         trans->block_rsv = block_rsv;
1384         btrfs_end_transaction_dmeta(trans, root);
1385         btrfs_btree_balance_dirty_nodelay(root);
1386
1387 release_path:
1388         btrfs_release_path(path);
1389         total_done++;
1390
1391         btrfs_release_prepared_delayed_node(delayed_node);
1392         if (async_work->nr == 0 || total_done < async_work->nr)
1393                 goto again;
1394
1395 free_path:
1396         btrfs_free_path(path);
1397 out:
1398         wake_up(&delayed_root->wait);
1399         kfree(async_work);
1400 }
1401
1402
1403 static int btrfs_wq_run_delayed_node(struct btrfs_delayed_root *delayed_root,
1404                                      struct btrfs_root *root, int nr)
1405 {
1406         struct btrfs_async_delayed_work *async_work;
1407
1408         if (atomic_read(&delayed_root->items) < BTRFS_DELAYED_BACKGROUND)
1409                 return 0;
1410
1411         async_work = kmalloc(sizeof(*async_work), GFP_NOFS);
1412         if (!async_work)
1413                 return -ENOMEM;
1414
1415         async_work->delayed_root = delayed_root;
1416         async_work->work.func = btrfs_async_run_delayed_root;
1417         async_work->work.flags = 0;
1418         async_work->nr = nr;
1419
1420         btrfs_queue_worker(&root->fs_info->delayed_workers, &async_work->work);
1421         return 0;
1422 }
1423
1424 void btrfs_assert_delayed_root_empty(struct btrfs_root *root)
1425 {
1426         struct btrfs_delayed_root *delayed_root;
1427         delayed_root = btrfs_get_delayed_root(root);
1428         WARN_ON(btrfs_first_delayed_node(delayed_root));
1429 }
1430
1431 static int refs_newer(struct btrfs_delayed_root *delayed_root,
1432                       int seq, int count)
1433 {
1434         int val = atomic_read(&delayed_root->items_seq);
1435
1436         if (val < seq || val >= seq + count)
1437                 return 1;
1438         return 0;
1439 }
1440
1441 void btrfs_balance_delayed_items(struct btrfs_root *root)
1442 {
1443         struct btrfs_delayed_root *delayed_root;
1444         int seq;
1445
1446         delayed_root = btrfs_get_delayed_root(root);
1447
1448         if (atomic_read(&delayed_root->items) < BTRFS_DELAYED_BACKGROUND)
1449                 return;
1450
1451         seq = atomic_read(&delayed_root->items_seq);
1452
1453         if (atomic_read(&delayed_root->items) >= BTRFS_DELAYED_WRITEBACK) {
1454                 int ret;
1455                 DEFINE_WAIT(__wait);
1456
1457                 ret = btrfs_wq_run_delayed_node(delayed_root, root, 0);
1458                 if (ret)
1459                         return;
1460
1461                 while (1) {
1462                         prepare_to_wait(&delayed_root->wait, &__wait,
1463                                         TASK_INTERRUPTIBLE);
1464
1465                         if (refs_newer(delayed_root, seq,
1466                                        BTRFS_DELAYED_BATCH) ||
1467                             atomic_read(&delayed_root->items) <
1468                             BTRFS_DELAYED_BACKGROUND) {
1469                                 break;
1470                         }
1471                         if (!signal_pending(current))
1472                                 schedule();
1473                         else
1474                                 break;
1475                 }
1476                 finish_wait(&delayed_root->wait, &__wait);
1477         }
1478
1479         btrfs_wq_run_delayed_node(delayed_root, root, BTRFS_DELAYED_BATCH);
1480 }
1481
1482 /* Will return 0 or -ENOMEM */
1483 int btrfs_insert_delayed_dir_index(struct btrfs_trans_handle *trans,
1484                                    struct btrfs_root *root, const char *name,
1485                                    int name_len, struct inode *dir,
1486                                    struct btrfs_disk_key *disk_key, u8 type,
1487                                    u64 index)
1488 {
1489         struct btrfs_delayed_node *delayed_node;
1490         struct btrfs_delayed_item *delayed_item;
1491         struct btrfs_dir_item *dir_item;
