fs/btrfs/tree-log.c

   1 /*
   2  * Copyright (C) 2008 Oracle.  All rights reserved.
   3  *
   4  * This program is free software; you can redistribute it and/or
   5  * modify it under the terms of the GNU General Public
   6  * License v2 as published by the Free Software Foundation.
   7  *
   8  * This program is distributed in the hope that it will be useful,
   9  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  10  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  11  * General Public License for more details.
  12  *
  13  * You should have received a copy of the GNU General Public
  14  * License along with this program; if not, write to the
  15  * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  16  * Boston, MA 021110-1307, USA.
  17  */
  18
  19 #include <linux/sched.h>
  20 #include <linux/slab.h>
  21 #include <linux/blkdev.h>
  22 #include <linux/list_sort.h>
  23 #include "ctree.h"
  24 #include "transaction.h"
  25 #include "disk-io.h"
  26 #include "locking.h"
  27 #include "print-tree.h"
  28 #include "backref.h"
  29 #include "compat.h"
  30 #include "tree-log.h"
  31 #include "hash.h"
  32
  33 /* magic values for the inode_only field in btrfs_log_inode:
  34  *
  35  * LOG_INODE_ALL means to log everything
  36  * LOG_INODE_EXISTS means to log just enough to recreate the inode
  37  * during log replay
  38  */
  39 #define LOG_INODE_ALL 0
  40 #define LOG_INODE_EXISTS 1
  41
  42 /*
  43  * directory trouble cases
  44  *
  45  * 1) on rename or unlink, if the inode being unlinked isn't in the fsync
  46  * log, we must force a full commit before doing an fsync of the directory
  47  * where the unlink was done.
  48  * ---> record transid of last unlink/rename per directory
  49  *
  50  * mkdir foo/some_dir
  51  * normal commit
  52  * rename foo/some_dir foo2/some_dir
  53  * mkdir foo/some_dir
  54  * fsync foo/some_dir/some_file
  55  *
  56  * The fsync above will unlink the original some_dir without recording
  57  * it in its new location (foo2).  After a crash, some_dir will be gone
  58  * unless the fsync of some_file forces a full commit
  59  *
  60  * 2) we must log any new names for any file or dir that is in the fsync
  61  * log. ---> check inode while renaming/linking.
  62  *
  63  * 2a) we must log any new names for any file or dir during rename
  64  * when the directory they are being removed from was logged.
  65  * ---> check inode and old parent dir during rename
  66  *
  67  *  2a is actually the more important variant.  With the extra logging
  68  *  a crash might unlink the old name without recreating the new one
  69  *
  70  * 3) after a crash, we must go through any directories with a link count
  71  * of zero and redo the rm -rf
  72  *
  73  * mkdir f1/foo
  74  * normal commit
  75  * rm -rf f1/foo
  76  * fsync(f1)
  77  *
  78  * The directory f1 was fully removed from the FS, but fsync was never
  79  * called on f1, only its parent dir.  After a crash the rm -rf must
  80  * be replayed.  This must be able to recurse down the entire
  81  * directory tree.  The inode link count fixup code takes care of the
  82  * ugly details.
  83  */
  84
  85 /*
  86  * stages for the tree walking.  The first
  87  * stage (0) is to only pin down the blocks we find
  88  * the second stage (1) is to make sure that all the inodes
  89  * we find in the log are created in the subvolume.
  90  *
  91  * The last stage is to deal with directories and links and extents
  92  * and all the other fun semantics
  93  */
  94 #define LOG_WALK_PIN_ONLY 0
  95 #define LOG_WALK_REPLAY_INODES 1
  96 #define LOG_WALK_REPLAY_DIR_INDEX 2
  97 #define LOG_WALK_REPLAY_ALL 3
  98
  99 static int btrfs_log_inode(struct btrfs_trans_handle *trans,
 100                              struct btrfs_root *root, struct inode *inode,
 101                              int inode_only);
 102 static int link_to_fixup_dir(struct btrfs_trans_handle *trans,
 103                              struct btrfs_root *root,
 104                              struct btrfs_path *path, u64 objectid);
 105 static noinline int replay_dir_deletes(struct btrfs_trans_handle *trans,
 106                                        struct btrfs_root *root,
 107                                        struct btrfs_root *log,
 108                                        struct btrfs_path *path,
 109                                        u64 dirid, int del_all);
 110
 111 /*
 112  * tree logging is a special write ahead log used to make sure that
 113  * fsyncs and O_SYNCs can happen without doing full tree commits.
 114  *
 115  * Full tree commits are expensive because they require commonly
 116  * modified blocks to be recowed, creating many dirty pages in the
 117  * extent tree an 4x-6x higher write load than ext3.
 118  *
 119  * Instead of doing a tree commit on every fsync, we use the
 120  * key ranges and transaction ids to find items for a given file or directory
 121  * that have changed in this transaction.  Those items are copied into
 122  * a special tree (one per subvolume root), that tree is written to disk
 123  * and then the fsync is considered complete.
 124  *
 125  * After a crash, items are copied out of the log-tree back into the
 126  * subvolume tree.  Any file data extents found are recorded in the extent
 127  * allocation tree, and the log-tree freed.
 128  *
 129  * The log tree is read three times, once to pin down all the extents it is
 130  * using in ram and once, once to create all the inodes logged in the tree
 131  * and once to do all the other items.
 132  */
 133
 134 /*
 135  * start a sub transaction and setup the log tree
 136  * this increments the log tree writer count to make the people
 137  * syncing the tree wait for us to finish
 138  */
 139 static int start_log_trans(struct btrfs_trans_handle *trans,
 140                            struct btrfs_root *root)
 141 {
 142         int ret;
 143         int err = 0;
 144
 145         mutex_lock(&root->log_mutex);
 146         if (root->log_root) {
 147                 if (!root->log_start_pid) {
 148                         root->log_start_pid = current->pid;
 149                         root->log_multiple_pids = false;
 150                 } else if (root->log_start_pid != current->pid) {
 151                         root->log_multiple_pids = true;
 152                 }
 153
 154                 atomic_inc(&root->log_batch);
 155                 atomic_inc(&root->log_writers);
 156                 mutex_unlock(&root->log_mutex);
 157                 return 0;
 158         }
 159         root->log_multiple_pids = false;
 160         root->log_start_pid = current->pid;
 161         mutex_lock(&root->fs_info->tree_log_mutex);
 162         if (!root->fs_info->log_root_tree) {
 163                 ret = btrfs_init_log_root_tree(trans, root->fs_info);
 164                 if (ret)
 165                         err = ret;
 166         }
 167         if (err == 0 && !root->log_root) {
 168                 ret = btrfs_add_log_tree(trans, root);
 169                 if (ret)
 170                         err = ret;
 171         }
 172         mutex_unlock(&root->fs_info->tree_log_mutex);
 173         atomic_inc(&root->log_batch);
 174         atomic_inc(&root->log_writers);
 175         mutex_unlock(&root->log_mutex);
 176         return err;
 177 }
 178
 179 /*
 180  * returns 0 if there was a log transaction running and we were able
 181  * to join, or returns -ENOENT if there were not transactions
 182  * in progress
 183  */
 184 static int join_running_log_trans(struct btrfs_root *root)
 185 {
 186         int ret = -ENOENT;
 187
 188         smp_mb();
 189         if (!root->log_root)
 190                 return -ENOENT;
 191
 192         mutex_lock(&root->log_mutex);
 193         if (root->log_root) {
 194                 ret = 0;
 195                 atomic_inc(&root->log_writers);
 196         }
 197         mutex_unlock(&root->log_mutex);
 198         return ret;
 199 }
 200
 201 /*
 202  * This either makes the current running log transaction wait
 203  * until you call btrfs_end_log_trans() or it makes any future
 204  * log transactions wait until you call btrfs_end_log_trans()
 205  */
 206 int btrfs_pin_log_trans(struct btrfs_root *root)
 207 {
 208         int ret = -ENOENT;
 209
 210         mutex_lock(&root->log_mutex);
 211         atomic_inc(&root->log_writers);
 212         mutex_unlock(&root->log_mutex);
 213         return ret;
 214 }
 215
 216 /*
 217  * indicate we're done making changes to the log tree
 218  * and wake up anyone waiting to do a sync
 219  */
 220 void btrfs_end_log_trans(struct btrfs_root *root)
 221 {
 222         if (atomic_dec_and_test(&root->log_writers)) {
 223                 smp_mb();
 224                 if (waitqueue_active(&root->log_writer_wait))
 225                         wake_up(&root->log_writer_wait);
 226         }
 227 }
 228
 229
 230 /*
 231  * the walk control struct is used to pass state down the chain when
 232  * processing the log tree.  The stage field tells us which part
 233  * of the log tree processing we are currently doing.  The others
 234  * are state fields used for that specific part
 235  */
 236 struct walk_control {
 237         /* should we free the extent on disk when done?  This is used
 238          * at transaction commit time while freeing a log tree
 239          */
 240         int free;
 241
 242         /* should we write out the extent buffer?  This is used
 243          * while flushing the log tree to disk during a sync
 244          */
 245         int write;
 246
 247         /* should we wait for the extent buffer io to finish?  Also used
 248          * while flushing the log tree to disk for a sync
 249          */
 250         int wait;
 251
 252         /* pin only walk, we record which extents on disk belong to the
 253          * log trees
 254          */
 255         int pin;
 256
 257         /* what stage of the replay code we're currently in */
 258         int stage;
 259
 260         /* the root we are currently replaying */
 261         struct btrfs_root *replay_dest;
 262
 263         /* the trans handle for the current replay */
 264         struct btrfs_trans_handle *trans;
 265
 266         /* the function that gets used to process blocks we find in the
 267          * tree.  Note the extent_buffer might not be up to date when it is
 268          * passed in, and it must be checked or read if you need the data
 269          * inside it
 270          */
 271         int (*process_func)(struct btrfs_root *log, struct extent_buffer *eb,
 272                             struct walk_control *wc, u64 gen);
 273 };
 274
 275 /*
 276  * process_func used to pin down extents, write them or wait on them
 277  */
 278 static int process_one_buffer(struct btrfs_root *log,
 279                               struct extent_buffer *eb,
 280                               struct walk_control *wc, u64 gen)
 281 {
 282         int ret = 0;
 283
 284         /*
 285          * If this fs is mixed then we need to be able to process the leaves to
 286          * pin down any logged extents, so we have to read the block.
 287          */
 288         if (btrfs_fs_incompat(log->fs_info, MIXED_GROUPS)) {
 289                 ret = btrfs_read_buffer(eb, gen);
 290                 if (ret)
 291                         return ret;
 292         }
 293
 294         if (wc->pin)
 295                 ret = btrfs_pin_extent_for_log_replay(log->fs_info->extent_root,
 296                                                       eb->start, eb->len);
 297
 298         if (!ret && btrfs_buffer_uptodate(eb, gen, 0)) {
 299                 if (wc->pin && btrfs_header_level(eb) == 0)
 300                         ret = btrfs_exclude_logged_extents(log, eb);
 301                 if (wc->write)
 302                         btrfs_write_tree_block(eb);
 303                 if (wc->wait)
 304                         btrfs_wait_tree_block_writeback(eb);
 305         }
 306         return ret;
 307 }
 308
 309 /*
 310  * Item overwrite used by replay and tree logging.  eb, slot and key all refer
 311  * to the src data we are copying out.
 312  *
 313  * root is the tree we are copying into, and path is a scratch
 314  * path for use in this function (it should be released on entry and
 315  * will be released on exit).
 316  *
 317  * If the key is already in the destination tree the existing item is
 318  * overwritten.  If the existing item isn't big enough, it is extended.
 319  * If it is too large, it is truncated.
 320  *
 321  * If the key isn't in the destination yet, a new item is inserted.
 322  */
 323 static noinline int overwrite_item(struct btrfs_trans_handle *trans,
 324                                    struct btrfs_root *root,
 325                                    struct btrfs_path *path,
 326                                    struct extent_buffer *eb, int slot,
 327                                    struct btrfs_key *key)
 328 {
 329         int ret;
 330         u32 item_size;
 331         u64 saved_i_size = 0;
 332         int save_old_i_size = 0;
 333         unsigned long src_ptr;
 334         unsigned long dst_ptr;
 335         int overwrite_root = 0;
 336         bool inode_item = key->type == BTRFS_INODE_ITEM_KEY;
 337
 338         if (root->root_key.objectid != BTRFS_TREE_LOG_OBJECTID)
 339                 overwrite_root = 1;
 340
 341         item_size = btrfs_item_size_nr(eb, slot);
 342         src_ptr = btrfs_item_ptr_offset(eb, slot);
 343
 344         /* look for the key in the destination tree */
 345         ret = btrfs_search_slot(NULL, root, key, path, 0, 0);
 346         if (ret < 0)
 347                 return ret;
 348
 349         if (ret == 0) {
 350                 char *src_copy;
 351                 char *dst_copy;
 352                 u32 dst_size = btrfs_item_size_nr(path->nodes[0],
 353                                                   path->slots[0]);
 354                 if (dst_size != item_size)
 355                         goto insert;
 356
 357                 if (item_size == 0) {
 358                         btrfs_release_path(path);
 359                         return 0;
 360                 }
 361                 dst_copy = kmalloc(item_size, GFP_NOFS);
 362                 src_copy = kmalloc(item_size, GFP_NOFS);
 363                 if (!dst_copy || !src_copy) {
 364                         btrfs_release_path(path);
 365                         kfree(dst_copy);
 366                         kfree(src_copy);
 367                         return -ENOMEM;
 368                 }
 369
 370                 read_extent_buffer(eb, src_copy, src_ptr, item_size);
 371
 372                 dst_ptr = btrfs_item_ptr_offset(path->nodes[0], path->slots[0]);
 373                 read_extent_buffer(path->nodes[0], dst_copy, dst_ptr,
 374                                    item_size);
 375                 ret = memcmp(dst_copy, src_copy, item_size);
 376
 377                 kfree(dst_copy);
 378                 kfree(src_copy);
 379                 /*
 380                  * they have the same contents, just return, this saves
 381                  * us from cowing blocks in the destination tree and doing
 382                  * extra writes that may not have been done by a previous
 383                  * sync
 384                  */
 385                 if (ret == 0) {
 386                         btrfs_release_path(path);
 387                         return 0;
 388                 }
 389
 390                 /*
 391                  * We need to load the old nbytes into the inode so when we
 392                  * replay the extents we've logged we get the right nbytes.
 393                  */
 394                 if (inode_item) {
 395                         struct btrfs_inode_item *item;
 396                         u64 nbytes;
 397                         u32 mode;
 398
 399                         item = btrfs_item_ptr(path->nodes[0], path->slots[0],
 400                                               struct btrfs_inode_item);
 401                         nbytes = btrfs_inode_nbytes(path->nodes[0], item);
 402                         item = btrfs_item_ptr(eb, slot,
 403                                               struct btrfs_inode_item);
 404                         btrfs_set_inode_nbytes(eb, item, nbytes);
 405
 406                         /*
 407                          * If this is a directory we need to reset the i_size to
 408                          * 0 so that we can set it up properly when replaying
 409                          * the rest of the items in this log.
 410                          */
 411                         mode = btrfs_inode_mode(eb, item);
 412                         if (S_ISDIR(mode))
 413                                 btrfs_set_inode_size(eb, item, 0);
 414                 }
 415         } else if (inode_item) {
 416                 struct btrfs_inode_item *item;
 417                 u32 mode;
 418
 419                 /*
 420                  * New inode, set nbytes to 0 so that the nbytes comes out
 421                  * properly when we replay the extents.
 422                  */
 423                 item = btrfs_item_ptr(eb, slot, struct btrfs_inode_item);
 424                 btrfs_set_inode_nbytes(eb, item, 0);
 425
 426                 /*
 427                  * If this is a directory we need to reset the i_size to 0 so
 428                  * that we can set it up properly when replaying the rest of
 429                  * the items in this log.
 430                  */
 431                 mode = btrfs_inode_mode(eb, item);
 432                 if (S_ISDIR(mode))
 433                         btrfs_set_inode_size(eb, item, 0);
 434         }
 435 insert:
 436         btrfs_release_path(path);
 437         /* try to insert the key into the destination tree */
 438         ret = btrfs_insert_empty_item(trans, root, path,
 439                                       key, item_size);
 440
 441         /* make sure any existing item is the correct size */
 442         if (ret == -EEXIST) {
 443                 u32 found_size;
 444                 found_size = btrfs_item_size_nr(path->nodes[0],
 445                                                 path->slots[0]);
 446                 if (found_size > item_size)
 447                         btrfs_truncate_item(root, path, item_size, 1);
 448                 else if (found_size < item_size)
 449                         btrfs_extend_item(root, path,
 450                                           item_size - found_size);
 451         } else if (ret) {
 452                 return ret;
 453         }
 454         dst_ptr = btrfs_item_ptr_offset(path->nodes[0],
 455                                         path->slots[0]);
 456
 457         /* don't overwrite an existing inode if the generation number
 458          * was logged as zero.  This is done when the tree logging code
 459          * is just logging an inode to make sure it exists after recovery.
 460          *
 461          * Also, don't overwrite i_size on directories during replay.