1492         int ret;
1493
1494         delayed_node = btrfs_get_or_create_delayed_node(dir);
1495         if (IS_ERR(delayed_node))
1496                 return PTR_ERR(delayed_node);
1497
1498         delayed_item = btrfs_alloc_delayed_item(sizeof(*dir_item) + name_len);
1499         if (!delayed_item) {
1500                 ret = -ENOMEM;
1501                 goto release_node;
1502         }
1503
1504         delayed_item->key.objectid = btrfs_ino(dir);
1505         btrfs_set_key_type(&delayed_item->key, BTRFS_DIR_INDEX_KEY);
1506         delayed_item->key.offset = index;
1507
1508         dir_item = (struct btrfs_dir_item *)delayed_item->data;
1509         dir_item->location = *disk_key;
1510         dir_item->transid = cpu_to_le64(trans->transid);
1511         dir_item->data_len = 0;
1512         dir_item->name_len = cpu_to_le16(name_len);
1513         dir_item->type = type;
1514         memcpy((char *)(dir_item + 1), name, name_len);
1515
1516         ret = btrfs_delayed_item_reserve_metadata(trans, root, delayed_item);
1517         /*
1518          * we have reserved enough space when we start a new transaction,
1519          * so reserving metadata failure is impossible
1520          */
1521         BUG_ON(ret);
1522
1523
1524         mutex_lock(&delayed_node->mutex);
1525         ret = __btrfs_add_delayed_insertion_item(delayed_node, delayed_item);
1526         if (unlikely(ret)) {
1527                 printk(KERN_ERR "err add delayed dir index item(name: %s) into "
1528                                 "the insertion tree of the delayed node"
1529                                 "(root id: %llu, inode id: %llu, errno: %d)\n",
1530                                 name,
1531                                 (unsigned long long)delayed_node->root->objectid,
1532                                 (unsigned long long)delayed_node->inode_id,
1533                                 ret);
1534                 BUG();
1535         }
1536         mutex_unlock(&delayed_node->mutex);
1537
1538 release_node:
1539         btrfs_release_delayed_node(delayed_node);
1540         return ret;
1541 }
1542
1543 static int btrfs_delete_delayed_insertion_item(struct btrfs_root *root,
1544                                                struct btrfs_delayed_node *node,
1545                                                struct btrfs_key *key)
1546 {
1547         struct btrfs_delayed_item *item;
1548
1549         mutex_lock(&node->mutex);
1550         item = __btrfs_lookup_delayed_insertion_item(node, key);
1551         if (!item) {
1552                 mutex_unlock(&node->mutex);
1553                 return 1;
1554         }
1555
1556         btrfs_delayed_item_release_metadata(root, item);
1557         btrfs_release_delayed_item(item);
1558         mutex_unlock(&node->mutex);
1559         return 0;
1560 }
1561
1562 int btrfs_delete_delayed_dir_index(struct btrfs_trans_handle *trans,
1563                                    struct btrfs_root *root, struct inode *dir,
1564                                    u64 index)
1565 {
1566         struct btrfs_delayed_node *node;
1567         struct btrfs_delayed_item *item;
1568         struct btrfs_key item_key;
1569         int ret;
1570
1571         node = btrfs_get_or_create_delayed_node(dir);
1572         if (IS_ERR(node))
1573                 return PTR_ERR(node);
1574
1575         item_key.objectid = btrfs_ino(dir);
1576         btrfs_set_key_type(&item_key, BTRFS_DIR_INDEX_KEY);
1577         item_key.offset = index;
1578
1579         ret = btrfs_delete_delayed_insertion_item(root, node, &item_key);
1580         if (!ret)
1581                 goto end;
1582
1583         item = btrfs_alloc_delayed_item(0);
1584         if (!item) {
1585                 ret = -ENOMEM;
1586                 goto end;
1587         }
1588
1589         item->key = item_key;
1590
1591         ret = btrfs_delayed_item_reserve_metadata(trans, root, item);
1592         /*
1593          * we have reserved enough space when we start a new transaction,
1594          * so reserving metadata failure is impossible.