 462          * log replay inserts and removes directory items based on the
 463          * state of the tree found in the subvolume, and i_size is modified
 464          * as it goes
 465          */
 466         if (key->type == BTRFS_INODE_ITEM_KEY && ret == -EEXIST) {
 467                 struct btrfs_inode_item *src_item;
 468                 struct btrfs_inode_item *dst_item;
 469
 470                 src_item = (struct btrfs_inode_item *)src_ptr;
 471                 dst_item = (struct btrfs_inode_item *)dst_ptr;
 472
 473                 if (btrfs_inode_generation(eb, src_item) == 0)
 474                         goto no_copy;
 475
 476                 if (overwrite_root &&
 477                     S_ISDIR(btrfs_inode_mode(eb, src_item)) &&
 478                     S_ISDIR(btrfs_inode_mode(path->nodes[0], dst_item))) {
 479                         save_old_i_size = 1;
 480                         saved_i_size = btrfs_inode_size(path->nodes[0],
 481                                                         dst_item);
 482                 }
 483         }
 484
 485         copy_extent_buffer(path->nodes[0], eb, dst_ptr,
 486                            src_ptr, item_size);
 487
 488         if (save_old_i_size) {
 489                 struct btrfs_inode_item *dst_item;
 490                 dst_item = (struct btrfs_inode_item *)dst_ptr;
 491                 btrfs_set_inode_size(path->nodes[0], dst_item, saved_i_size);
 492         }
 493
 494         /* make sure the generation is filled in */
 495         if (key->type == BTRFS_INODE_ITEM_KEY) {
 496                 struct btrfs_inode_item *dst_item;
 497                 dst_item = (struct btrfs_inode_item *)dst_ptr;
 498                 if (btrfs_inode_generation(path->nodes[0], dst_item) == 0) {
 499                         btrfs_set_inode_generation(path->nodes[0], dst_item,
 500                                                    trans->transid);
 501                 }
 502         }
 503 no_copy:
 504         btrfs_mark_buffer_dirty(path->nodes[0]);
 505         btrfs_release_path(path);
 506         return 0;
 507 }
 508
 509 /*
 510  * simple helper to read an inode off the disk from a given root
 511  * This can only be called for subvolume roots and not for the log
 512  */
 513 static noinline struct inode *read_one_inode(struct btrfs_root *root,
 514                                              u64 objectid)
 515 {
 516         struct btrfs_key key;
 517         struct inode *inode;
 518
 519         key.objectid = objectid;
 520         key.type = BTRFS_INODE_ITEM_KEY;
 521         key.offset = 0;
 522         inode = btrfs_iget(root->fs_info->sb, &key, root, NULL);
 523         if (IS_ERR(inode)) {
 524                 inode = NULL;
 525         } else if (is_bad_inode(inode)) {
 526                 iput(inode);
 527                 inode = NULL;
 528         }
 529         return inode;
 530 }
 531
 532 /* replays a single extent in 'eb' at 'slot' with 'key' into the
 533  * subvolume 'root'.  path is released on entry and should be released
 534  * on exit.
 535  *
 536  * extents in the log tree have not been allocated out of the extent
 537  * tree yet.  So, this completes the allocation, taking a reference
 538  * as required if the extent already exists or creating a new extent
 539  * if it isn't in the extent allocation tree yet.
 540  *
 541  * The extent is inserted into the file, dropping any existing extents
 542  * from the file that overlap the new one.
 543  */
 544 static noinline int replay_one_extent(struct btrfs_trans_handle *trans,
 545                                       struct btrfs_root *root,
 546                                       struct btrfs_path *path,
 547                                       struct extent_buffer *eb, int slot,
 548                                       struct btrfs_key *key)
 549 {
 550         int found_type;
 551         u64 extent_end;
 552         u64 start = key->offset;
 553         u64 nbytes = 0;
 554         struct btrfs_file_extent_item *item;
 555         struct inode *inode = NULL;
 556         unsigned long size;
 557         int ret = 0;
 558
 559         item = btrfs_item_ptr(eb, slot, struct btrfs_file_extent_item);
 560         found_type = btrfs_file_extent_type(eb, item);
 561
 562         if (found_type == BTRFS_FILE_EXTENT_REG ||
 563             found_type == BTRFS_FILE_EXTENT_PREALLOC) {
 564                 nbytes = btrfs_file_extent_num_bytes(eb, item);
 565                 extent_end = start + nbytes;
 566
 567                 /*
 568                  * We don't add to the inodes nbytes if we are prealloc or a
 569                  * hole.
 570                  */
 571                 if (btrfs_file_extent_disk_bytenr(eb, item) == 0)
 572                         nbytes = 0;
 573         } else if (found_type == BTRFS_FILE_EXTENT_INLINE) {
 574                 size = btrfs_file_extent_inline_len(eb, item);
 575                 nbytes = btrfs_file_extent_ram_bytes(eb, item);
 576                 extent_end = ALIGN(start + size, root->sectorsize);
 577         } else {
 578                 ret = 0;
 579                 goto out;
 580         }
 581
 582         inode = read_one_inode(root, key->objectid);
 583         if (!inode) {
 584                 ret = -EIO;
 585                 goto out;
 586         }
 587
 588         /*
 589          * first check to see if we already have this extent in the
 590          * file.  This must be done before the btrfs_drop_extents run
 591          * so we don't try to drop this extent.
 592          */
 593         ret = btrfs_lookup_file_extent(trans, root, path, btrfs_ino(inode),
 594                                        start, 0);
 595
 596         if (ret == 0 &&
 597             (found_type == BTRFS_FILE_EXTENT_REG ||
 598              found_type == BTRFS_FILE_EXTENT_PREALLOC)) {
 599                 struct btrfs_file_extent_item cmp1;
 600                 struct btrfs_file_extent_item cmp2;
 601                 struct btrfs_file_extent_item *existing;
 602                 struct extent_buffer *leaf;
 603
 604                 leaf = path->nodes[0];
 605                 existing = btrfs_item_ptr(leaf, path->slots[0],
 606                                           struct btrfs_file_extent_item);
 607
 608                 read_extent_buffer(eb, &cmp1, (unsigned long)item,
 609                                    sizeof(cmp1));
 610                 read_extent_buffer(leaf, &cmp2, (unsigned long)existing,
 611                                    sizeof(cmp2));
 612
 613                 /*
 614                  * we already have a pointer to this exact extent,
 615                  * we don't have to do anything
 616                  */
 617                 if (memcmp(&cmp1, &cmp2, sizeof(cmp1)) == 0) {
 618                         btrfs_release_path(path);
 619                         goto out;
 620                 }
 621         }
 622         btrfs_release_path(path);
 623
 624         /* drop any overlapping extents */
 625         ret = btrfs_drop_extents(trans, root, inode, start, extent_end, 1);
 626         if (ret)
 627                 goto out;
 628
 629         if (found_type == BTRFS_FILE_EXTENT_REG ||
 630             found_type == BTRFS_FILE_EXTENT_PREALLOC) {
 631                 u64 offset;
 632                 unsigned long dest_offset;
 633                 struct btrfs_key ins;
 634
 635                 ret = btrfs_insert_empty_item(trans, root, path, key,
 636                                               sizeof(*item));
 637                 if (ret)
 638                         goto out;
 639                 dest_offset = btrfs_item_ptr_offset(path->nodes[0],
 640                                                     path->slots[0]);
 641                 copy_extent_buffer(path->nodes[0], eb, dest_offset,
 642                                 (unsigned long)item,  sizeof(*item));
 643
 644                 ins.objectid = btrfs_file_extent_disk_bytenr(eb, item);
 645                 ins.offset = btrfs_file_extent_disk_num_bytes(eb, item);
 646                 ins.type = BTRFS_EXTENT_ITEM_KEY;
 647                 offset = key->offset - btrfs_file_extent_offset(eb, item);
 648
 649                 if (ins.objectid > 0) {
 650                         u64 csum_start;
 651                         u64 csum_end;
 652                         LIST_HEAD(ordered_sums);
 653                         /*
 654                          * is this extent already allocated in the extent
 655                          * allocation tree?  If so, just add a reference
 656                          */
 657                         ret = btrfs_lookup_extent(root, ins.objectid,
 658                                                 ins.offset);
 659                         if (ret == 0) {
 660                                 ret = btrfs_inc_extent_ref(trans, root,
 661                                                 ins.objectid, ins.offset,
 662                                                 0, root->root_key.objectid,
 663                                                 key->objectid, offset, 0);
 664                                 if (ret)
 665                                         goto out;
 666                         } else {
 667                                 /*
 668                                  * insert the extent pointer in the extent
 669                                  * allocation tree
 670                                  */
 671                                 ret = btrfs_alloc_logged_file_extent(trans,
 672                                                 root, root->root_key.objectid,
 673                                                 key->objectid, offset, &ins);
 674                                 if (ret)
 675                                         goto out;
 676                         }
 677                         btrfs_release_path(path);
 678
 679                         if (btrfs_file_extent_compression(eb, item)) {
 680                                 csum_start = ins.objectid;
 681                                 csum_end = csum_start + ins.offset;
 682                         } else {
 683                                 csum_start = ins.objectid +
 684                                         btrfs_file_extent_offset(eb, item);
 685                                 csum_end = csum_start +
 686                                         btrfs_file_extent_num_bytes(eb, item);
 687                         }
 688
 689                         ret = btrfs_lookup_csums_range(root->log_root,
 690                                                 csum_start, csum_end - 1,
 691                                                 &ordered_sums, 0);
 692                         if (ret)
 693                                 goto out;
 694                         while (!list_empty(&ordered_sums)) {
 695                                 struct btrfs_ordered_sum *sums;
 696                                 sums = list_entry(ordered_sums.next,
 697                                                 struct btrfs_ordered_sum,
 698                                                 list);
 699                                 if (!ret)
 700                                         ret = btrfs_csum_file_blocks(trans,
 701                                                 root->fs_info->csum_root,
 702                                                 sums);
 703                                 list_del(&sums->list);
 704                                 kfree(sums);
 705                         }
 706                         if (ret)
 707                                 goto out;
 708                 } else {
 709                         btrfs_release_path(path);
 710                 }
 711         } else if (found_type == BTRFS_FILE_EXTENT_INLINE) {
 712                 /* inline extents are easy, we just overwrite them */
 713                 ret = overwrite_item(trans, root, path, eb, slot, key);
 714                 if (ret)
 715                         goto out;
 716         }
 717
 718         inode_add_bytes(inode, nbytes);
 719         ret = btrfs_update_inode(trans, root, inode);
 720 out:
 721         if (inode)
 722                 iput(inode);
 723         return ret;
 724 }
 725
 726 /*
 727  * when cleaning up conflicts between the directory names in the
 728  * subvolume, directory names in the log and directory names in the
 729  * inode back references, we may have to unlink inodes from directories.
 730  *
 731  * This is a helper function to do the unlink of a specific directory
 732  * item
 733  */
 734 static noinline int drop_one_dir_item(struct btrfs_trans_handle *trans,
 735                                       struct btrfs_root *root,
 736                                       struct btrfs_path *path,
 737                                       struct inode *dir,
 738                                       struct btrfs_dir_item *di)
 739 {
 740         struct inode *inode;
 741         char *name;
 742         int name_len;
 743         struct extent_buffer *leaf;
 744         struct btrfs_key location;
 745         int ret;
 746
 747         leaf = path->nodes[0];
 748
 749         btrfs_dir_item_key_to_cpu(leaf, di, &location);
 750         name_len = btrfs_dir_name_len(leaf, di);
 751         name = kmalloc(name_len, GFP_NOFS);
 752         if (!name)
 753                 return -ENOMEM;
 754
 755         read_extent_buffer(leaf, name, (unsigned long)(di + 1), name_len);
 756         btrfs_release_path(path);
 757
 758         inode = read_one_inode(root, location.objectid);
 759         if (!inode) {
 760                 ret = -EIO;
 761                 goto out;
 762         }
 763
 764         ret = link_to_fixup_dir(trans, root, path, location.objectid);
 765         if (ret)
 766                 goto out;
 767
 768         ret = btrfs_unlink_inode(trans, root, dir, inode, name, name_len);
 769         if (ret)
 770                 goto out;
 771         else
 772                 ret = btrfs_run_delayed_items(trans, root);
 773 out:
 774         kfree(name);
 775         iput(inode);
 776         return ret;
 777 }
 778
 779 /*
 780  * helper function to see if a given name and sequence number found
 781  * in an inode back reference are already in a directory and correctly
 782  * point to this inode
 783  */
 784 static noinline int inode_in_dir(struct btrfs_root *root,
 785                                  struct btrfs_path *path,
 786                                  u64 dirid, u64 objectid, u64 index,
 787                                  const char *name, int name_len)
 788 {
 789         struct btrfs_dir_item *di;
 790         struct btrfs_key location;
 791         int match = 0;
 792
 793         di = btrfs_lookup_dir_index_item(NULL, root, path, dirid,
 794                                          index, name, name_len, 0);
 795         if (di && !IS_ERR(di)) {
 796                 btrfs_dir_item_key_to_cpu(path->nodes[0], di, &location);
 797                 if (location.objectid != objectid)
 798                         goto out;
 799         } else
 800                 goto out;
 801         btrfs_release_path(path);
 802
 803         di = btrfs_lookup_dir_item(NULL, root, path, dirid, name, name_len, 0);
 804         if (di && !IS_ERR(di)) {
 805                 btrfs_dir_item_key_to_cpu(path->nodes[0], di, &location);
 806                 if (location.objectid != objectid)
 807                         goto out;
 808         } else
 809                 goto out;
 810         match = 1;
 811 out:
 812         btrfs_release_path(path);
 813         return match;
 814 }
 815
 816 /*
 817  * helper function to check a log tree for a named back reference in
 818  * an inode.  This is used to decide if a back reference that is
 819  * found in the subvolume conflicts with what we find in the log.
 820  *
 821  * inode backreferences may have multiple refs in a single item,
 822  * during replay we process one reference at a time, and we don't
 823  * want to delete valid links to a file from the subvolume if that
 824  * link is also in the log.
 825  */
 826 static noinline int backref_in_log(struct btrfs_root *log,
 827                                    struct btrfs_key *key,
 828                                    u64 ref_objectid,
 829                                    char *name, int namelen)
 830 {
 831         struct btrfs_path *path;
 832         struct btrfs_inode_ref *ref;
 833         unsigned long ptr;
 834         unsigned long ptr_end;
 835         unsigned long name_ptr;
 836         int found_name_len;
 837         int item_size;
 838         int ret;
 839         int match = 0;
 840
 841         path = btrfs_alloc_path();
 842         if (!path)
 843                 return -ENOMEM;
 844
 845         ret = btrfs_search_slot(NULL, log, key, path, 0, 0);
 846         if (ret != 0)
 847                 goto out;
 848
 849         ptr = btrfs_item_ptr_offset(path->nodes[0], path->slots[0]);
 850
 851         if (key->type == BTRFS_INODE_EXTREF_KEY) {
 852                 if (btrfs_find_name_in_ext_backref(path, ref_objectid,
 853                                                    name, namelen, NULL))
 854                         match = 1;
 855
 856                 goto out;
 857         }
 858
 859         item_size = btrfs_item_size_nr(path->nodes[0], path->slots[0]);
 860         ptr_end = ptr + item_size;
 861         while (ptr < ptr_end) {
 862                 ref = (struct btrfs_inode_ref *)ptr;
 863                 found_name_len = btrfs_inode_ref_name_len(path->nodes[0], ref);
 864                 if (found_name_len == namelen) {
 865                         name_ptr = (unsigned long)(ref + 1);
 866                         ret = memcmp_extent_buffer(path->nodes[0], name,
 867                                                    name_ptr, namelen);
 868                         if (ret == 0) {
 869                                 match = 1;
 870                                 goto out;
 871                         }
 872                 }
 873                 ptr = (unsigned long)(ref + 1) + found_name_len;
 874         }
 875 out:
 876         btrfs_free_path(path);
 877         return match;
 878 }
 879
 880 static inline int __add_inode_ref(struct btrfs_trans_handle *trans,
 881                                   struct btrfs_root *root,
 882                                   struct btrfs_path *path,
 883                                   struct btrfs_root *log_root,
 884                                   struct inode *dir, struct inode *inode,
 885                                   struct extent_buffer *eb,
 886                                   u64 inode_objectid, u64 parent_objectid,
 887                                   u64 ref_index, char *name, int namelen,
 888                                   int *search_done)
 889 {
 890         int ret;
 891         char *victim_name;
 892         int victim_name_len;
 893         struct extent_buffer *leaf;
 894         struct btrfs_dir_item *di;
 895         struct btrfs_key search_key;
 896         struct btrfs_inode_extref *extref;
 897
 898 again:
 899         /* Search old style refs */
 900         search_key.objectid = inode_objectid;
 901         search_key.type = BTRFS_INODE_REF_KEY;
 902         search_key.offset = parent_objectid;
 903         ret = btrfs_search_slot(NULL, root, &search_key, path, 0, 0);
 904         if (ret == 0) {
 905                 struct btrfs_inode_ref *victim_ref;
 906                 unsigned long ptr;
 907                 unsigned long ptr_end;
 908
 909                 leaf = path->nodes[0];
 910
 911                 /* are we trying to overwrite a back ref for the root directory
 912                  * if so, just jump out, we're done
 913                  */
 914                 if (search_key.objectid == search_key.offset)
 915                         return 1;
 916
 917                 /* check all the names in this back reference to see
 918                  * if they are in the log.  if so, we allow them to stay
 919                  * otherwise they must be unlinked as a conflict
 920                  */
 921                 ptr = btrfs_item_ptr_offset(leaf, path->slots[0]);
 922                 ptr_end = ptr + btrfs_item_size_nr(leaf, path->slots[0]);
 923                 while (ptr < ptr_end) {
 924                         victim_ref = (struct btrfs_inode_ref *)ptr;
 925                         victim_name_len = btrfs_inode_ref_name_len(leaf,
 926                                                                    victim_ref);
 927                         victim_name = kmalloc(victim_name_len, GFP_NOFS);
 928                         if (!victim_name)
 929                                 return -ENOMEM;
 930
 931                         read_extent_buffer(leaf, victim_name,
 932                                            (unsigned long)(victim_ref + 1),
 933                                            victim_name_len);
 934
 935                         if (!backref_in_log(log_root, &search_key,
 936                                             parent_objectid,
 937                                             victim_name,
 938                                             victim_name_len)) {
 939                                 btrfs_inc_nlink(inode);
 940                                 btrfs_release_path(path);
 941
 942                                 ret = btrfs_unlink_inode(trans, root, dir,
 943                                                          inode, victim_name,
 944                                                          victim_name_len);
 945                                 kfree(victim_name);
 946                                 if (ret)
 947                                         return ret;
 948                                 ret = btrfs_run_delayed_items(trans, root);
 949                                 if (ret)
 950                                         return ret;
 951                                 *search_done = 1;
 952                                 goto again;
 953                         }
 954                         kfree(victim_name);
 955
 956                         ptr = (unsigned long)(victim_ref + 1) + victim_name_len;
 957                 }
 958
 959                 /*
 960                  * NOTE: we have searched root tree and checked the
 961                  * coresponding ref, it does not need to check again.