1595          */
1596         BUG_ON(ret);
1597
1598         mutex_lock(&node->mutex);
1599         ret = __btrfs_add_delayed_deletion_item(node, item);
1600         if (unlikely(ret)) {
1601                 printk(KERN_ERR "err add delayed dir index item(index: %llu) "
1602                                 "into the deletion tree of the delayed node"
1603                                 "(root id: %llu, inode id: %llu, errno: %d)\n",
1604                                 (unsigned long long)index,
1605                                 (unsigned long long)node->root->objectid,
1606                                 (unsigned long long)node->inode_id,
1607                                 ret);
1608                 BUG();
1609         }
1610         mutex_unlock(&node->mutex);
1611 end:
1612         btrfs_release_delayed_node(node);
1613         return ret;
1614 }
1615
1616 int btrfs_inode_delayed_dir_index_count(struct inode *inode)
1617 {
1618         struct btrfs_delayed_node *delayed_node = btrfs_get_delayed_node(inode);
1619
1620         if (!delayed_node)
1621                 return -ENOENT;
1622
1623         /*
1624          * Since we have held i_mutex of this directory, it is impossible that
1625          * a new directory index is added into the delayed node and index_cnt
1626          * is updated now. So we needn't lock the delayed node.
1627          */
1628         if (!delayed_node->index_cnt) {
1629                 btrfs_release_delayed_node(delayed_node);
1630                 return -EINVAL;
1631         }
1632
1633         BTRFS_I(inode)->index_cnt = delayed_node->index_cnt;
1634         btrfs_release_delayed_node(delayed_node);
1635         return 0;
1636 }
1637
1638 void btrfs_get_delayed_items(struct inode *inode, struct list_head *ins_list,
1639                              struct list_head *del_list)
1640 {
1641         struct btrfs_delayed_node *delayed_node;
1642         struct btrfs_delayed_item *item;
1643
1644         delayed_node = btrfs_get_delayed_node(inode);
1645         if (!delayed_node)
1646                 return;
1647
1648         mutex_lock(&delayed_node->mutex);
1649         item = __btrfs_first_delayed_insertion_item(delayed_node);
1650         while (item) {
1651                 atomic_inc(&item->refs);
1652                 list_add_tail(&item->readdir_list, ins_list);
1653                 item = __btrfs_next_delayed_item(item);
1654         }
1655
1656         item = __btrfs_first_delayed_deletion_item(delayed_node);
1657         while (item) {
1658                 atomic_inc(&item->refs);
1659                 list_add_tail(&item->readdir_list, del_list);
1660                 item = __btrfs_next_delayed_item(item);
1661         }
1662         mutex_unlock(&delayed_node->mutex);
1663         /*
1664          * This delayed node is still cached in the btrfs inode, so refs
1665          * must be > 1 now, and we needn't check it is going to be freed
1666          * or not.
1667          *
1668          * Besides that, this function is used to read dir, we do not
1669          * insert/delete delayed items in this period. So we also needn't
1670          * requeue or dequeue this delayed node.
1671          */
1672         atomic_dec(&delayed_node->refs);
1673 }
1674
1675 void btrfs_put_delayed_items(struct list_head *ins_list,
1676                              struct list_head *del_list)
1677 {
1678         struct btrfs_delayed_item *curr, *next;
1679
1680         list_for_each_entry_safe(curr, next, ins_list, readdir_list) {
1681                 list_del(&curr->readdir_list);
1682                 if (atomic_dec_and_test(&curr->refs))
1683                         kfree(curr);
1684         }
1685
1686         list_for_each_entry_safe(curr, next, del_list, readdir_list) {
1687                 list_del(&curr->readdir_list);
1688                 if (atomic_dec_and_test(&curr->refs))
1689                         kfree(curr);
1690         }
1691 }
1692
1693 int btrfs_should_delete_dir_index(struct list_head *del_list,
1694                                   u64 index)
1695 {
1696         struct btrfs_delayed_item *curr, *next;
1697         int ret;
1698
1699         if (list_empty(del_list))
1700                 return 0;
1701
1702         list_for_each_entry_safe(curr, next, del_list, readdir_list) {
1703                 if (curr->key.offset > index)
1704                         break;
1705
1706                 list_del(&curr->readdir_list);
1707                 ret = (curr->key.offset == index);
1708
1709                 if (atomic_dec_and_test(&curr->refs))
1710                         kfree(curr);
1711
1712                 if (ret)
1713                         return 1;
1714                 else
1715                         continue;
1716         }
1717         return 0;
1718 }
1719
1720 /*
1721  * btrfs_readdir_delayed_dir_index - read dir info stored in the delayed tree
1722  *
1723  */
1724 int btrfs_readdir_delayed_dir_index(struct file *filp, void *dirent,
1725                                     filldir_t filldir,
1726                                     struct list_head *ins_list)
1727 {
1728         struct btrfs_dir_item *di;
1729         struct btrfs_delayed_item *curr, *next;
1730         struct btrfs_key location;
1731         char *name;
1732         int name_len;
1733         int over = 0;
1734         unsigned char d_type;
1735
1736         if (list_empty(ins_list))
1737                 return 0;
1738
1739         /*
1740          * Changing the data of the delayed item is impossible. So
1741          * we needn't lock them. And we have held i_mutex of the
1742          * directory, nobody can delete any directory indexes now.
1743          */
1744         list_for_each_entry_safe(curr, next, ins_list, readdir_list) {
1745                 list_del(&curr->readdir_list);
1746
1747                 if (curr->key.offset < filp->f_pos) {
1748                         if (atomic_dec_and_test(&curr->refs))
1749                                 kfree(curr);
1750                         continue;
1751                 }
1752
1753                 filp->f_pos = curr->key.offset;
1754
1755                 di = (struct btrfs_dir_item *)curr->data;
1756                 name = (char *)(di + 1);
1757                 name_len = le16_to_cpu(di->name_len);
1758
1759                 d_type = btrfs_filetype_table[di->type];
1760                 btrfs_disk_key_to_cpu(&location, &di->location);
1761
1762                 over = filldir(dirent, name, name_len, curr->key.offset,
1763                                location.objectid, d_type);
1764
1765                 if (atomic_dec_and_test(&curr->refs))
1766                         kfree(curr);
1767
1768                 if (over)
1769                         return 1;
1770         }
1771         return 0;
1772 }
1773
1774 BTRFS_SETGET_STACK_FUNCS(stack_inode_generation, struct btrfs_inode_item,
1775                          generation, 64);
1776 BTRFS_SETGET_STACK_FUNCS(stack_inode_sequence, struct btrfs_inode_item,
1777                          sequence, 64);
1778 BTRFS_SETGET_STACK_FUNCS(stack_inode_transid, struct btrfs_inode_item,
1779                          transid, 64);
1780 BTRFS_SETGET_STACK_FUNCS(stack_inode_size, struct btrfs_inode_item, size, 64);
1781 BTRFS_SETGET_STACK_FUNCS(stack_inode_nbytes, struct btrfs_inode_item,
1782                          nbytes, 64);
1783 BTRFS_SETGET_STACK_FUNCS(stack_inode_block_group, struct btrfs_inode_item,
1784                          block_group, 64);
1785 BTRFS_SETGET_STACK_FUNCS(stack_inode_nlink, struct btrfs_inode_item, nlink, 32);
1786 BTRFS_SETGET_STACK_FUNCS(stack_inode_uid, struct btrfs_inode_item, uid, 32);
1787 BTRFS_SETGET_STACK_FUNCS(stack_inode_gid, struct btrfs_inode_item, gid, 32);
1788 BTRFS_SETGET_STACK_FUNCS(stack_inode_mode, struct btrfs_inode_item, mode, 32);
1789 BTRFS_SETGET_STACK_FUNCS(stack_inode_rdev, struct btrfs_inode_item, rdev, 64);
1790 BTRFS_SETGET_STACK_FUNCS(stack_inode_flags, struct btrfs_inode_item, flags, 64);
1791
1792 BTRFS_SETGET_STACK_FUNCS(stack_timespec_sec, struct btrfs_timespec, sec, 64);
1793 BTRFS_SETGET_STACK_FUNCS(stack_timespec_nsec, struct btrfs_timespec, nsec, 32);
1794
1795 static void fill_stack_inode_item(struct btrfs_trans_handle *trans,
1796                                   struct btrfs_inode_item *inode_item,
1797                                   struct inode *inode)
1798 {
1799         btrfs_set_stack_inode_uid(inode_item, i_uid_read(inode));
1800         btrfs_set_stack_inode_gid(inode_item, i_gid_read(inode));
1801         btrfs_set_stack_inode_size(inode_item, BTRFS_I(inode)->disk_i_size);
1802         btrfs_set_stack_inode_mode(inode_item, inode->i_mode);
1803         btrfs_set_stack_inode_nlink(inode_item, inode->i_nlink);
1804         btrfs_set_stack_inode_nbytes(inode_item, inode_get_bytes(inode));
1805         btrfs_set_stack_inode_generation(inode_item,
1806                                          BTRFS_I(inode)->generation);
1807         btrfs_set_stack_inode_sequence(inode_item, inode->i_version);
1808         btrfs_set_stack_inode_transid(inode_item, trans->transid);
1809         btrfs_set_stack_inode_rdev(inode_item, inode->i_rdev);
1810         btrfs_set_stack_inode_flags(inode_item, BTRFS_I(inode)->flags);
1811         btrfs_set_stack_inode_block_group(inode_item, 0);
1812
1813         btrfs_set_stack_timespec_sec(btrfs_inode_atime(inode_item),
1814                                      inode->i_atime.tv_sec);