 962                  */
 963                 *search_done = 1;
 964         }
 965         btrfs_release_path(path);
 966
 967         /* Same search but for extended refs */
 968         extref = btrfs_lookup_inode_extref(NULL, root, path, name, namelen,
 969                                            inode_objectid, parent_objectid, 0,
 970                                            0);
 971         if (!IS_ERR_OR_NULL(extref)) {
 972                 u32 item_size;
 973                 u32 cur_offset = 0;
 974                 unsigned long base;
 975                 struct inode *victim_parent;
 976
 977                 leaf = path->nodes[0];
 978
 979                 item_size = btrfs_item_size_nr(leaf, path->slots[0]);
 980                 base = btrfs_item_ptr_offset(leaf, path->slots[0]);
 981
 982                 while (cur_offset < item_size) {
 983                         extref = (struct btrfs_inode_extref *)base + cur_offset;
 984
 985                         victim_name_len = btrfs_inode_extref_name_len(leaf, extref);
 986
 987                         if (btrfs_inode_extref_parent(leaf, extref) != parent_objectid)
 988                                 goto next;
 989
 990                         victim_name = kmalloc(victim_name_len, GFP_NOFS);
 991                         if (!victim_name)
 992                                 return -ENOMEM;
 993                         read_extent_buffer(leaf, victim_name, (unsigned long)&extref->name,
 994                                            victim_name_len);
 995
 996                         search_key.objectid = inode_objectid;
 997                         search_key.type = BTRFS_INODE_EXTREF_KEY;
 998                         search_key.offset = btrfs_extref_hash(parent_objectid,
 999                                                               victim_name,
1000                                                               victim_name_len);
1001                         ret = 0;
1002                         if (!backref_in_log(log_root, &search_key,
1003                                             parent_objectid, victim_name,
1004                                             victim_name_len)) {
1005                                 ret = -ENOENT;
1006                                 victim_parent = read_one_inode(root,
1007                                                                parent_objectid);
1008                                 if (victim_parent) {
1009                                         btrfs_inc_nlink(inode);
1010                                         btrfs_release_path(path);
1011
1012                                         ret = btrfs_unlink_inode(trans, root,
1013                                                                  victim_parent,
1014                                                                  inode,
1015                                                                  victim_name,
1016                                                                  victim_name_len);
1017                                         if (!ret)
1018                                                 ret = btrfs_run_delayed_items(
1019                                                                   trans, root);
1020                                 }
1021                                 iput(victim_parent);
1022                                 kfree(victim_name);
1023                                 if (ret)
1024                                         return ret;
1025                                 *search_done = 1;
1026                                 goto again;
1027                         }
1028                         kfree(victim_name);
1029                         if (ret)
1030                                 return ret;
1031 next:
1032                         cur_offset += victim_name_len + sizeof(*extref);
1033                 }
1034                 *search_done = 1;
1035         }
1036         btrfs_release_path(path);
1037
1038         /* look for a conflicting sequence number */
1039         di = btrfs_lookup_dir_index_item(trans, root, path, btrfs_ino(dir),
1040                                          ref_index, name, namelen, 0);
1041         if (di && !IS_ERR(di)) {
1042                 ret = drop_one_dir_item(trans, root, path, dir, di);
1043                 if (ret)
1044                         return ret;
1045         }
1046         btrfs_release_path(path);
1047
1048         /* look for a conflicing name */
1049         di = btrfs_lookup_dir_item(trans, root, path, btrfs_ino(dir),
1050                                    name, namelen, 0);
1051         if (di && !IS_ERR(di)) {
1052                 ret = drop_one_dir_item(trans, root, path, dir, di);
1053                 if (ret)
1054                         return ret;
1055         }
1056         btrfs_release_path(path);
1057
1058         return 0;
1059 }
1060
1061 static int extref_get_fields(struct extent_buffer *eb, unsigned long ref_ptr,
1062                              u32 *namelen, char **name, u64 *index,
1063                              u64 *parent_objectid)
1064 {
1065         struct btrfs_inode_extref *extref;
1066
1067         extref = (struct btrfs_inode_extref *)ref_ptr;
1068
1069         *namelen = btrfs_inode_extref_name_len(eb, extref);
1070         *name = kmalloc(*namelen, GFP_NOFS);
1071         if (*name == NULL)
1072                 return -ENOMEM;
1073
1074         read_extent_buffer(eb, *name, (unsigned long)&extref->name,
1075                            *namelen);
1076
1077         *index = btrfs_inode_extref_index(eb, extref);
1078         if (parent_objectid)
1079                 *parent_objectid = btrfs_inode_extref_parent(eb, extref);
1080
1081         return 0;
1082 }
1083
1084 static int ref_get_fields(struct extent_buffer *eb, unsigned long ref_ptr,
1085                           u32 *namelen, char **name, u64 *index)
1086 {
1087         struct btrfs_inode_ref *ref;
1088
1089         ref = (struct btrfs_inode_ref *)ref_ptr;
1090
1091         *namelen = btrfs_inode_ref_name_len(eb, ref);
1092         *name = kmalloc(*namelen, GFP_NOFS);
1093         if (*name == NULL)
1094                 return -ENOMEM;
1095
1096         read_extent_buffer(eb, *name, (unsigned long)(ref + 1), *namelen);
1097
1098         *index = btrfs_inode_ref_index(eb, ref);
1099
1100         return 0;
1101 }
1102
1103 /*
1104  * replay one inode back reference item found in the log tree.
1105  * eb, slot and key refer to the buffer and key found in the log tree.
1106  * root is the destination we are replaying into, and path is for temp
1107  * use by this function.  (it should be released on return).
1108  */
1109 static noinline int add_inode_ref(struct btrfs_trans_handle *trans,
1110                                   struct btrfs_root *root,
1111                                   struct btrfs_root *log,
1112                                   struct btrfs_path *path,
1113                                   struct extent_buffer *eb, int slot,
1114                                   struct btrfs_key *key)
1115 {
1116         struct inode *dir;
1117         struct inode *inode;
1118         unsigned long ref_ptr;
1119         unsigned long ref_end;
1120         char *name;
1121         int namelen;
1122         int ret;
1123         int search_done = 0;
1124         int log_ref_ver = 0;
1125         u64 parent_objectid;
1126         u64 inode_objectid;
1127         u64 ref_index = 0;
1128         int ref_struct_size;
1129
1130         ref_ptr = btrfs_item_ptr_offset(eb, slot);
1131         ref_end = ref_ptr + btrfs_item_size_nr(eb, slot);
1132
1133         if (key->type == BTRFS_INODE_EXTREF_KEY) {
1134                 struct btrfs_inode_extref *r;
1135
1136                 ref_struct_size = sizeof(struct btrfs_inode_extref);
1137                 log_ref_ver = 1;
1138                 r = (struct btrfs_inode_extref *)ref_ptr;
1139                 parent_objectid = btrfs_inode_extref_parent(eb, r);
1140         } else {
1141                 ref_struct_size = sizeof(struct btrfs_inode_ref);
1142                 parent_objectid = key->offset;
1143         }
1144         inode_objectid = key->objectid;
1145
1146         /*
1147          * it is possible that we didn't log all the parent directories
1148          * for a given inode.  If we don't find the dir, just don't
1149          * copy the back ref in.  The link count fixup code will take
1150          * care of the rest
1151          */
1152         dir = read_one_inode(root, parent_objectid);
1153         if (!dir)
1154                 return -ENOENT;
1155
1156         inode = read_one_inode(root, inode_objectid);
1157         if (!inode) {
1158                 iput(dir);
1159                 return -EIO;
1160         }
1161
1162         while (ref_ptr < ref_end) {
1163                 if (log_ref_ver) {
1164                         ret = extref_get_fields(eb, ref_ptr, &namelen, &name,
1165                                                 &ref_index, &parent_objectid);
1166                         /*
1167                          * parent object can change from one array
1168                          * item to another.
1169                          */
1170                         if (!dir)
1171                                 dir = read_one_inode(root, parent_objectid);
1172                         if (!dir)
1173                                 return -ENOENT;
1174                 } else {
1175                         ret = ref_get_fields(eb, ref_ptr, &namelen, &name,
1176                                              &ref_index);
1177                 }
1178                 if (ret)
1179                         return ret;
1180
1181                 /* if we already have a perfect match, we're done */
1182                 if (!inode_in_dir(root, path, btrfs_ino(dir), btrfs_ino(inode),
1183                                   ref_index, name, namelen)) {
1184                         /*
1185                          * look for a conflicting back reference in the
1186                          * metadata. if we find one we have to unlink that name
1187                          * of the file before we add our new link.  Later on, we
1188                          * overwrite any existing back reference, and we don't
1189                          * want to create dangling pointers in the directory.
1190                          */
1191
1192                         if (!search_done) {
1193                                 ret = __add_inode_ref(trans, root, path, log,
1194                                                       dir, inode, eb,
1195                                                       inode_objectid,
1196                                                       parent_objectid,
1197                                                       ref_index, name, namelen,
1198                                                       &search_done);
1199                                 if (ret == 1) {
1200                                         ret = 0;
1201                                         goto out;
1202                                 }
1203                                 if (ret)
1204                                         goto out;
1205                         }
1206
1207                         /* insert our name */
1208                         ret = btrfs_add_link(trans, dir, inode, name, namelen,
1209                                              0, ref_index);
1210                         if (ret)
1211                                 goto out;
1212
1213                         btrfs_update_inode(trans, root, inode);
1214                 }
1215
1216                 ref_ptr = (unsigned long)(ref_ptr + ref_struct_size) + namelen;
1217                 kfree(name);
1218                 if (log_ref_ver) {
1219                         iput(dir);
1220                         dir = NULL;
1221                 }
1222         }
1223
1224         /* finally write the back reference in the inode */
1225         ret = overwrite_item(trans, root, path, eb, slot, key);
1226 out:
1227         btrfs_release_path(path);
1228         iput(dir);
1229         iput(inode);
1230         return ret;
1231 }
1232
1233 static int insert_orphan_item(struct btrfs_trans_handle *trans,
1234                               struct btrfs_root *root, u64 offset)
1235 {
1236         int ret;
1237         ret = btrfs_find_orphan_item(root, offset);
1238         if (ret > 0)
1239                 ret = btrfs_insert_orphan_item(trans, root, offset);
1240         return ret;
1241 }
1242
1243 static int count_inode_extrefs(struct btrfs_root *root,
1244                                struct inode *inode, struct btrfs_path *path)
1245 {
1246         int ret = 0;
1247         int name_len;
1248         unsigned int nlink = 0;
1249         u32 item_size;
1250         u32 cur_offset = 0;
1251         u64 inode_objectid = btrfs_ino(inode);
1252         u64 offset = 0;
1253         unsigned long ptr;
1254         struct btrfs_inode_extref *extref;
1255         struct extent_buffer *leaf;
1256
1257         while (1) {
1258                 ret = btrfs_find_one_extref(root, inode_objectid, offset, path,
1259                                             &extref, &offset);
1260                 if (ret)
1261                         break;
1262
1263                 leaf = path->nodes[0];
1264                 item_size = btrfs_item_size_nr(leaf, path->slots[0]);
1265                 ptr = btrfs_item_ptr_offset(leaf, path->slots[0]);
1266
1267                 while (cur_offset < item_size) {
1268                         extref = (struct btrfs_inode_extref *) (ptr + cur_offset);
1269                         name_len = btrfs_inode_extref_name_len(leaf, extref);
1270
1271                         nlink++;
1272
1273                         cur_offset += name_len + sizeof(*extref);
1274                 }
1275
1276                 offset++;
1277                 btrfs_release_path(path);
1278         }
1279         btrfs_release_path(path);
1280
1281         if (ret < 0)
1282                 return ret;
1283         return nlink;
1284 }
1285
1286 static int count_inode_refs(struct btrfs_root *root,
1287                                struct inode *inode, struct btrfs_path *path)
1288 {
1289         int ret;
1290         struct btrfs_key key;
1291         unsigned int nlink = 0;
1292         unsigned long ptr;
1293         unsigned long ptr_end;
1294         int name_len;
1295         u64 ino = btrfs_ino(inode);
1296
1297         key.objectid = ino;
1298         key.type = BTRFS_INODE_REF_KEY;
1299         key.offset = (u64)-1;
1300
1301         while (1) {
1302                 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
1303                 if (ret < 0)
1304                         break;
1305                 if (ret > 0) {
1306                         if (path->slots[0] == 0)
1307                                 break;
1308                         path->slots[0]--;
1309                 }
1310                 btrfs_item_key_to_cpu(path->nodes[0], &key,
1311                                       path->slots[0]);
1312                 if (key.objectid != ino ||
1313                     key.type != BTRFS_INODE_REF_KEY)
1314                         break;
1315                 ptr = btrfs_item_ptr_offset(path->nodes[0], path->slots[0]);
1316                 ptr_end = ptr + btrfs_item_size_nr(path->nodes[0],
1317                                                    path->slots[0]);
1318                 while (ptr < ptr_end) {
1319                         struct btrfs_inode_ref *ref;
1320
1321                         ref = (struct btrfs_inode_ref *)ptr;
1322                         name_len = btrfs_inode_ref_name_len(path->nodes[0],
1323                                                             ref);
1324                         ptr = (unsigned long)(ref + 1) + name_len;
1325                         nlink++;
1326                 }
1327
1328                 if (key.offset == 0)
1329                         break;
1330                 key.offset--;
1331                 btrfs_release_path(path);
1332         }
1333         btrfs_release_path(path);
1334
1335         return nlink;
1336 }
1337
1338 /*
1339  * There are a few corners where the link count of the file can't
1340  * be properly maintained during replay.  So, instead of adding
1341  * lots of complexity to the log code, we just scan the backrefs
1342  * for any file that has been through replay.
1343  *
1344  * The scan will update the link count on the inode to reflect the
1345  * number of back refs found.  If it goes down to zero, the iput
1346  * will free the inode.
1347  */
1348 static noinline int fixup_inode_link_count(struct btrfs_trans_handle *trans,
1349                                            struct btrfs_root *root,
1350                                            struct inode *inode)
1351 {
1352         struct btrfs_path *path;
1353         int ret;
1354         u64 nlink = 0;
1355         u64 ino = btrfs_ino(inode);
1356
1357         path = btrfs_alloc_path();
1358         if (!path)
1359                 return -ENOMEM;
1360
1361         ret = count_inode_refs(root, inode, path);
1362         if (ret < 0)
1363                 goto out;
1364
1365         nlink = ret;
1366
1367         ret = count_inode_extrefs(root, inode, path);
1368         if (ret == -ENOENT)
1369                 ret = 0;
1370
1371         if (ret < 0)
1372                 goto out;
1373
1374         nlink += ret;
1375
1376         ret = 0;
1377
1378         if (nlink != inode->i_nlink) {
1379                 set_nlink(inode, nlink);
1380                 btrfs_update_inode(trans, root, inode);
1381         }
1382         BTRFS_I(inode)->index_cnt = (u64)-1;
1383
1384         if (inode->i_nlink == 0) {
1385                 if (S_ISDIR(inode->i_mode)) {
1386                         ret = replay_dir_deletes(trans, root, NULL, path,
1387                                                  ino, 1);
1388                         if (ret)
1389                                 goto out;
1390                 }
1391                 ret = insert_orphan_item(trans, root, ino);
1392         }
1393
1394 out:
1395         btrfs_free_path(path);
1396         return ret;
1397 }
1398
1399 static noinline int fixup_inode_link_counts(struct btrfs_trans_handle *trans,
1400                                             struct btrfs_root *root,
1401                                             struct btrfs_path *path)
1402 {
1403         int ret;
1404         struct btrfs_key key;
1405         struct inode *inode;
1406
1407         key.objectid = BTRFS_TREE_LOG_FIXUP_OBJECTID;
1408         key.type = BTRFS_ORPHAN_ITEM_KEY;
1409         key.offset = (u64)-1;
1410         while (1) {
1411                 ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
1412                 if (ret < 0)
1413                         break;
1414
1415                 if (ret == 1) {
1416                         if (path->slots[0] == 0)
1417                                 break;
1418                         path->slots[0]--;
1419                 }
1420
1421                 btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
1422                 if (key.objectid != BTRFS_TREE_LOG_FIXUP_OBJECTID ||
1423                     key.type != BTRFS_ORPHAN_ITEM_KEY)
1424                         break;
1425
1426                 ret = btrfs_del_item(trans, root, path);
1427                 if (ret)
1428                         goto out;
1429
1430                 btrfs_release_path(path);
1431                 inode = read_one_inode(root, key.offset);
1432                 if (!inode)
1433                         return -EIO;
1434
1435                 ret = fixup_inode_link_count(trans, root, inode);
1436                 iput(inode);
1437                 if (ret)
1438                         goto out;
1439
1440                 /*
1441                  * fixup on a directory may create new entries,
1442                  * make sure we always look for the highset possible
1443                  * offset
1444                  */
1445                 key.offset = (u64)-1;
1446         }
1447         ret = 0;
1448 out:
1449         btrfs_release_path(path);
1450         return ret;
1451 }
1452
1453
1454 /*
1455  * record a given inode in the fixup dir so we can check its link
1456  * count when replay is done.  The link count is incremented here
1457  * so the inode won't go away until we check it
1458  */
1459 static noinline int link_to_fixup_dir(struct btrfs_trans_handle *trans,
1460                                       struct btrfs_root *root,
1461                                       struct btrfs_path *path,
1462                                       u64 objectid)
1463 {
1464         struct btrfs_key key;
1465         int ret = 0;
1466         struct inode *inode;
1467
1468         inode = read_one_inode(root, objectid);
1469         if (!inode)
1470                 return -EIO;
1471
1472         key.objectid = BTRFS_TREE_LOG_FIXUP_OBJECTID;
1473         btrfs_set_key_type(&key, BTRFS_ORPHAN_ITEM_KEY);
1474         key.offset = objectid;
1475
1476         ret = btrfs_insert_empty_item(trans, root, path, &key, 0);
1477
1478         btrfs_release_path(path);
1479         if (ret == 0) {
1480                 if (!inode->i_nlink)
1481                         set_nlink(inode, 1);
1482                 else
1483                         btrfs_inc_nlink(inode);
1484                 ret = btrfs_update_inode(trans, root, inode);
1485         } else if (ret == -EEXIST) {
1486                 ret = 0;
1487         } else {
1488                 BUG(); /* Logic Error */
1489         }
1490         iput(inode);
1491
1492         return ret;
1493 }
1494
1495 /*
1496  * when replaying the log for a directory, we only insert names
1497  * for inodes that actually exist.  This means an fsync on a directory
1498  * does not implicitly fsync all the new files in it
1499  */
1500 static noinline int insert_one_name(struct btrfs_trans_handle *trans,
1501                                     struct btrfs_root *root,
1502                                     struct btrfs_path *path,
1503                                     u64 dirid, u64 index,
1504                                     char *name, int name_len, u8 type,
1505                                     struct btrfs_key *location)
1506 {
1507         struct inode *inode;
1508         struct inode *dir;
1509         int ret;
1510
1511         inode = read_one_inode(root, location->objectid);
1512         if (!inode)
1513                 return -ENOENT;
1514
1515         dir = read_one_inode(root, dirid);
1516         if (!dir) {
1517                 iput(inode);
1518                 return -EIO;
1519         }
1520
1521         ret = btrfs_add_link(trans, dir, inode, name, name_len, 1, index);
1522
1523         /* FIXME, put inode into FIXUP list */
1524
1525         iput(inode);
1526         iput(dir);
1527         return ret;
1528 }
1529
1530 /*
1531  * take a single entry in a log directory item and replay it into
1532  * the subvolume.