1815         btrfs_set_stack_timespec_nsec(btrfs_inode_atime(inode_item),
1816                                       inode->i_atime.tv_nsec);
1817
1818         btrfs_set_stack_timespec_sec(btrfs_inode_mtime(inode_item),
1819                                      inode->i_mtime.tv_sec);
1820         btrfs_set_stack_timespec_nsec(btrfs_inode_mtime(inode_item),
1821                                       inode->i_mtime.tv_nsec);
1822
1823         btrfs_set_stack_timespec_sec(btrfs_inode_ctime(inode_item),
1824                                      inode->i_ctime.tv_sec);
1825         btrfs_set_stack_timespec_nsec(btrfs_inode_ctime(inode_item),
1826                                       inode->i_ctime.tv_nsec);
1827 }
1828
1829 int btrfs_fill_inode(struct inode *inode, u32 *rdev)
1830 {
1831         struct btrfs_delayed_node *delayed_node;
1832         struct btrfs_inode_item *inode_item;
1833         struct btrfs_timespec *tspec;
1834
1835         delayed_node = btrfs_get_delayed_node(inode);
1836         if (!delayed_node)
1837                 return -ENOENT;
1838
1839         mutex_lock(&delayed_node->mutex);
1840         if (!delayed_node->inode_dirty) {
1841                 mutex_unlock(&delayed_node->mutex);
1842                 btrfs_release_delayed_node(delayed_node);
1843                 return -ENOENT;
1844         }
1845
1846         inode_item = &delayed_node->inode_item;
1847
1848         i_uid_write(inode, btrfs_stack_inode_uid(inode_item));
1849         i_gid_write(inode, btrfs_stack_inode_gid(inode_item));
1850         btrfs_i_size_write(inode, btrfs_stack_inode_size(inode_item));
1851         inode->i_mode = btrfs_stack_inode_mode(inode_item);
1852         set_nlink(inode, btrfs_stack_inode_nlink(inode_item));
1853         inode_set_bytes(inode, btrfs_stack_inode_nbytes(inode_item));
1854         BTRFS_I(inode)->generation = btrfs_stack_inode_generation(inode_item);
1855         inode->i_version = btrfs_stack_inode_sequence(inode_item);
1856         inode->i_rdev = 0;
1857         *rdev = btrfs_stack_inode_rdev(inode_item);
1858         BTRFS_I(inode)->flags = btrfs_stack_inode_flags(inode_item);
1859
1860         tspec = btrfs_inode_atime(inode_item);
1861         inode->i_atime.tv_sec = btrfs_stack_timespec_sec(tspec);
1862         inode->i_atime.tv_nsec = btrfs_stack_timespec_nsec(tspec);
1863
1864         tspec = btrfs_inode_mtime(inode_item);
1865         inode->i_mtime.tv_sec = btrfs_stack_timespec_sec(tspec);
1866         inode->i_mtime.tv_nsec = btrfs_stack_timespec_nsec(tspec);
1867
1868         tspec = btrfs_inode_ctime(inode_item);
1869         inode->i_ctime.tv_sec = btrfs_stack_timespec_sec(tspec);
1870         inode->i_ctime.tv_nsec = btrfs_stack_timespec_nsec(tspec);
1871
1872         inode->i_generation = BTRFS_I(inode)->generation;
1873         BTRFS_I(inode)->index_cnt = (u64)-1;
1874
1875         mutex_unlock(&delayed_node->mutex);
1876         btrfs_release_delayed_node(delayed_node);
1877         return 0;
1878 }
1879
1880 int btrfs_delayed_update_inode(struct btrfs_trans_handle *trans,
1881                                struct btrfs_root *root, struct inode *inode)
1882 {
1883         struct btrfs_delayed_node *delayed_node;
1884         int ret = 0;
1885
1886         delayed_node = btrfs_get_or_create_delayed_node(inode);
1887         if (IS_ERR(delayed_node))
1888                 return PTR_ERR(delayed_node);
1889
1890         mutex_lock(&delayed_node->mutex);
1891         if (delayed_node->inode_dirty) {
1892                 fill_stack_inode_item(trans, &delayed_node->inode_item, inode);
1893                 goto release_node;
1894         }
1895
1896         ret = btrfs_delayed_inode_reserve_metadata(trans, root, inode,
1897                                                    delayed_node);
1898         if (ret)
1899                 goto release_node;