1533  *
1534  * if a conflicting item exists in the subdirectory already,
1535  * the inode it points to is unlinked and put into the link count
1536  * fix up tree.
1537  *
1538  * If a name from the log points to a file or directory that does
1539  * not exist in the FS, it is skipped.  fsyncs on directories
1540  * do not force down inodes inside that directory, just changes to the
1541  * names or unlinks in a directory.
1542  */
1543 static noinline int replay_one_name(struct btrfs_trans_handle *trans,
1544                                     struct btrfs_root *root,
1545                                     struct btrfs_path *path,
1546                                     struct extent_buffer *eb,
1547                                     struct btrfs_dir_item *di,
1548                                     struct btrfs_key *key)
1549 {
1550         char *name;
1551         int name_len;
1552         struct btrfs_dir_item *dst_di;
1553         struct btrfs_key found_key;
1554         struct btrfs_key log_key;
1555         struct inode *dir;
1556         u8 log_type;
1557         int exists;
1558         int ret = 0;
1559         bool update_size = (key->type == BTRFS_DIR_INDEX_KEY);
1560
1561         dir = read_one_inode(root, key->objectid);
1562         if (!dir)
1563                 return -EIO;
1564
1565         name_len = btrfs_dir_name_len(eb, di);
1566         name = kmalloc(name_len, GFP_NOFS);
1567         if (!name) {
1568                 ret = -ENOMEM;
1569                 goto out;
1570         }
1571
1572         log_type = btrfs_dir_type(eb, di);
1573         read_extent_buffer(eb, name, (unsigned long)(di + 1),
1574                    name_len);
1575
1576         btrfs_dir_item_key_to_cpu(eb, di, &log_key);
1577         exists = btrfs_lookup_inode(trans, root, path, &log_key, 0);
1578         if (exists == 0)
1579                 exists = 1;
1580         else
1581                 exists = 0;
1582         btrfs_release_path(path);
1583
1584         if (key->type == BTRFS_DIR_ITEM_KEY) {
1585                 dst_di = btrfs_lookup_dir_item(trans, root, path, key->objectid,
1586                                        name, name_len, 1);
1587         } else if (key->type == BTRFS_DIR_INDEX_KEY) {
1588                 dst_di = btrfs_lookup_dir_index_item(trans, root, path,
1589                                                      key->objectid,
1590                                                      key->offset, name,
1591                                                      name_len, 1);
1592         } else {
1593                 /* Corruption */
1594                 ret = -EINVAL;
1595                 goto out;
1596         }
1597         if (IS_ERR_OR_NULL(dst_di)) {
1598                 /* we need a sequence number to insert, so we only
1599                  * do inserts for the BTRFS_DIR_INDEX_KEY types
1600                  */
1601                 if (key->type != BTRFS_DIR_INDEX_KEY)
1602                         goto out;
1603                 goto insert;
1604         }
1605
1606         btrfs_dir_item_key_to_cpu(path->nodes[0], dst_di, &found_key);
1607         /* the existing item matches the logged item */
1608         if (found_key.objectid == log_key.objectid &&
1609             found_key.type == log_key.type &&
1610             found_key.offset == log_key.offset &&
1611             btrfs_dir_type(path->nodes[0], dst_di) == log_type) {
1612                 goto out;
1613         }
1614
1615         /*
1616          * don't drop the conflicting directory entry if the inode
1617          * for the new entry doesn't exist
1618          */
1619         if (!exists)
1620                 goto out;
1621
1622         ret = drop_one_dir_item(trans, root, path, dir, dst_di);
1623         if (ret)
1624                 goto out;
1625
1626         if (key->type == BTRFS_DIR_INDEX_KEY)
1627                 goto insert;
1628 out:
1629         btrfs_release_path(path);
1630         if (!ret && update_size) {
1631                 btrfs_i_size_write(dir, dir->i_size + name_len * 2);
1632                 ret = btrfs_update_inode(trans, root, dir);
1633         }
1634         kfree(name);
1635         iput(dir);
1636         return ret;
1637
1638 insert:
1639         btrfs_release_path(path);
1640         ret = insert_one_name(trans, root, path, key->objectid, key->offset,
1641                               name, name_len, log_type, &log_key);
1642         if (ret && ret != -ENOENT)
1643                 goto out;
1644         update_size = false;
1645         ret = 0;
1646         goto out;
1647 }
1648
1649 /*
1650  * find all the names in a directory item and reconcile them into
1651  * the subvolume.  Only BTRFS_DIR_ITEM_KEY types will have more than
1652  * one name in a directory item, but the same code gets used for
1653  * both directory index types
1654  */
1655 static noinline int replay_one_dir_item(struct btrfs_trans_handle *trans,
1656                                         struct btrfs_root *root,
1657                                         struct btrfs_path *path,
1658                                         struct extent_buffer *eb, int slot,
1659                                         struct btrfs_key *key)
1660 {
1661         int ret;
1662         u32 item_size = btrfs_item_size_nr(eb, slot);
1663         struct btrfs_dir_item *di;
1664         int name_len;
1665         unsigned long ptr;
1666         unsigned long ptr_end;
1667
1668         ptr = btrfs_item_ptr_offset(eb, slot);
1669         ptr_end = ptr + item_size;
1670         while (ptr < ptr_end) {
1671                 di = (struct btrfs_dir_item *)ptr;
1672                 if (verify_dir_item(root, eb, di))
1673                         return -EIO;
1674                 name_len = btrfs_dir_name_len(eb, di);
1675                 ret = replay_one_name(trans, root, path, eb, di, key);
1676                 if (ret)
1677                         return ret;
1678                 ptr = (unsigned long)(di + 1);
1679                 ptr += name_len;
1680         }
1681         return 0;
1682 }
1683
1684 /*
1685  * directory replay has two parts.  There are the standard directory
1686  * items in the log copied from the subvolume, and range items
1687  * created in the log while the subvolume was logged.
1688  *
1689  * The range items tell us which parts of the key space the log
1690  * is authoritative for.  During replay, if a key in the subvolume
1691  * directory is in a logged range item, but not actually in the log
1692  * that means it was deleted from the directory before the fsync
1693  * and should be removed.
1694  */
1695 static noinline int find_dir_range(struct btrfs_root *root,
1696                                    struct btrfs_path *path,
1697                                    u64 dirid, int key_type,
1698                                    u64 *start_ret, u64 *end_ret)
1699 {
1700         struct btrfs_key key;
1701         u64 found_end;
1702         struct btrfs_dir_log_item *item;
1703         int ret;
1704         int nritems;
1705
1706         if (*start_ret == (u64)-1)
1707                 return 1;
1708
1709         key.objectid = dirid;
1710         key.type = key_type;
1711         key.offset = *start_ret;
1712
1713         ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
1714         if (ret < 0)
1715                 goto out;
1716         if (ret > 0) {
1717                 if (path->slots[0] == 0)
1718                         goto out;
1719                 path->slots[0]--;
1720         }
1721         if (ret != 0)
1722                 btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
1723
1724         if (key.type != key_type || key.objectid != dirid) {
1725                 ret = 1;
1726                 goto next;
1727         }
1728         item = btrfs_item_ptr(path->nodes[0], path->slots[0],
1729                               struct btrfs_dir_log_item);
1730         found_end = btrfs_dir_log_end(path->nodes[0], item);
1731
1732         if (*start_ret >= key.offset && *start_ret <= found_end) {
1733                 ret = 0;
1734                 *start_ret = key.offset;
1735                 *end_ret = found_end;
1736                 goto out;
1737         }
1738         ret = 1;
1739 next:
1740         /* check the next slot in the tree to see if it is a valid item */
1741         nritems = btrfs_header_nritems(path->nodes[0]);
1742         if (path->slots[0] >= nritems) {
1743                 ret = btrfs_next_leaf(root, path);
1744                 if (ret)
1745                         goto out;
1746         } else {
1747                 path->slots[0]++;
1748         }
1749
1750         btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
1751
1752         if (key.type != key_type || key.objectid != dirid) {
1753                 ret = 1;
1754                 goto out;
1755         }
1756         item = btrfs_item_ptr(path->nodes[0], path->slots[0],
1757                               struct btrfs_dir_log_item);
1758         found_end = btrfs_dir_log_end(path->nodes[0], item);
1759         *start_ret = key.offset;
1760         *end_ret = found_end;
1761         ret = 0;
1762 out:
1763         btrfs_release_path(path);
1764         return ret;
1765 }
1766
1767 /*
1768  * this looks for a given directory item in the log.  If the directory
1769  * item is not in the log, the item is removed and the inode it points
1770  * to is unlinked
1771  */
1772 static noinline int check_item_in_log(struct btrfs_trans_handle *trans,
1773                                       struct btrfs_root *root,
1774                                       struct btrfs_root *log,
1775                                       struct btrfs_path *path,
1776                                       struct btrfs_path *log_path,
1777                                       struct inode *dir,
1778                                       struct btrfs_key *dir_key)
1779 {
1780         int ret;
1781         struct extent_buffer *eb;
1782         int slot;
1783         u32 item_size;
1784         struct btrfs_dir_item *di;
1785         struct btrfs_dir_item *log_di;
1786         int name_len;
1787         unsigned long ptr;
1788         unsigned long ptr_end;
1789         char *name;
1790         struct inode *inode;
1791         struct btrfs_key location;
1792
1793 again:
1794         eb = path->nodes[0];
1795         slot = path->slots[0];
1796         item_size = btrfs_item_size_nr(eb, slot);
1797         ptr = btrfs_item_ptr_offset(eb, slot);
1798         ptr_end = ptr + item_size;
1799         while (ptr < ptr_end) {
1800                 di = (struct btrfs_dir_item *)ptr;
1801                 if (verify_dir_item(root, eb, di)) {
1802                         ret = -EIO;
1803                         goto out;
1804                 }
1805
1806                 name_len = btrfs_dir_name_len(eb, di);
1807                 name = kmalloc(name_len, GFP_NOFS);
1808                 if (!name) {
1809                         ret = -ENOMEM;
1810                         goto out;
1811                 }
1812                 read_extent_buffer(eb, name, (unsigned long)(di + 1),
1813                                   name_len);
1814                 log_di = NULL;
1815                 if (log && dir_key->type == BTRFS_DIR_ITEM_KEY) {
1816                         log_di = btrfs_lookup_dir_item(trans, log, log_path,
1817                                                        dir_key->objectid,
1818                                                        name, name_len, 0);
1819                 } else if (log && dir_key->type == BTRFS_DIR_INDEX_KEY) {
1820                         log_di = btrfs_lookup_dir_index_item(trans, log,
1821                                                      log_path,
1822                                                      dir_key->objectid,
1823                                                      dir_key->offset,
1824                                                      name, name_len, 0);
1825                 }
1826                 if (IS_ERR_OR_NULL(log_di)) {
1827                         btrfs_dir_item_key_to_cpu(eb, di, &location);
1828                         btrfs_release_path(path);
1829                         btrfs_release_path(log_path);
1830                         inode = read_one_inode(root, location.objectid);
1831                         if (!inode) {
1832                                 kfree(name);
1833                                 return -EIO;
1834                         }
1835
1836                         ret = link_to_fixup_dir(trans, root,
1837                                                 path, location.objectid);
1838                         if (ret) {
1839                                 kfree(name);
1840                                 iput(inode);
1841                                 goto out;
1842                         }
1843
1844                         btrfs_inc_nlink(inode);
1845                         ret = btrfs_unlink_inode(trans, root, dir, inode,
1846                                                  name, name_len);
1847                         if (!ret)
1848                                 ret = btrfs_run_delayed_items(trans, root);
1849                         kfree(name);
1850                         iput(inode);
1851                         if (ret)
1852                                 goto out;
1853
1854                         /* there might still be more names under this key
1855                          * check and repeat if required
1856                          */
1857                         ret = btrfs_search_slot(NULL, root, dir_key, path,
1858                                                 0, 0);
1859                         if (ret == 0)
1860                                 goto again;
1861                         ret = 0;
1862                         goto out;
1863                 }
1864                 btrfs_release_path(log_path);
1865                 kfree(name);
1866
1867                 ptr = (unsigned long)(di + 1);
1868                 ptr += name_len;
1869         }
1870         ret = 0;
1871 out:
1872         btrfs_release_path(path);
1873         btrfs_release_path(log_path);
1874         return ret;
1875 }
1876
1877 /*
1878  * deletion replay happens before we copy any new directory items
1879  * out of the log or out of backreferences from inodes.  It
1880  * scans the log to find ranges of keys that log is authoritative for,
1881  * and then scans the directory to find items in those ranges that are
1882  * not present in the log.
1883  *
1884  * Anything we don't find in the log is unlinked and removed from the
1885  * directory.
1886  */
1887 static noinline int replay_dir_deletes(struct btrfs_trans_handle *trans,
1888                                        struct btrfs_root *root,
1889                                        struct btrfs_root *log,
1890                                        struct btrfs_path *path,
1891                                        u64 dirid, int del_all)
1892 {
1893         u64 range_start;
1894         u64 range_end;
1895         int key_type = BTRFS_DIR_LOG_ITEM_KEY;
1896         int ret = 0;
1897         struct btrfs_key dir_key;
1898         struct btrfs_key found_key;
1899         struct btrfs_path *log_path;
1900         struct inode *dir;
1901
1902         dir_key.objectid = dirid;
1903         dir_key.type = BTRFS_DIR_ITEM_KEY;
1904         log_path = btrfs_alloc_path();
1905         if (!log_path)
1906                 return -ENOMEM;
1907
1908         dir = read_one_inode(root, dirid);
1909         /* it isn't an error if the inode isn't there, that can happen
1910          * because we replay the deletes before we copy in the inode item
1911          * from the log
1912          */
1913         if (!dir) {
1914                 btrfs_free_path(log_path);
1915                 return 0;
1916         }
1917 again:
1918         range_start = 0;
1919         range_end = 0;
1920         while (1) {
1921                 if (del_all)
1922                         range_end = (u64)-1;
1923                 else {
1924                         ret = find_dir_range(log, path, dirid, key_type,
1925                                              &range_start, &range_end);
1926                         if (ret != 0)
1927                                 break;
1928                 }
1929
1930                 dir_key.offset = range_start;
1931                 while (1) {
1932                         int nritems;
1933                         ret = btrfs_search_slot(NULL, root, &dir_key, path,
1934                                                 0, 0);
1935                         if (ret < 0)
1936                                 goto out;
1937
1938                         nritems = btrfs_header_nritems(path->nodes[0]);
1939                         if (path->slots[0] >= nritems) {
1940                                 ret = btrfs_next_leaf(root, path);
1941                                 if (ret)
1942                                         break;
1943                         }
1944                         btrfs_item_key_to_cpu(path->nodes[0], &found_key,
1945                                               path->slots[0]);
1946                         if (found_key.objectid != dirid ||
1947                             found_key.type != dir_key.type)
1948                                 goto next_type;
1949
1950                         if (found_key.offset > range_end)
1951                                 break;
1952
1953                         ret = check_item_in_log(trans, root, log, path,
1954                                                 log_path, dir,
1955                                                 &found_key);
1956                         if (ret)
1957                                 goto out;
1958                         if (found_key.offset == (u64)-1)
1959                                 break;
1960                         dir_key.offset = found_key.offset + 1;
1961                 }
1962                 btrfs_release_path(path);
1963                 if (range_end == (u64)-1)
1964                         break;
1965                 range_start = range_end + 1;
1966         }
1967
1968 next_type:
1969         ret = 0;
1970         if (key_type == BTRFS_DIR_LOG_ITEM_KEY) {
1971                 key_type = BTRFS_DIR_LOG_INDEX_KEY;
1972                 dir_key.type = BTRFS_DIR_INDEX_KEY;
1973                 btrfs_release_path(path);
1974                 goto again;
1975         }
1976 out:
1977         btrfs_release_path(path);
1978         btrfs_free_path(log_path);
1979         iput(dir);
1980         return ret;
1981 }
1982
1983 /*
1984  * the process_func used to replay items from the log tree.  This
1985  * gets called in two different stages.  The first stage just looks
1986  * for inodes and makes sure they are all copied into the subvolume.
1987  *
1988  * The second stage copies all the other item types from the log into
1989  * the subvolume.  The two stage approach is slower, but gets rid of
1990  * lots of complexity around inodes referencing other inodes that exist
1991  * only in the log (references come from either directory items or inode
1992  * back refs).
1993  */
1994 static int replay_one_buffer(struct btrfs_root *log, struct extent_buffer *eb,
1995                              struct walk_control *wc, u64 gen)
1996 {
1997         int nritems;
1998         struct btrfs_path *path;
1999         struct btrfs_root *root = wc->replay_dest;
2000         struct btrfs_key key;
2001         int level;
2002         int i;
2003         int ret;
2004
2005         ret = btrfs_read_buffer(eb, gen);
2006         if (ret)
2007                 return ret;
2008
2009         level = btrfs_header_level(eb);
2010
2011         if (level != 0)
2012                 return 0;
2013
2014         path = btrfs_alloc_path();
2015         if (!path)
2016                 return -ENOMEM;
2017
2018         nritems = btrfs_header_nritems(eb);
2019         for (i = 0; i < nritems; i++) {
2020                 btrfs_item_key_to_cpu(eb, &key, i);
2021
2022                 /* inode keys are done during the first stage */
2023                 if (key.type == BTRFS_INODE_ITEM_KEY &&
2024                     wc->stage == LOG_WALK_REPLAY_INODES) {
2025                         struct btrfs_inode_item *inode_item;
2026                         u32 mode;
2027
2028                         inode_item = btrfs_item_ptr(eb, i,
2029                                             struct btrfs_inode_item);
2030                         mode = btrfs_inode_mode(eb, inode_item);
2031                         if (S_ISDIR(mode)) {
2032                                 ret = replay_dir_deletes(wc->trans,
2033                                          root, log, path, key.objectid, 0);
2034                                 if (ret)
2035                                         break;
2036                         }
2037                         ret = overwrite_item(wc->trans, root, path,
2038                                              eb, i, &key);
2039                         if (ret)
2040                                 break;
2041
2042                         /* for regular files, make sure corresponding
2043                          * orhpan item exist. extents past the new EOF
2044                          * will be truncated later by orphan cleanup.
2045                          */
2046                         if (S_ISREG(mode)) {
2047                                 ret = insert_orphan_item(wc->trans, root,
2048                                                          key.objectid);
2049                                 if (ret)
2050                                         break;
2051                         }
2052
2053                         ret = link_to_fixup_dir(wc->trans, root,
2054                                                 path, key.objectid);
2055                         if (ret)
2056                                 break;
2057                 }
2058
2059                 if (key.type == BTRFS_DIR_INDEX_KEY &&
2060                     wc->stage == LOG_WALK_REPLAY_DIR_INDEX) {
2061                         ret = replay_one_dir_item(wc->trans, root, path,
2062                                                   eb, i, &key);
2063                         if (ret)
2064                                 break;
2065                 }
2066
2067                 if (wc->stage < LOG_WALK_REPLAY_ALL)
2068                         continue;
2069
2070                 /* these keys are simply copied */
2071                 if (key.type == BTRFS_XATTR_ITEM_KEY) {
2072                         ret = overwrite_item(wc->trans, root, path,
2073                                              eb, i, &key);
2074                         if (ret)
2075                                 break;
2076                 } else if (key.type == BTRFS_INODE_REF_KEY ||
2077                            key.type == BTRFS_INODE_EXTREF_KEY) {
2078                         ret = add_inode_ref(wc->trans, root, log, path,
2079                                             eb, i, &key);
2080                         if (ret && ret != -ENOENT)
2081                                 break;
2082                         ret = 0;
2083                 } else if (key.type == BTRFS_EXTENT_DATA_KEY) {
2084                         ret = replay_one_extent(wc->trans, root, path,
2085                                                 eb, i, &key);
2086                         if (ret)
2087                                 break;
2088                 } else if (key.type == BTRFS_DIR_ITEM_KEY) {
2089                         ret = replay_one_dir_item(wc->trans, root, path,
2090                                                   eb, i, &key);
2091                         if (ret)
2092                                 break;
2093                 }
2094         }
2095         btrfs_free_path(path);
2096         return ret;
2097 }
2098
2099 static noinline int walk_down_log_tree(struct btrfs_trans_handle *trans,
2100                                    struct btrfs_root *root,
2101                                    struct btrfs_path *path, int *level,
2102                                    struct walk_control *wc)
2103 {
2104         u64 root_owner;
2105         u64 bytenr;
2106         u64 ptr_gen;
2107         struct extent_buffer *next;
2108         struct extent_buffer *cur;
2109         struct extent_buffer *parent;
2110         u32 blocksize;
2111         int ret = 0;
2112
2113         WARN_ON(*level < 0);
2114         WARN_ON(*level >= BTRFS_MAX_LEVEL);
2115
2116         while (*level > 0) {
2117                 WARN_ON(*level < 0);
2118                 WARN_ON(*level >= BTRFS_MAX_LEVEL);
2119                 cur = path->nodes[*level];
2120
2121                 if (btrfs_header_level(cur) != *level)
2122                         WARN_ON(1);
2123
2124                 if (path->slots[*level] >=
2125                     btrfs_header_nritems(cur))
2126                         break;
2127
2128                 bytenr = btrfs_node_blockptr(cur, path->slots[*level]);
2129                 ptr_gen = btrfs_node_ptr_generation(cur, path->slots[*level]);
2130                 blocksize = btrfs_level_size(root, *level - 1);
2131
2132                 parent = path->nodes[*level];
2133                 root_owner = btrfs_header_owner(parent);
2134
2135                 next = btrfs_find_create_tree_block(root, bytenr, blocksize);
2136                 if (!next)
2137                         return -ENOMEM;
2138
2139                 if (*level == 1) {
2140                         ret = wc->process_func(root, next, wc, ptr_gen);
2141                         if (ret) {
2142                                 free_extent_buffer(next);
2143                                 return ret;
2144                         }
2145
2146                         path->slots[*level]++;
2147                         if (wc->free) {
2148                                 ret = btrfs_read_buffer(next, ptr_gen);
2149                                 if (ret) {
2150                                         free_extent_buffer(next);
2151                                         return ret;
2152                                 }
2153
2154                                 btrfs_tree_lock(next);
2155                                 btrfs_set_lock_blocking(next);
2156                                 clean_tree_block(trans, root, next);
2157                                 btrfs_wait_tree_block_writeback(next);
2158                                 btrfs_tree_unlock(next);
2159
2160                                 WARN_ON(root_owner !=
2161                                         BTRFS_TREE_LOG_OBJECTID);
2162                                 ret = btrfs_free_and_pin_reserved_extent(root,
2163                                                          bytenr, blocksize);
2164                                 if (ret) {
2165                                         free_extent_buffer(next);
2166                                         return ret;
2167                                 }
2168                         }
2169                         free_extent_buffer(next);
2170                         continue;
2171                 }
2172                 ret = btrfs_read_buffer(next, ptr_gen);
2173                 if (ret) {
2174                         free_extent_buffer(next);
2175                         return ret;
2176                 }
2177
2178                 WARN_ON(*level <= 0);
2179                 if (path->nodes[*level-1])
2180                         free_extent_buffer(path->nodes[*level-1]);
2181                 path->nodes[*level-1] = next;
2182                 *level = btrfs_header_level(next);
2183                 path->slots[*level] = 0;
2184                 cond_resched();
2185         }
2186         WARN_ON(*level < 0);
2187         WARN_ON(*level >= BTRFS_MAX_LEVEL);
2188
2189         path->slots[*level] = btrfs_header_nritems(path->nodes[*level]);
2190
2191         cond_resched();
2192         return 0;
2193 }
2194
2195 static noinline int walk_up_log_tree(struct btrfs_trans_handle *trans,
2196                                  struct btrfs_root *root,
2197                                  struct btrfs_path *path, int *level,
2198                                  struct walk_control *wc)
2199 {
2200         u64 root_owner;
2201         int i;
2202         int slot;
2203         int ret;
2204
2205         for (i = *level; i < BTRFS_MAX_LEVEL - 1 && path->nodes[i]; i++) {
2206                 slot = path->slots[i];
2207                 if (slot + 1 < btrfs_header_nritems(path->nodes[i])) {
2208                         path->slots[i]++;
2209                         *level = i;
2210                         WARN_ON(*level == 0);
2211                         return 0;
2212                 } else {
2213                         struct extent_buffer *parent;
2214                         if (path->nodes[*level] == root->node)
2215                                 parent = path->nodes[*level];
2216                         else
2217                                 parent = path->nodes[*level + 1];
2218
2219                         root_owner = btrfs_header_owner(parent);
2220                         ret = wc->process_func(root, path->nodes[*level], wc,
2221                                  btrfs_header_generation(path->nodes[*level]));
2222                         if (ret)
2223                                 return ret;
2224
2225                         if (wc->free) {
2226                                 struct extent_buffer *next;
2227
2228                                 next = path->nodes[*level];
2229
2230                                 btrfs_tree_lock(next);
2231                                 btrfs_set_lock_blocking(next);
2232                                 clean_tree_block(trans, root, next);
2233                                 btrfs_wait_tree_block_writeback(next);
2234                                 btrfs_tree_unlock(next);
2235
2236                                 WARN_ON(root_owner != BTRFS_TREE_LOG_OBJECTID);
2237                                 ret = btrfs_free_and_pin_reserved_extent(root,
2238                                                 path->nodes[*level]->start,
2239                                                 path->nodes[*level]->len);
2240                                 if (ret)
2241                                         return ret;
2242                         }
2243                         free_extent_buffer(path->nodes[*level]);
2244                         path->nodes[*level] = NULL;
2245                         *level = i + 1;
2246                 }
2247         }
2248         return 1;
2249 }
2250
2251 /*
2252  * drop the reference count on the tree rooted at 'snap'.  This traverses
2253  * the tree freeing any blocks that have a ref count of zero after being
2254  * decremented.
2255  */
2256 static int walk_log_tree(struct btrfs_trans_handle *trans,
2257                          struct btrfs_root *log, struct walk_control *wc)
2258 {
2259         int ret = 0;
2260         int wret;
2261         int level;
2262         struct btrfs_path *path;
2263         int orig_level;
2264
2265         path = btrfs_alloc_path();
2266         if (!path)
2267                 return -ENOMEM;
2268
2269         level = btrfs_header_level(log->node);
2270         orig_level = level;
2271         path->nodes[level] = log->node;
2272         extent_buffer_get(log->node);
2273         path->slots[level] = 0;
2274
2275         while (1) {
2276                 wret = walk_down_log_tree(trans, log, path, &level, wc);
2277                 if (wret > 0)
2278                         break;
2279                 if (wret < 0) {
2280                         ret = wret;
2281                         goto out;
2282                 }
2283
2284                 wret = walk_up_log_tree(trans, log, path, &level, wc);
2285                 if (wret > 0)
2286                         break;
2287                 if (wret < 0) {
2288                         ret = wret;
2289                         goto out;
2290                 }
2291         }
2292
2293         /* was the root node processed? if not, catch it here */
2294         if (path->nodes[orig_level]) {
2295                 ret = wc->process_func(log, path->nodes[orig_level], wc,
2296                          btrfs_header_generation(path->nodes[orig_level]));
2297                 if (ret)
2298                         goto out;
2299                 if (wc->free) {
2300                         struct extent_buffer *next;
2301
2302                         next = path->nodes[orig_level];
2303
2304                         btrfs_tree_lock(next);
2305                         btrfs_set_lock_blocking(next);
2306                         clean_tree_block(trans, log, next);
2307                         btrfs_wait_tree_block_writeback(next);
2308                         btrfs_tree_unlock(next);
2309
2310                         WARN_ON(log->root_key.objectid !=
2311                                 BTRFS_TREE_LOG_OBJECTID);
2312                         ret = btrfs_free_and_pin_reserved_extent(log, next->start,
2313                                                          next->len);
2314                         if (ret)
2315                                 goto out;
2316                 }
2317         }
2318
2319 out:
2320         btrfs_free_path(path);
2321         return ret;
2322 }
2323
2324 /*
2325  * helper function to update the item for a given subvolumes log root
2326  * in the tree of log roots
2327  */
2328 static int update_log_root(struct btrfs_trans_handle *trans,
2329                            struct btrfs_root *log)
2330 {
2331         int ret;
2332
2333         if (log->log_transid == 1) {
2334                 /* insert root item on the first sync */
2335                 ret = btrfs_insert_root(trans, log->fs_info->log_root_tree,
2336                                 &log->root_key, &log->root_item);
2337         } else {
2338                 ret = btrfs_update_root(trans, log->fs_info->log_root_tree,
2339                                 &log->root_key, &log->root_item);
2340         }
2341         return ret;
2342 }
2343
2344 static int wait_log_commit(struct btrfs_trans_handle *trans,
2345                            struct btrfs_root *root, unsigned long transid)
2346 {
2347         DEFINE_WAIT(wait);
2348         int index = transid % 2;
2349
2350         /*
2351          * we only allow two pending log transactions at a time,
2352          * so we know that if ours is more than 2 older than the
2353          * current transaction, we're done
2354          */
2355         do {
2356                 prepare_to_wait(&root->log_commit_wait[index],
2357                                 &wait, TASK_UNINTERRUPTIBLE);
2358                 mutex_unlock(&root->log_mutex);
2359
2360                 if (root->fs_info->last_trans_log_full_commit !=
2361                     trans->transid && root->log_transid < transid + 2 &&
2362                     atomic_read(&root->log_commit[index]))
2363                         schedule();
2364
2365                 finish_wait(&root->log_commit_wait[index], &wait);
2366                 mutex_lock(&root->log_mutex);
2367         } while (root->fs_info->last_trans_log_full_commit !=
2368                  trans->transid && root->log_transid < transid + 2 &&
2369                  atomic_read(&root->log_commit[index]));
2370         return 0;
2371 }
2372
2373 static void wait_for_writer(struct btrfs_trans_handle *trans,
2374                             struct btrfs_root *root)
2375 {
2376         DEFINE_WAIT(wait);
2377         while (root->fs_info->last_trans_log_full_commit !=
2378                trans->transid && atomic_read(&root->log_writers)) {
2379                 prepare_to_wait(&root->log_writer_wait,
2380                                 &wait, TASK_UNINTERRUPTIBLE);
2381                 mutex_unlock(&root->log_mutex);
2382                 if (root->fs_info->last_trans_log_full_commit !=
2383                     trans->transid && atomic_read(&root->log_writers))
2384                         schedule();
2385                 mutex_lock(&root->log_mutex);
2386                 finish_wait(&root->log_writer_wait, &wait);
2387         }
2388 }
2389
2390 /*
2391  * btrfs_sync_log does sends a given tree log down to the disk and
2392  * updates the super blocks to record it.  When this call is done,
2393  * you know that any inodes previously logged are safely on disk only
2394  * if it returns 0.
2395  *
2396  * Any other return value means you need to call btrfs_commit_transaction.
2397  * Some of the edge cases for fsyncing directories that have had unlinks
2398  * or renames done in the past mean that sometimes the only safe
2399  * fsync is to commit the whole FS.  When btrfs_sync_log returns -EAGAIN,
2400  * that has happened.
2401  */
2402 int btrfs_sync_log(struct btrfs_trans_handle *trans,
2403                    struct btrfs_root *root)
2404 {
2405         int index1;
2406         int index2;
2407         int mark;
2408         int ret;
2409         struct btrfs_root *log = root->log_root;
2410         struct btrfs_root *log_root_tree = root->fs_info->log_root_tree;
2411         unsigned long log_transid = 0;
2412         struct blk_plug plug;
2413
2414         mutex_lock(&root->log_mutex);
2415         log_transid = root->log_transid;
2416         index1 = root->log_transid % 2;
2417         if (atomic_read(&root->log_commit[index1])) {
2418                 wait_log_commit(trans, root, root->log_transid);
2419                 mutex_unlock(&root->log_mutex);
2420                 return 0;
2421         }
2422         atomic_set(&root->log_commit[index1], 1);
2423
2424         /* wait for previous tree log sync to complete */
2425         if (atomic_read(&root->log_commit[(index1 + 1) % 2]))
2426                 wait_log_commit(trans, root, root->log_transid - 1);
2427         while (1) {
2428                 int batch = atomic_read(&root->log_batch);
2429                 /* when we're on an ssd, just kick the log commit out */
2430                 if (!btrfs_test_opt(root, SSD) && root->log_multiple_pids) {
2431                         mutex_unlock(&root->log_mutex);
2432                         schedule_timeout_uninterruptible(1);
2433                         mutex_lock(&root->log_mutex);
2434                 }
2435                 wait_for_writer(trans, root);
2436                 if (batch == atomic_read(&root->log_batch))
2437                         break;
2438         }
2439
2440         /* bail out if we need to do a full commit */
2441         if (root->fs_info->last_trans_log_full_commit == trans->transid) {
2442                 ret = -EAGAIN;
2443                 btrfs_free_logged_extents(log, log_transid);
2444                 mutex_unlock(&root->log_mutex);
2445                 goto out;
2446         }
2447
2448         if (log_transid % 2 == 0)
2449                 mark = EXTENT_DIRTY;
2450         else
2451                 mark = EXTENT_NEW;
2452
2453         /* we start IO on  all the marked extents here, but we don't actually
2454          * wait for them until later.
2455          */
2456         blk_start_plug(&plug);
2457         ret = btrfs_write_marked_extents(log, &log->dirty_log_pages, mark);
2458         if (ret) {
2459                 blk_finish_plug(&plug);
2460                 btrfs_abort_transaction(trans, root, ret);
2461                 btrfs_free_logged_extents(log, log_transid);
2462                 mutex_unlock(&root->log_mutex);
2463                 goto out;
2464         }
2465
2466         btrfs_set_root_node(&log->root_item, log->node);
2467
2468         root->log_transid++;
2469         log->log_transid = root->log_transid;
2470         root->log_start_pid = 0;
2471         smp_mb();
2472         /*
2473          * IO has been started, blocks of the log tree have WRITTEN flag set
2474          * in their headers. new modifications of the log will be written to
2475          * new positions. so it's safe to allow log writers to go in.