1900
1901         fill_stack_inode_item(trans, &delayed_node->inode_item, inode);
1902         delayed_node->inode_dirty = 1;
1903         delayed_node->count++;
1904         atomic_inc(&root->fs_info->delayed_root->items);
1905 release_node:
1906         mutex_unlock(&delayed_node->mutex);
1907         btrfs_release_delayed_node(delayed_node);
1908         return ret;
1909 }
1910
1911 static void __btrfs_kill_delayed_node(struct btrfs_delayed_node *delayed_node)
1912 {
1913         struct btrfs_root *root = delayed_node->root;
1914         struct btrfs_delayed_item *curr_item, *prev_item;
1915
1916         mutex_lock(&delayed_node->mutex);
1917         curr_item = __btrfs_first_delayed_insertion_item(delayed_node);
1918         while (curr_item) {
1919                 btrfs_delayed_item_release_metadata(root, curr_item);
1920                 prev_item = curr_item;
1921                 curr_item = __btrfs_next_delayed_item(prev_item);
1922                 btrfs_release_delayed_item(prev_item);
1923         }
1924
1925         curr_item = __btrfs_first_delayed_deletion_item(delayed_node);
1926         while (curr_item) {
1927                 btrfs_delayed_item_release_metadata(root, curr_item);
1928                 prev_item = curr_item;
1929                 curr_item = __btrfs_next_delayed_item(prev_item);
1930                 btrfs_release_delayed_item(prev_item);
1931         }
1932
1933         if (delayed_node->inode_dirty) {
1934                 btrfs_delayed_inode_release_metadata(root, delayed_node);
1935                 btrfs_release_delayed_inode(delayed_node);
1936         }
1937         mutex_unlock(&delayed_node->mutex);
1938 }
1939
1940 void btrfs_kill_delayed_inode_items(struct inode *inode)
1941 {
1942         struct btrfs_delayed_node *delayed_node;
1943
1944         delayed_node = btrfs_get_delayed_node(inode);
1945         if (!delayed_node)
1946                 return;
1947
1948         __btrfs_kill_delayed_node(delayed_node);
1949         btrfs_release_delayed_node(delayed_node);
1950 }
1951
1952 void btrfs_kill_all_delayed_nodes(struct btrfs_root *root)
1953 {
1954         u64 inode_id = 0;
1955         struct btrfs_delayed_node *delayed_nodes[8];
1956         int i, n;
1957
1958         while (1) {
1959                 spin_lock(&root->inode_lock);
1960                 n = radix_tree_gang_lookup(&root->delayed_nodes_tree,
1961                                            (void **)delayed_nodes, inode_id,
1962                                            ARRAY_SIZE(delayed_nodes));
1963                 if (!n) {
1964                         spin_unlock(&root->inode_lock);
1965                         break;
1966                 }
1967
1968                 inode_id = delayed_nodes[n - 1]->inode_id + 1;
1969
1970                 for (i = 0; i < n; i++)
1971                         atomic_inc(&delayed_nodes[i]->refs);
1972                 spin_unlock(&root->inode_lock);
1973
1974                 for (i = 0; i < n; i++) {
1975                         __btrfs_kill_delayed_node(delayed_nodes[i]);
1976                         btrfs_release_delayed_node(delayed_nodes[i]);
1977                 }
1978         }
1979 }
1980
1981 void btrfs_destroy_delayed_inodes(struct btrfs_root *root)
1982 {
1983         struct btrfs_delayed_root *delayed_root;
1984         struct btrfs_delayed_node *curr_node, *prev_node;
1985
1986         delayed_root = btrfs_get_delayed_root(root);
1987
1988         curr_node = btrfs_first_delayed_node(delayed_root);
1989         while (curr_node) {
1990                 __btrfs_kill_delayed_node(curr_node);
1991
1992                 prev_node = curr_node;
1993                 curr_node = btrfs_next_delayed_node(curr_node);
1994                 btrfs_release_delayed_node(prev_node);
1995         }
1996 }
1997