2476          */
2477         mutex_unlock(&root->log_mutex);
2478
2479         mutex_lock(&log_root_tree->log_mutex);
2480         atomic_inc(&log_root_tree->log_batch);
2481         atomic_inc(&log_root_tree->log_writers);
2482         mutex_unlock(&log_root_tree->log_mutex);
2483
2484         ret = update_log_root(trans, log);
2485
2486         mutex_lock(&log_root_tree->log_mutex);
2487         if (atomic_dec_and_test(&log_root_tree->log_writers)) {
2488                 smp_mb();
2489                 if (waitqueue_active(&log_root_tree->log_writer_wait))
2490                         wake_up(&log_root_tree->log_writer_wait);
2491         }
2492
2493         if (ret) {
2494                 blk_finish_plug(&plug);
2495                 if (ret != -ENOSPC) {
2496                         btrfs_abort_transaction(trans, root, ret);
2497                         mutex_unlock(&log_root_tree->log_mutex);
2498                         goto out;
2499                 }
2500                 root->fs_info->last_trans_log_full_commit = trans->transid;
2501                 btrfs_wait_marked_extents(log, &log->dirty_log_pages, mark);
2502                 btrfs_free_logged_extents(log, log_transid);
2503                 mutex_unlock(&log_root_tree->log_mutex);
2504                 ret = -EAGAIN;
2505                 goto out;
2506         }
2507
2508         index2 = log_root_tree->log_transid % 2;
2509         if (atomic_read(&log_root_tree->log_commit[index2])) {
2510                 blk_finish_plug(&plug);
2511                 btrfs_wait_marked_extents(log, &log->dirty_log_pages, mark);
2512                 wait_log_commit(trans, log_root_tree,
2513                                 log_root_tree->log_transid);
2514                 btrfs_free_logged_extents(log, log_transid);
2515                 mutex_unlock(&log_root_tree->log_mutex);
2516                 ret = 0;
2517                 goto out;
2518         }
2519         atomic_set(&log_root_tree->log_commit[index2], 1);
2520
2521         if (atomic_read(&log_root_tree->log_commit[(index2 + 1) % 2])) {
2522                 wait_log_commit(trans, log_root_tree,
2523                                 log_root_tree->log_transid - 1);
2524         }
2525
2526         wait_for_writer(trans, log_root_tree);
2527
2528         /*
2529          * now that we've moved on to the tree of log tree roots,
2530          * check the full commit flag again
2531          */
2532         if (root->fs_info->last_trans_log_full_commit == trans->transid) {
2533                 blk_finish_plug(&plug);
2534                 btrfs_wait_marked_extents(log, &log->dirty_log_pages, mark);
2535                 btrfs_free_logged_extents(log, log_transid);
2536                 mutex_unlock(&log_root_tree->log_mutex);
2537                 ret = -EAGAIN;
2538                 goto out_wake_log_root;
2539         }
2540
2541         ret = btrfs_write_marked_extents(log_root_tree,
2542                                          &log_root_tree->dirty_log_pages,
2543                                          EXTENT_DIRTY | EXTENT_NEW);
2544         blk_finish_plug(&plug);
2545         if (ret) {
2546                 btrfs_abort_transaction(trans, root, ret);
2547                 btrfs_free_logged_extents(log, log_transid);
2548                 mutex_unlock(&log_root_tree->log_mutex);
2549                 goto out_wake_log_root;
2550         }
2551         btrfs_wait_marked_extents(log, &log->dirty_log_pages, mark);
2552         btrfs_wait_marked_extents(log_root_tree,
2553                                   &log_root_tree->dirty_log_pages,
2554                                   EXTENT_NEW | EXTENT_DIRTY);
2555         btrfs_wait_logged_extents(log, log_transid);
2556
2557         btrfs_set_super_log_root(root->fs_info->super_for_commit,
2558                                 log_root_tree->node->start);
2559         btrfs_set_super_log_root_level(root->fs_info->super_for_commit,
2560                                 btrfs_header_level(log_root_tree->node));
2561
2562         log_root_tree->log_transid++;
2563         smp_mb();
2564
2565         mutex_unlock(&log_root_tree->log_mutex);
2566
2567         /*
2568          * nobody else is going to jump in and write the the ctree
2569          * super here because the log_commit atomic below is protecting
2570          * us.  We must be called with a transaction handle pinning
2571          * the running transaction open, so a full commit can't hop
2572          * in and cause problems either.
2573          */
2574         btrfs_scrub_pause_super(root);
2575         ret = write_ctree_super(trans, root->fs_info->tree_root, 1);
2576         btrfs_scrub_continue_super(root);
2577         if (ret) {
2578                 btrfs_abort_transaction(trans, root, ret);
2579                 goto out_wake_log_root;
2580         }
2581
2582         mutex_lock(&root->log_mutex);
2583         if (root->last_log_commit < log_transid)
2584                 root->last_log_commit = log_transid;
2585         mutex_unlock(&root->log_mutex);
2586
2587 out_wake_log_root:
2588         atomic_set(&log_root_tree->log_commit[index2], 0);
2589         smp_mb();
2590         if (waitqueue_active(&log_root_tree->log_commit_wait[index2]))
2591                 wake_up(&log_root_tree->log_commit_wait[index2]);
2592 out:
2593         atomic_set(&root->log_commit[index1], 0);
2594         smp_mb();
2595         if (waitqueue_active(&root->log_commit_wait[index1]))
2596                 wake_up(&root->log_commit_wait[index1]);
2597         return ret;
2598 }
2599
2600 static void free_log_tree(struct btrfs_trans_handle *trans,
2601                           struct btrfs_root *log)
2602 {
2603         int ret;
2604         u64 start;
2605         u64 end;
2606         struct walk_control wc = {
2607                 .free = 1,
2608                 .process_func = process_one_buffer
2609         };
2610
2611         if (trans) {
2612                 ret = walk_log_tree(trans, log, &wc);
2613
2614                 /* I don't think this can happen but just in case */
2615                 if (ret)
2616                         btrfs_abort_transaction(trans, log, ret);
2617         }
2618
2619         while (1) {
2620                 ret = find_first_extent_bit(&log->dirty_log_pages,
2621                                 0, &start, &end, EXTENT_DIRTY | EXTENT_NEW,
2622                                 NULL);
2623                 if (ret)
2624                         break;
2625
2626                 clear_extent_bits(&log->dirty_log_pages, start, end,
2627                                   EXTENT_DIRTY | EXTENT_NEW, GFP_NOFS);
2628         }
2629
2630         /*
2631          * We may have short-circuited the log tree with the full commit logic
2632          * and left ordered extents on our list, so clear these out to keep us
2633          * from leaking inodes and memory.
2634          */
2635         btrfs_free_logged_extents(log, 0);
2636         btrfs_free_logged_extents(log, 1);
2637
2638         free_extent_buffer(log->node);
2639         kfree(log);
2640 }
2641
2642 /*
2643  * free all the extents used by the tree log.  This should be called
2644  * at commit time of the full transaction
2645  */
2646 int btrfs_free_log(struct btrfs_trans_handle *trans, struct btrfs_root *root)
2647 {
2648         if (root->log_root) {
2649                 free_log_tree(trans, root->log_root);
2650                 root->log_root = NULL;
2651         }
2652         return 0;
2653 }
2654
2655 int btrfs_free_log_root_tree(struct btrfs_trans_handle *trans,
2656                              struct btrfs_fs_info *fs_info)
2657 {
2658         if (fs_info->log_root_tree) {
2659                 free_log_tree(trans, fs_info->log_root_tree);
2660                 fs_info->log_root_tree = NULL;
2661         }
2662         return 0;
2663 }
2664
2665 /*
2666  * If both a file and directory are logged, and unlinks or renames are
2667  * mixed in, we have a few interesting corners:
2668  *
2669  * create file X in dir Y
2670  * link file X to X.link in dir Y
2671  * fsync file X
2672  * unlink file X but leave X.link
2673  * fsync dir Y
2674  *
2675  * After a crash we would expect only X.link to exist.  But file X
2676  * didn't get fsync'd again so the log has back refs for X and X.link.
2677  *
2678  * We solve this by removing directory entries and inode backrefs from the
2679  * log when a file that was logged in the current transaction is
2680  * unlinked.  Any later fsync will include the updated log entries, and
2681  * we'll be able to reconstruct the proper directory items from backrefs.
2682  *
2683  * This optimizations allows us to avoid relogging the entire inode
2684  * or the entire directory.
2685  */
2686 int btrfs_del_dir_entries_in_log(struct btrfs_trans_handle *trans,
2687                                  struct btrfs_root *root,
2688                                  const char *name, int name_len,
2689                                  struct inode *dir, u64 index)
2690 {
2691         struct btrfs_root *log;
2692         struct btrfs_dir_item *di;
2693         struct btrfs_path *path;
2694         int ret;
2695         int err = 0;
2696         int bytes_del = 0;
2697         u64 dir_ino = btrfs_ino(dir);
2698
2699         if (BTRFS_I(dir)->logged_trans < trans->transid)
2700                 return 0;
2701
2702         ret = join_running_log_trans(root);
2703         if (ret)
2704                 return 0;
2705
2706         mutex_lock(&BTRFS_I(dir)->log_mutex);
2707
2708         log = root->log_root;
2709         path = btrfs_alloc_path();
2710         if (!path) {
2711                 err = -ENOMEM;
2712                 goto out_unlock;
2713         }
2714
2715         di = btrfs_lookup_dir_item(trans, log, path, dir_ino,
2716                                    name, name_len, -1);
2717         if (IS_ERR(di)) {
2718                 err = PTR_ERR(di);
2719                 goto fail;
2720         }
2721         if (di) {
2722                 ret = btrfs_delete_one_dir_name(trans, log, path, di);
2723                 bytes_del += name_len;
2724                 if (ret) {
2725                         err = ret;
2726                         goto fail;
2727                 }
2728         }
2729         btrfs_release_path(path);
2730         di = btrfs_lookup_dir_index_item(trans, log, path, dir_ino,
2731                                          index, name, name_len, -1);
2732         if (IS_ERR(di)) {
2733                 err = PTR_ERR(di);
2734                 goto fail;
2735         }
2736         if (di) {
2737                 ret = btrfs_delete_one_dir_name(trans, log, path, di);
2738                 bytes_del += name_len;
2739                 if (ret) {
2740                         err = ret;
2741                         goto fail;
2742                 }
2743         }
2744
2745         /* update the directory size in the log to reflect the names
2746          * we have removed
2747          */
2748         if (bytes_del) {
2749                 struct btrfs_key key;
2750
2751                 key.objectid = dir_ino;
2752                 key.offset = 0;
2753                 key.type = BTRFS_INODE_ITEM_KEY;
2754                 btrfs_release_path(path);
2755
2756                 ret = btrfs_search_slot(trans, log, &key, path, 0, 1);
2757                 if (ret < 0) {
2758                         err = ret;
2759                         goto fail;
2760                 }
2761                 if (ret == 0) {
2762                         struct btrfs_inode_item *item;
2763                         u64 i_size;
2764
2765                         item = btrfs_item_ptr(path->nodes[0], path->slots[0],
2766                                               struct btrfs_inode_item);
2767                         i_size = btrfs_inode_size(path->nodes[0], item);
2768                         if (i_size > bytes_del)
2769                                 i_size -= bytes_del;
2770                         else
2771                                 i_size = 0;
2772                         btrfs_set_inode_size(path->nodes[0], item, i_size);
2773                         btrfs_mark_buffer_dirty(path->nodes[0]);
2774                 } else
2775                         ret = 0;
2776                 btrfs_release_path(path);
2777         }
2778 fail:
2779         btrfs_free_path(path);
2780 out_unlock:
2781         mutex_unlock(&BTRFS_I(dir)->log_mutex);
2782         if (ret == -ENOSPC) {
2783                 root->fs_info->last_trans_log_full_commit = trans->transid;
2784                 ret = 0;
2785         } else if (ret < 0)
2786                 btrfs_abort_transaction(trans, root, ret);
2787
2788         btrfs_end_log_trans(root);
2789
2790         return err;
2791 }
2792
2793 /* see comments for btrfs_del_dir_entries_in_log */
2794 int btrfs_del_inode_ref_in_log(struct btrfs_trans_handle *trans,
2795                                struct btrfs_root *root,
2796                                const char *name, int name_len,
2797                                struct inode *inode, u64 dirid)
2798 {
2799         struct btrfs_root *log;
2800         u64 index;
2801         int ret;
2802
2803         if (BTRFS_I(inode)->logged_trans < trans->transid)
2804                 return 0;
2805
2806         ret = join_running_log_trans(root);
2807         if (ret)
2808                 return 0;
2809         log = root->log_root;
2810         mutex_lock(&BTRFS_I(inode)->log_mutex);
2811
2812         ret = btrfs_del_inode_ref(trans, log, name, name_len, btrfs_ino(inode),
2813                                   dirid, &index);
2814         mutex_unlock(&BTRFS_I(inode)->log_mutex);
2815         if (ret == -ENOSPC) {
2816                 root->fs_info->last_trans_log_full_commit = trans->transid;
2817                 ret = 0;
2818         } else if (ret < 0 && ret != -ENOENT)
2819                 btrfs_abort_transaction(trans, root, ret);
2820         btrfs_end_log_trans(root);
2821
2822         return ret;
2823 }
2824
2825 /*
2826  * creates a range item in the log for 'dirid'.  first_offset and
2827  * last_offset tell us which parts of the key space the log should
2828  * be considered authoritative for.
2829  */
2830 static noinline int insert_dir_log_key(struct btrfs_trans_handle *trans,
2831                                        struct btrfs_root *log,
2832                                        struct btrfs_path *path,
2833                                        int key_type, u64 dirid,
2834                                        u64 first_offset, u64 last_offset)
2835 {
2836         int ret;
2837         struct btrfs_key key;
2838         struct btrfs_dir_log_item *item;
2839
2840         key.objectid = dirid;
2841         key.offset = first_offset;
2842         if (key_type == BTRFS_DIR_ITEM_KEY)
2843                 key.type = BTRFS_DIR_LOG_ITEM_KEY;
2844         else
2845                 key.type = BTRFS_DIR_LOG_INDEX_KEY;
2846         ret = btrfs_insert_empty_item(trans, log, path, &key, sizeof(*item));
2847         if (ret)
2848                 return ret;
2849
2850         item = btrfs_item_ptr(path->nodes[0], path->slots[0],
2851                               struct btrfs_dir_log_item);
2852         btrfs_set_dir_log_end(path->nodes[0], item, last_offset);
2853         btrfs_mark_buffer_dirty(path->nodes[0]);
2854         btrfs_release_path(path);
2855         return 0;
2856 }
2857
2858 /*
2859  * log all the items included in the current transaction for a given
2860  * directory.  This also creates the range items in the log tree required
2861  * to replay anything deleted before the fsync
2862  */
2863 static noinline int log_dir_items(struct btrfs_trans_handle *trans,
2864                           struct btrfs_root *root, struct inode *inode,
2865                           struct btrfs_path *path,
2866                           struct btrfs_path *dst_path, int key_type,
2867                           u64 min_offset, u64 *last_offset_ret)
2868 {
2869         struct btrfs_key min_key;
2870         struct btrfs_key max_key;
2871         struct btrfs_root *log = root->log_root;
2872         struct extent_buffer *src;
2873         int err = 0;
2874         int ret;
2875         int i;
2876         int nritems;
2877         u64 first_offset = min_offset;
2878         u64 last_offset = (u64)-1;
2879         u64 ino = btrfs_ino(inode);
2880
2881         log = root->log_root;
2882         max_key.objectid = ino;
2883         max_key.offset = (u64)-1;
2884         max_key.type = key_type;
2885
2886         min_key.objectid = ino;
2887         min_key.type = key_type;
2888         min_key.offset = min_offset;
2889
2890         path->keep_locks = 1;
2891
2892         ret = btrfs_search_forward(root, &min_key, &max_key,
2893                                    path, trans->transid);
2894
2895         /*
2896          * we didn't find anything from this transaction, see if there
2897          * is anything at all
2898          */
2899         if (ret != 0 || min_key.objectid != ino || min_key.type != key_type) {
2900                 min_key.objectid = ino;
2901                 min_key.type = key_type;
2902                 min_key.offset = (u64)-1;
2903                 btrfs_release_path(path);
2904                 ret = btrfs_search_slot(NULL, root, &min_key, path, 0, 0);
2905                 if (ret < 0) {
2906                         btrfs_release_path(path);
2907                         return ret;
2908                 }
2909                 ret = btrfs_previous_item(root, path, ino, key_type);
2910
2911                 /* if ret == 0 there are items for this type,
2912                  * create a range to tell us the last key of this type.
2913                  * otherwise, there are no items in this directory after
2914                  * *min_offset, and we create a range to indicate that.
2915                  */
2916                 if (ret == 0) {
2917                         struct btrfs_key tmp;
2918                         btrfs_item_key_to_cpu(path->nodes[0], &tmp,
2919                                               path->slots[0]);
2920                         if (key_type == tmp.type)
2921                                 first_offset = max(min_offset, tmp.offset) + 1;
2922                 }
2923                 goto done;
2924         }
2925
2926         /* go backward to find any previous key */
2927         ret = btrfs_previous_item(root, path, ino, key_type);
2928         if (ret == 0) {
2929                 struct btrfs_key tmp;
2930                 btrfs_item_key_to_cpu(path->nodes[0], &tmp, path->slots[0]);
2931                 if (key_type == tmp.type) {
2932                         first_offset = tmp.offset;
2933                         ret = overwrite_item(trans, log, dst_path,
2934                                              path->nodes[0], path->slots[0],
2935                                              &tmp);
2936                         if (ret) {
2937                                 err = ret;
2938                                 goto done;
2939                         }
2940                 }
2941         }
2942         btrfs_release_path(path);
2943
2944         /* find the first key from this transaction again */
2945         ret = btrfs_search_slot(NULL, root, &min_key, path, 0, 0);
2946         if (ret != 0) {
2947                 WARN_ON(1);
2948                 goto done;
2949         }
2950
2951         /*
2952          * we have a block from this transaction, log every item in it
2953          * from our directory
2954          */
2955         while (1) {
2956                 struct btrfs_key tmp;
2957                 src = path->nodes[0];
2958                 nritems = btrfs_header_nritems(src);
2959                 for (i = path->slots[0]; i < nritems; i++) {
2960                         btrfs_item_key_to_cpu(src, &min_key, i);
2961
2962                         if (min_key.objectid != ino || min_key.type != key_type)
2963                                 goto done;
2964                         ret = overwrite_item(trans, log, dst_path, src, i,
2965                                              &min_key);
2966                         if (ret) {
2967                                 err = ret;
2968                                 goto done;
2969                         }
2970                 }
2971                 path->slots[0] = nritems;
2972
2973                 /*
2974                  * look ahead to the next item and see if it is also
2975                  * from this directory and from this transaction
2976                  */
2977                 ret = btrfs_next_leaf(root, path);
2978                 if (ret == 1) {
2979                         last_offset = (u64)-1;
2980                         goto done;
2981                 }
2982                 btrfs_item_key_to_cpu(path->nodes[0], &tmp, path->slots[0]);
2983                 if (tmp.objectid != ino || tmp.type != key_type) {
2984                         last_offset = (u64)-1;
2985                         goto done;
2986                 }
2987                 if (btrfs_header_generation(path->nodes[0]) != trans->transid) {
2988                         ret = overwrite_item(trans, log, dst_path,
2989                                              path->nodes[0], path->slots[0],
2990                                              &tmp);
2991                         if (ret)
2992                                 err = ret;
2993                         else
2994                                 last_offset = tmp.offset;
2995                         goto done;
2996                 }
2997         }
2998 done:
2999         btrfs_release_path(path);
3000         btrfs_release_path(dst_path);
3001
3002         if (err == 0) {
3003                 *last_offset_ret = last_offset;
3004                 /*
3005                  * insert the log range keys to indicate where the log
3006                  * is valid
3007                  */
3008                 ret = insert_dir_log_key(trans, log, path, key_type,
3009                                          ino, first_offset, last_offset);
3010                 if (ret)
3011                         err = ret;
3012         }
3013         return err;
3014 }
3015
3016 /*
3017  * logging directories is very similar to logging inodes, We find all the items
3018  * from the current transaction and write them to the log.
3019  *
3020  * The recovery code scans the directory in the subvolume, and if it finds a
3021  * key in the range logged that is not present in the log tree, then it means
3022  * that dir entry was unlinked during the transaction.
3023  *
3024  * In order for that scan to work, we must include one key smaller than
3025  * the smallest logged by this transaction and one key larger than the largest
3026  * key logged by this transaction.
3027  */
3028 static noinline int log_directory_changes(struct btrfs_trans_handle *trans,
3029                           struct btrfs_root *root, struct inode *inode,
3030                           struct btrfs_path *path,
3031                           struct btrfs_path *dst_path)
3032 {
3033         u64 min_key;
3034         u64 max_key;
3035         int ret;
3036         int key_type = BTRFS_DIR_ITEM_KEY;
3037
3038 again:
3039         min_key = 0;
3040         max_key = 0;
3041         while (1) {
3042                 ret = log_dir_items(trans, root, inode, path,
3043                                     dst_path, key_type, min_key,
3044                                     &max_key);
3045                 if (ret)
3046                         return ret;
3047                 if (max_key == (u64)-1)
3048                         break;
3049                 min_key = max_key + 1;
3050         }
3051
3052         if (key_type == BTRFS_DIR_ITEM_KEY) {
3053                 key_type = BTRFS_DIR_INDEX_KEY;
3054                 goto again;
3055         }
3056         return 0;
3057 }
3058
3059 /*
3060  * a helper function to drop items from the log before we relog an
3061  * inode.  max_key_type indicates the highest item type to remove.
3062  * This cannot be run for file data extents because it does not
3063  * free the extents they point to.
3064  */
3065 static int drop_objectid_items(struct btrfs_trans_handle *trans,
3066                                   struct btrfs_root *log,
3067                                   struct btrfs_path *path,
3068                                   u64 objectid, int max_key_type)
3069 {
3070         int ret;
3071         struct btrfs_key key;
3072         struct btrfs_key found_key;
3073         int start_slot;
3074
3075         key.objectid = objectid;
3076         key.type = max_key_type;
3077         key.offset = (u64)-1;
3078
3079         while (1) {
3080                 ret = btrfs_search_slot(trans, log, &key, path, -1, 1);
3081                 BUG_ON(ret == 0); /* Logic error */
3082                 if (ret < 0)
3083                         break;
3084
3085                 if (path->slots[0] == 0)
3086                         break;
3087
3088                 path->slots[0]--;
3089                 btrfs_item_key_to_cpu(path->nodes[0], &found_key,
3090                                       path->slots[0]);
3091
3092                 if (found_key.objectid != objectid)
3093                         break;
3094
3095                 found_key.offset = 0;
3096                 found_key.type = 0;
3097                 ret = btrfs_bin_search(path->nodes[0], &found_key, 0,
3098                                        &start_slot);
3099
3100                 ret = btrfs_del_items(trans, log, path, start_slot,
3101                                       path->slots[0] - start_slot + 1);
3102                 /*
3103                  * If start slot isn't 0 then we don't need to re-search, we've
3104                  * found the last guy with the objectid in this tree.
3105                  */
3106                 if (ret || start_slot != 0)
3107                         break;
3108                 btrfs_release_path(path);
3109         }
3110         btrfs_release_path(path);
3111         if (ret > 0)
3112                 ret = 0;
3113         return ret;
3114 }
3115
3116 static void fill_inode_item(struct btrfs_trans_handle *trans,
3117                             struct extent_buffer *leaf,
3118                             struct btrfs_inode_item *item,
3119                             struct inode *inode, int log_inode_only)
3120 {
3121         struct btrfs_map_token token;
3122
3123         btrfs_init_map_token(&token);
3124
3125         if (log_inode_only) {
3126                 /* set the generation to zero so the recover code
3127                  * can tell the difference between an logging
3128                  * just to say 'this inode exists' and a logging
3129                  * to say 'update this inode with these values'
3130                  */
3131                 btrfs_set_token_inode_generation(leaf, item, 0, &token);
3132                 btrfs_set_token_inode_size(leaf, item, 0, &token);
3133         } else {
3134                 btrfs_set_token_inode_generation(leaf, item,
3135                                                  BTRFS_I(inode)->generation,
3136                                                  &token);
3137                 btrfs_set_token_inode_size(leaf, item, inode->i_size, &token);
3138         }
3139
3140         btrfs_set_token_inode_uid(leaf, item, i_uid_read(inode), &token);
3141         btrfs_set_token_inode_gid(leaf, item, i_gid_read(inode), &token);
3142         btrfs_set_token_inode_mode(leaf, item, inode->i_mode, &token);
3143         btrfs_set_token_inode_nlink(leaf, item, inode->i_nlink, &token);
3144
3145         btrfs_set_token_timespec_sec(leaf, btrfs_inode_atime(item),
3146                                      inode->i_atime.tv_sec, &token);
3147         btrfs_set_token_timespec_nsec(leaf, btrfs_inode_atime(item),
3148                                       inode->i_atime.tv_nsec, &token);
3149
3150         btrfs_set_token_timespec_sec(leaf, btrfs_inode_mtime(item),
3151                                      inode->i_mtime.tv_sec, &token);
3152         btrfs_set_token_timespec_nsec(leaf, btrfs_inode_mtime(item),
3153                                       inode->i_mtime.tv_nsec, &token);
3154
3155         btrfs_set_token_timespec_sec(leaf, btrfs_inode_ctime(item),
3156                                      inode->i_ctime.tv_sec, &token);
3157         btrfs_set_token_timespec_nsec(leaf, btrfs_inode_ctime(item),
3158                                       inode->i_ctime.tv_nsec, &token);
3159
3160         btrfs_set_token_inode_nbytes(leaf, item, inode_get_bytes(inode),
3161                                      &token);
3162
3163         btrfs_set_token_inode_sequence(leaf, item, inode->i_version, &token);
3164         btrfs_set_token_inode_transid(leaf, item, trans->transid, &token);
3165         btrfs_set_token_inode_rdev(leaf, item, inode->i_rdev, &token);
3166         btrfs_set_token_inode_flags(leaf, item, BTRFS_I(inode)->flags, &token);
3167         btrfs_set_token_inode_block_group(leaf, item, 0, &token);
3168 }
3169
3170 static int log_inode_item(struct btrfs_trans_handle *trans,
3171                           struct btrfs_root *log, struct btrfs_path *path,
3172                           struct inode *inode)
3173 {
3174         struct btrfs_inode_item *inode_item;
3175         struct btrfs_key key;
3176         int ret;
3177
3178         memcpy(&key, &BTRFS_I(inode)->location, sizeof(key));
3179         ret = btrfs_insert_empty_item(trans, log, path, &key,
3180                                       sizeof(*inode_item));
3181         if (ret && ret != -EEXIST)
3182                 return ret;
3183         inode_item = btrfs_item_ptr(path->nodes[0], path->slots[0],
3184                                     struct btrfs_inode_item);
3185         fill_inode_item(trans, path->nodes[0], inode_item, inode, 0);
3186         btrfs_release_path(path);
3187         return 0;
3188 }
3189
3190 static noinline int copy_items(struct btrfs_trans_handle *trans,
3191                                struct inode *inode,
3192                                struct btrfs_path *dst_path,
3193                                struct extent_buffer *src,
3194                                int start_slot, int nr, int inode_only)
3195 {
3196         unsigned long src_offset;
3197         unsigned long dst_offset;
3198         struct btrfs_root *log = BTRFS_I(inode)->root->log_root;
3199         struct btrfs_file_extent_item *extent;
3200         struct btrfs_inode_item *inode_item;
3201         int ret;
3202         struct btrfs_key *ins_keys;
3203         u32 *ins_sizes;
3204         char *ins_data;
3205         int i;
3206         struct list_head ordered_sums;
3207         int skip_csum = BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM;
3208
3209         INIT_LIST_HEAD(&ordered_sums);
3210
3211         ins_data = kmalloc(nr * sizeof(struct btrfs_key) +
3212                            nr * sizeof(u32), GFP_NOFS);
3213         if (!ins_data)
3214                 return -ENOMEM;
3215
3216         ins_sizes = (u32 *)ins_data;
3217         ins_keys = (struct btrfs_key *)(ins_data + nr * sizeof(u32));
3218
3219         for (i = 0; i < nr; i++) {
3220                 ins_sizes[i] = btrfs_item_size_nr(src, i + start_slot);
3221                 btrfs_item_key_to_cpu(src, ins_keys + i, i + start_slot);
3222         }
3223         ret = btrfs_insert_empty_items(trans, log, dst_path,
3224                                        ins_keys, ins_sizes, nr);
3225         if (ret) {
3226                 kfree(ins_data);
3227                 return ret;
3228         }
3229
3230         for (i = 0; i < nr; i++, dst_path->slots[0]++) {
3231                 dst_offset = btrfs_item_ptr_offset(dst_path->nodes[0],
3232                                                    dst_path->slots[0]);
3233
3234                 src_offset = btrfs_item_ptr_offset(src, start_slot + i);
3235
3236                 if (ins_keys[i].type == BTRFS_INODE_ITEM_KEY) {
3237                         inode_item = btrfs_item_ptr(dst_path->nodes[0],
3238                                                     dst_path->slots[0],
3239                                                     struct btrfs_inode_item);
3240                         fill_inode_item(trans, dst_path->nodes[0], inode_item,
3241                                         inode, inode_only == LOG_INODE_EXISTS);
3242                 } else {
3243                         copy_extent_buffer(dst_path->nodes[0], src, dst_offset,
3244                                            src_offset, ins_sizes[i]);
3245                 }
3246
3247                 /* take a reference on file data extents so that truncates
3248                  * or deletes of this inode don't have to relog the inode
3249                  * again
3250                  */
3251                 if (btrfs_key_type(ins_keys + i) == BTRFS_EXTENT_DATA_KEY &&
3252                     !skip_csum) {
3253                         int found_type;
3254                         extent = btrfs_item_ptr(src, start_slot + i,
3255                                                 struct btrfs_file_extent_item);
3256
3257                         if (btrfs_file_extent_generation(src, extent) < trans->transid)
3258                                 continue;
3259
3260                         found_type = btrfs_file_extent_type(src, extent);
3261                         if (found_type == BTRFS_FILE_EXTENT_REG) {
3262                                 u64 ds, dl, cs, cl;
3263                                 ds = btrfs_file_extent_disk_bytenr(src,
3264                                                                 extent);
3265                                 /* ds == 0 is a hole */
3266                                 if (ds == 0)
3267                                         continue;
3268
3269                                 dl = btrfs_file_extent_disk_num_bytes(src,
3270                                                                 extent);
3271                                 cs = btrfs_file_extent_offset(src, extent);
3272                                 cl = btrfs_file_extent_num_bytes(src,
3273                                                                 extent);
3274                                 if (btrfs_file_extent_compression(src,
3275                                                                   extent)) {
3276                                         cs = 0;
3277                                         cl = dl;
3278                                 }
3279
3280                                 ret = btrfs_lookup_csums_range(
3281                                                 log->fs_info->csum_root,
3282                                                 ds + cs, ds + cs + cl - 1,
3283                                                 &ordered_sums, 0);
3284                                 if (ret) {
3285                                         btrfs_release_path(dst_path);
3286                                         kfree(ins_data);
3287                                         return ret;
3288                                 }
3289                         }
3290                 }
3291         }
3292
3293         btrfs_mark_buffer_dirty(dst_path->nodes[0]);
3294         btrfs_release_path(dst_path);
3295         kfree(ins_data);
3296
3297         /*
3298          * we have to do this after the loop above to avoid changing the
3299          * log tree while trying to change the log tree.
3300          */
3301         ret = 0;
3302         while (!list_empty(&ordered_sums)) {
3303                 struct btrfs_ordered_sum *sums = list_entry(ordered_sums.next,
3304                                                    struct btrfs_ordered_sum,
3305                                                    list);
3306                 if (!ret)
3307                         ret = btrfs_csum_file_blocks(trans, log, sums);
3308                 list_del(&sums->list);
3309                 kfree(sums);
3310         }
3311         return ret;
3312 }
3313
3314 static int extent_cmp(void *priv, struct list_head *a, struct list_head *b)
3315 {
3316         struct extent_map *em1, *em2;
3317
3318         em1 = list_entry(a, struct extent_map, list);
3319         em2 = list_entry(b, struct extent_map, list);
3320
3321         if (em1->start < em2->start)
3322                 return -1;
3323         else if (em1->start > em2->start)
3324                 return 1;
3325         return 0;
3326 }
3327
3328 static int log_one_extent(struct btrfs_trans_handle *trans,
3329                           struct inode *inode, struct btrfs_root *root,
3330                           struct extent_map *em, struct btrfs_path *path)
3331 {
3332         struct btrfs_root *log = root->log_root;
3333         struct btrfs_file_extent_item *fi;
3334         struct extent_buffer *leaf;
3335         struct btrfs_ordered_extent *ordered;
3336         struct list_head ordered_sums;
3337         struct btrfs_map_token token;
3338         struct btrfs_key key;
3339         u64 mod_start = em->mod_start;
3340         u64 mod_len = em->mod_len;
3341         u64 csum_offset;
3342         u64 csum_len;
3343         u64 extent_offset = em->start - em->orig_start;
3344         u64 block_len;
3345         int ret;
3346         int index = log->log_transid % 2;
3347         bool skip_csum = BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM;
3348
3349         ret = __btrfs_drop_extents(trans, log, inode, path, em->start,
3350                                    em->start + em->len, NULL, 0);
3351         if (ret)
3352                 return ret;
3353
3354         INIT_LIST_HEAD(&ordered_sums);
3355         btrfs_init_map_token(&token);
3356         key.objectid = btrfs_ino(inode);
3357         key.type = BTRFS_EXTENT_DATA_KEY;
3358         key.offset = em->start;
3359
3360         ret = btrfs_insert_empty_item(trans, log, path, &key, sizeof(*fi));
3361         if (ret)
3362                 return ret;
3363         leaf = path->nodes[0];
3364         fi = btrfs_item_ptr(leaf, path->slots[0],
3365                             struct btrfs_file_extent_item);
3366
3367         btrfs_set_token_file_extent_generation(leaf, fi, em->generation,
3368                                                &token);
3369         if (test_bit(EXTENT_FLAG_PREALLOC, &em->flags)) {
3370                 skip_csum = true;
3371                 btrfs_set_token_file_extent_type(leaf, fi,
3372                                                  BTRFS_FILE_EXTENT_PREALLOC,
3373                                                  &token);
3374         } else {
3375                 btrfs_set_token_file_extent_type(leaf, fi,
3376                                                  BTRFS_FILE_EXTENT_REG,
3377                                                  &token);
3378                 if (em->block_start == 0)
3379                         skip_csum = true;
3380         }
3381
3382         block_len = max(em->block_len, em->orig_block_len);
3383         if (em->compress_type != BTRFS_COMPRESS_NONE) {
3384                 btrfs_set_token_file_extent_disk_bytenr(leaf, fi,
3385                                                         em->block_start,
3386                                                         &token);
3387                 btrfs_set_token_file_extent_disk_num_bytes(leaf, fi, block_len,
3388                                                            &token);
3389         } else if (em->block_start < EXTENT_MAP_LAST_BYTE) {
3390                 btrfs_set_token_file_extent_disk_bytenr(leaf, fi,
3391                                                         em->block_start -
3392                                                         extent_offset, &token);
3393                 btrfs_set_token_file_extent_disk_num_bytes(leaf, fi, block_len,
3394                                                            &token);
3395         } else {
3396                 btrfs_set_token_file_extent_disk_bytenr(leaf, fi, 0, &token);
3397                 btrfs_set_token_file_extent_disk_num_bytes(leaf, fi, 0,
3398                                                            &token);
3399         }
3400
3401         btrfs_set_token_file_extent_offset(leaf, fi,
3402                                            em->start - em->orig_start,
3403                                            &token);
3404         btrfs_set_token_file_extent_num_bytes(leaf, fi, em->len, &token);
3405         btrfs_set_token_file_extent_ram_bytes(leaf, fi, em->ram_bytes, &token);
3406         btrfs_set_token_file_extent_compression(leaf, fi, em->compress_type,
3407                                                 &token);
3408         btrfs_set_token_file_extent_encryption(leaf, fi, 0, &token);
3409         btrfs_set_token_file_extent_other_encoding(leaf, fi, 0, &token);
3410         btrfs_mark_buffer_dirty(leaf);
3411
3412         btrfs_release_path(path);
3413         if (ret) {
3414                 return ret;
3415         }
3416
3417         if (skip_csum)
3418                 return 0;
3419
3420         if (em->compress_type) {
3421                 csum_offset = 0;
3422                 csum_len = block_len;
3423         }
3424
3425         /*
3426          * First check and see if our csums are on our outstanding ordered
3427          * extents.
3428          */
3429 again:
3430         spin_lock_irq(&log->log_extents_lock[index]);
3431         list_for_each_entry(ordered, &log->logged_list[index], log_list) {
3432                 struct btrfs_ordered_sum *sum;
3433
3434                 if (!mod_len)
3435                         break;
3436
3437                 if (ordered->inode != inode)
3438                         continue;
3439
3440                 if (ordered->file_offset + ordered->len <= mod_start ||
3441                     mod_start + mod_len <= ordered->file_offset)
3442                         continue;
3443
3444                 /*
3445                  * We are going to copy all the csums on this ordered extent, so
3446                  * go ahead and adjust mod_start and mod_len in case this
3447                  * ordered extent has already been logged.
3448                  */
3449                 if (ordered->file_offset > mod_start) {
3450                         if (ordered->file_offset + ordered->len >=
3451                             mod_start + mod_len)
3452                                 mod_len = ordered->file_offset - mod_start;
3453                         /*
3454                          * If we have this case
3455                          *
3456                          * |--------- logged extent ---------|
3457                          *       |----- ordered extent ----|
3458                          *
3459                          * Just don't mess with mod_start and mod_len, we'll
3460                          * just end up logging more csums than we need and it
3461                          * will be ok.
3462                          */
3463                 } else {
3464                         if (ordered->file_offset + ordered->len <
3465                             mod_start + mod_len) {
3466                                 mod_len = (mod_start + mod_len) -
3467                                         (ordered->file_offset + ordered->len);
3468                                 mod_start = ordered->file_offset +
3469                                         ordered->len;
3470                         } else {
3471                                 mod_len = 0;
3472                         }
3473                 }
3474
3475                 /*
3476                  * To keep us from looping for the above case of an ordered
3477                  * extent that falls inside of the logged extent.
3478                  */
3479                 if (test_and_set_bit(BTRFS_ORDERED_LOGGED_CSUM,
3480                                      &ordered->flags))
3481                         continue;
3482                 atomic_inc(&ordered->refs);
3483                 spin_unlock_irq(&log->log_extents_lock[index]);
3484                 /*
3485                  * we've dropped the lock, we must either break or
3486                  * start over after this.
3487                  */
3488
3489                 wait_event(ordered->wait, ordered->csum_bytes_left == 0);
3490
3491                 list_for_each_entry(sum, &ordered->list, list) {
3492                         ret = btrfs_csum_file_blocks(trans, log, sum);
3493                         if (ret) {
3494                                 btrfs_put_ordered_extent(ordered);
3495                                 goto unlocked;
3496                         }
3497                 }
3498                 btrfs_put_ordered_extent(ordered);
3499                 goto again;
3500
3501         }
3502         spin_unlock_irq(&log->log_extents_lock[index]);
3503 unlocked:
3504
3505         if (!mod_len || ret)
3506                 return ret;
3507
3508         csum_offset = mod_start - em->start;
3509         csum_len = mod_len;
3510
3511         /* block start is already adjusted for the file extent offset. */
3512         ret = btrfs_lookup_csums_range(log->fs_info->csum_root,
3513                                        em->block_start + csum_offset,
3514                                        em->block_start + csum_offset +
3515                                        csum_len - 1, &ordered_sums, 0);
3516         if (ret)
3517                 return ret;
3518
3519         while (!list_empty(&ordered_sums)) {
3520                 struct btrfs_ordered_sum *sums = list_entry(ordered_sums.next,
3521                                                    struct btrfs_ordered_sum,
3522                                                    list);
3523                 if (!ret)
3524                         ret = btrfs_csum_file_blocks(trans, log, sums);
3525                 list_del(&sums->list);
3526                 kfree(sums);
3527         }
3528
3529         return ret;
3530 }
3531
3532 static int btrfs_log_changed_extents(struct btrfs_trans_handle *trans,
3533                                      struct btrfs_root *root,
3534                                      struct inode *inode,
3535                                      struct btrfs_path *path)
3536 {
3537         struct extent_map *em, *n;
3538         struct list_head extents;
3539         struct extent_map_tree *tree = &BTRFS_I(inode)->extent_tree;
3540         u64 test_gen;
3541         int ret = 0;
3542         int num = 0;
3543
3544         INIT_LIST_HEAD(&extents);
3545
3546         write_lock(&tree->lock);
3547         test_gen = root->fs_info->last_trans_committed;
3548
3549         list_for_each_entry_safe(em, n, &tree->modified_extents, list) {
3550                 list_del_init(&em->list);
3551
3552                 /*
3553                  * Just an arbitrary number, this can be really CPU intensive
3554                  * once we start getting a lot of extents, and really once we
3555                  * have a bunch of extents we just want to commit since it will
3556                  * be faster.
3557                  */
3558                 if (++num > 32768) {
3559                         list_del_init(&tree->modified_extents);
3560                         ret = -EFBIG;
3561                         goto process;
3562                 }
3563
3564                 if (em->generation <= test_gen)
3565                         continue;
3566                 /* Need a ref to keep it from getting evicted from cache */
3567                 atomic_inc(&em->refs);
3568                 set_bit(EXTENT_FLAG_LOGGING, &em->flags);
3569                 list_add_tail(&em->list, &extents);
3570                 num++;
3571         }
3572
3573         list_sort(NULL, &extents, extent_cmp);
3574
3575 process:
3576         while (!list_empty(&extents)) {
3577                 em = list_entry(extents.next, struct extent_map, list);
3578
3579                 list_del_init(&em->list);
3580
3581                 /*
3582                  * If we had an error we just need to delete everybody from our
3583                  * private list.
3584                  */
3585                 if (ret) {
3586                         clear_em_logging(tree, em);
3587                         free_extent_map(em);
3588                         continue;
3589                 }
3590
3591                 write_unlock(&tree->lock);
3592
3593                 ret = log_one_extent(trans, inode, root, em, path);
3594                 write_lock(&tree->lock);
3595                 clear_em_logging(tree, em);
3596                 free_extent_map(em);
3597         }
3598         WARN_ON(!list_empty(&extents));
3599         write_unlock(&tree->lock);
3600
3601         btrfs_release_path(path);
3602         return ret;
3603 }
3604
3605 /* log a single inode in the tree log.
3606  * At least one parent directory for this inode must exist in the tree
3607  * or be logged already.
3608  *
3609  * Any items from this inode changed by the current transaction are copied
3610  * to the log tree.  An extra reference is taken on any extents in this
3611  * file, allowing us to avoid a whole pile of corner cases around logging
3612  * blocks that have been removed from the tree.
3613  *
3614  * See LOG_INODE_ALL and related defines for a description of what inode_only
3615  * does.
3616  *
3617  * This handles both files and directories.
3618  */
3619 static int btrfs_log_inode(struct btrfs_trans_handle *trans,
3620                              struct btrfs_root *root, struct inode *inode,
3621                              int inode_only)
3622 {
3623         struct btrfs_path *path;
3624         struct btrfs_path *dst_path;
3625         struct btrfs_key min_key;
3626         struct btrfs_key max_key;
3627         struct btrfs_root *log = root->log_root;
3628         struct extent_buffer *src = NULL;
3629         int err = 0;
3630         int ret;
3631         int nritems;
3632         int ins_start_slot = 0;
3633         int ins_nr;
3634         bool fast_search = false;
3635         u64 ino = btrfs_ino(inode);
3636
3637         path = btrfs_alloc_path();
3638         if (!path)
3639                 return -ENOMEM;
3640         dst_path = btrfs_alloc_path();
3641         if (!dst_path) {
3642                 btrfs_free_path(path);
3643                 return -ENOMEM;
3644         }
3645
3646         min_key.objectid = ino;
3647         min_key.type = BTRFS_INODE_ITEM_KEY;
3648         min_key.offset = 0;
3649
3650         max_key.objectid = ino;
3651
3652
3653         /* today the code can only do partial logging of directories */
3654         if (S_ISDIR(inode->i_mode) ||
3655             (!test_bit(BTRFS_INODE_NEEDS_FULL_SYNC,
3656                        &BTRFS_I(inode)->runtime_flags) &&
3657              inode_only == LOG_INODE_EXISTS))
3658                 max_key.type = BTRFS_XATTR_ITEM_KEY;
3659         else
3660                 max_key.type = (u8)-1;
3661         max_key.offset = (u64)-1;
3662
3663         /* Only run delayed items if we are a dir or a new file */
3664         if (S_ISDIR(inode->i_mode) ||
3665             BTRFS_I(inode)->generation > root->fs_info->last_trans_committed) {
3666                 ret = btrfs_commit_inode_delayed_items(trans, inode);
3667                 if (ret) {
3668                         btrfs_free_path(path);
3669                         btrfs_free_path(dst_path);
3670                         return ret;
3671                 }
3672         }
3673
3674         mutex_lock(&BTRFS_I(inode)->log_mutex);
3675
3676         btrfs_get_logged_extents(log, inode);
3677
3678         /*
3679          * a brute force approach to making sure we get the most uptodate
3680          * copies of everything.
3681          */
3682         if (S_ISDIR(inode->i_mode)) {
3683                 int max_key_type = BTRFS_DIR_LOG_INDEX_KEY;
3684
3685                 if (inode_only == LOG_INODE_EXISTS)
3686                         max_key_type = BTRFS_XATTR_ITEM_KEY;
3687                 ret = drop_objectid_items(trans, log, path, ino, max_key_type);
3688         } else {
3689                 if (test_and_clear_bit(BTRFS_INODE_NEEDS_FULL_SYNC,
3690                                        &BTRFS_I(inode)->runtime_flags)) {
3691                         clear_bit(BTRFS_INODE_COPY_EVERYTHING,
3692                                   &BTRFS_I(inode)->runtime_flags);
3693                         ret = btrfs_truncate_inode_items(trans, log,
3694                                                          inode, 0, 0);
3695                 } else if (test_and_clear_bit(BTRFS_INODE_COPY_EVERYTHING,
3696                                               &BTRFS_I(inode)->runtime_flags)) {
3697                         if (inode_only == LOG_INODE_ALL)
3698                                 fast_search = true;
3699                         max_key.type = BTRFS_XATTR_ITEM_KEY;
3700                         ret = drop_objectid_items(trans, log, path, ino,
3701                                                   max_key.type);
3702                 } else {
3703                         if (inode_only == LOG_INODE_ALL)
3704                                 fast_search = true;
3705                         ret = log_inode_item(trans, log, dst_path, inode);
3706                         if (ret) {
3707                                 err = ret;
3708                                 goto out_unlock;
3709                         }
3710                         goto log_extents;
3711                 }
3712
3713         }
3714         if (ret) {
3715                 err = ret;
3716                 goto out_unlock;
3717         }
3718         path->keep_locks = 1;
3719
3720         while (1) {
3721                 ins_nr = 0;
3722                 ret = btrfs_search_forward(root, &min_key, &max_key,
3723                                            path, trans->transid);
3724                 if (ret != 0)
3725                         break;
3726 again:
3727                 /* note, ins_nr might be > 0 here, cleanup outside the loop */
3728                 if (min_key.objectid != ino)
3729                         break;
3730                 if (min_key.type > max_key.type)
3731                         break;
3732
3733                 src = path->nodes[0];
3734                 if (ins_nr && ins_start_slot + ins_nr == path->slots[0]) {
3735                         ins_nr++;
3736                         goto next_slot;
3737                 } else if (!ins_nr) {
3738                         ins_start_slot = path->slots[0];
3739                         ins_nr = 1;
3740                         goto next_slot;
3741                 }
3742
3743                 ret = copy_items(trans, inode, dst_path, src, ins_start_slot,
3744                                  ins_nr, inode_only);
3745                 if (ret) {
3746                         err = ret;
3747                         goto out_unlock;
3748                 }
3749                 ins_nr = 1;
3750                 ins_start_slot = path->slots[0];
3751 next_slot:
3752
3753                 nritems = btrfs_header_nritems(path->nodes[0]);
3754                 path->slots[0]++;
3755                 if (path->slots[0] < nritems) {
3756                         btrfs_item_key_to_cpu(path->nodes[0], &min_key,
3757                                               path->slots[0]);
3758                         goto again;
3759                 }
3760                 if (ins_nr) {
3761                         ret = copy_items(trans, inode, dst_path, src,
3762                                          ins_start_slot,
3763                                          ins_nr, inode_only);
3764                         if (ret) {
3765                                 err = ret;
3766                                 goto out_unlock;
3767                         }
3768                         ins_nr = 0;
3769                 }
3770                 btrfs_release_path(path);
3771
3772                 if (min_key.offset < (u64)-1)
3773                         min_key.offset++;
3774                 else if (min_key.type < (u8)-1)
3775                         min_key.type++;
3776                 else if (min_key.objectid < (u64)-1)
3777                         min_key.objectid++;
3778                 else
3779                         break;
3780         }
3781         if (ins_nr) {
3782                 ret = copy_items(trans, inode, dst_path, src, ins_start_slot,
3783                                  ins_nr, inode_only);
3784                 if (ret) {
3785                         err = ret;
3786                         goto out_unlock;
3787                 }
3788                 ins_nr = 0;
3789         }
3790
3791 log_extents:
3792         btrfs_release_path(path);
3793         btrfs_release_path(dst_path);
3794         if (fast_search) {
3795                 ret = btrfs_log_changed_extents(trans, root, inode, dst_path);
3796                 if (ret) {
3797                         err = ret;
3798                         goto out_unlock;
3799                 }
3800         } else {
3801                 struct extent_map_tree *tree = &BTRFS_I(inode)->extent_tree;
3802                 struct extent_map *em, *n;
3803
3804                 write_lock(&tree->lock);
3805                 list_for_each_entry_safe(em, n, &tree->modified_extents, list)
3806                         list_del_init(&em->list);
3807                 write_unlock(&tree->lock);
3808         }
3809
3810         if (inode_only == LOG_INODE_ALL && S_ISDIR(inode->i_mode)) {
3811                 ret = log_directory_changes(trans, root, inode, path, dst_path);
3812                 if (ret) {
3813                         err = ret;
3814                         goto out_unlock;
3815                 }
3816         }
3817         BTRFS_I(inode)->logged_trans = trans->transid;
3818         BTRFS_I(inode)->last_log_commit = BTRFS_I(inode)->last_sub_trans;
3819 out_unlock:
3820         if (err)
3821                 btrfs_free_logged_extents(log, log->log_transid);
3822         mutex_unlock(&BTRFS_I(inode)->log_mutex);
3823
3824         btrfs_free_path(path);
3825         btrfs_free_path(dst_path);
3826         return err;
3827 }
3828
3829 /*
3830  * follow the dentry parent pointers up the chain and see if any
3831  * of the directories in it require a full commit before they can
3832  * be logged.  Returns zero if nothing special needs to be done or 1 if
3833  * a full commit is required.
3834  */
3835 static noinline int check_parent_dirs_for_sync(struct btrfs_trans_handle *trans,
3836                                                struct inode *inode,
3837                                                struct dentry *parent,
3838                                                struct super_block *sb,
3839                                                u64 last_committed)
3840 {
3841         int ret = 0;
3842         struct btrfs_root *root;
3843         struct dentry *old_parent = NULL;
3844         struct inode *orig_inode = inode;
3845
3846         /*
3847          * for regular files, if its inode is already on disk, we don't
3848          * have to worry about the parents at all.  This is because
3849          * we can use the last_unlink_trans field to record renames
3850          * and other fun in this file.
3851          */
3852         if (S_ISREG(inode->i_mode) &&
3853             BTRFS_I(inode)->generation <= last_committed &&
3854             BTRFS_I(inode)->last_unlink_trans <= last_committed)
3855                         goto out;
3856
3857         if (!S_ISDIR(inode->i_mode)) {
3858                 if (!parent || !parent->d_inode || sb != parent->d_inode->i_sb)
3859                         goto out;
3860                 inode = parent->d_inode;
3861         }
3862
3863         while (1) {
3864                 /*
3865                  * If we are logging a directory then we start with our inode,
3866                  * not our parents inode, so we need to skipp setting the
3867                  * logged_trans so that further down in the log code we don't
3868                  * think this inode has already been logged.
3869                  */
3870                 if (inode != orig_inode)
3871                         BTRFS_I(inode)->logged_trans = trans->transid;
3872                 smp_mb();
3873
3874                 if (BTRFS_I(inode)->last_unlink_trans > last_committed) {
3875                         root = BTRFS_I(inode)->root;
3876
3877                         /*
3878                          * make sure any commits to the log are forced
3879                          * to be full commits
3880                          */
3881                         root->fs_info->last_trans_log_full_commit =
3882                                 trans->transid;
3883                         ret = 1;
3884                         break;
3885                 }
3886
3887                 if (!parent || !parent->d_inode || sb != parent->d_inode->i_sb)
3888                         break;
3889
3890                 if (IS_ROOT(parent))
3891                         break;
3892
3893                 parent = dget_parent(parent);
3894                 dput(old_parent);
3895                 old_parent = parent;
3896                 inode = parent->d_inode;
3897
3898         }
3899         dput(old_parent);
3900 out:
3901         return ret;
3902 }
3903
3904 /*
3905  * helper function around btrfs_log_inode to make sure newly created
3906  * parent directories also end up in the log.  A minimal inode and backref
3907  * only logging is done of any parent directories that are older than
3908  * the last committed transaction
3909  */
3910 static int btrfs_log_inode_parent(struct btrfs_trans_handle *trans,
3911                                   struct btrfs_root *root, struct inode *inode,
3912                                   struct dentry *parent, int exists_only)
3913 {
3914         int inode_only = exists_only ? LOG_INODE_EXISTS : LOG_INODE_ALL;
3915         struct super_block *sb;
3916         struct dentry *old_parent = NULL;
3917         int ret = 0;
3918         u64 last_committed = root->fs_info->last_trans_committed;
3919
3920         sb = inode->i_sb;
3921
3922         if (btrfs_test_opt(root, NOTREELOG)) {
3923                 ret = 1;
3924                 goto end_no_trans;
3925         }
3926
3927         if (root->fs_info->last_trans_log_full_commit >
3928             root->fs_info->last_trans_committed) {
3929                 ret = 1;
3930                 goto end_no_trans;
3931         }
3932
3933         if (root != BTRFS_I(inode)->root ||
3934             btrfs_root_refs(&root->root_item) == 0) {
3935                 ret = 1;
3936                 goto end_no_trans;
3937         }
3938
3939         ret = check_parent_dirs_for_sync(trans, inode, parent,
3940                                          sb, last_committed);
3941         if (ret)
3942                 goto end_no_trans;
3943
3944         if (btrfs_inode_in_log(inode, trans->transid)) {
3945                 ret = BTRFS_NO_LOG_SYNC;
3946                 goto end_no_trans;
3947         }
3948
3949         ret = start_log_trans(trans, root);
3950         if (ret)
3951                 goto end_trans;
3952
3953         ret = btrfs_log_inode(trans, root, inode, inode_only);
3954         if (ret)
3955                 goto end_trans;
3956
3957         /*
3958          * for regular files, if its inode is already on disk, we don't
3959          * have to worry about the parents at all.  This is because
3960          * we can use the last_unlink_trans field to record renames
3961          * and other fun in this file.
3962          */
3963         if (S_ISREG(inode->i_mode) &&
3964             BTRFS_I(inode)->generation <= last_committed &&
3965             BTRFS_I(inode)->last_unlink_trans <= last_committed) {
3966                 ret = 0;
3967                 goto end_trans;
3968         }
3969
3970         inode_only = LOG_INODE_EXISTS;
3971         while (1) {
3972                 if (!parent || !parent->d_inode || sb != parent->d_inode->i_sb)
3973                         break;
3974
3975                 inode = parent->d_inode;
3976                 if (root != BTRFS_I(inode)->root)
3977                         break;
3978
3979                 if (BTRFS_I(inode)->generation >
3980                     root->fs_info->last_trans_committed) {
3981                         ret = btrfs_log_inode(trans, root, inode, inode_only);
3982                         if (ret)
3983                                 goto end_trans;
3984                 }
3985                 if (IS_ROOT(parent))
3986                         break;
3987
3988                 parent = dget_parent(parent);
3989                 dput(old_parent);
3990                 old_parent = parent;
3991         }
3992         ret = 0;
3993 end_trans:
3994         dput(old_parent);
3995         if (ret < 0) {
3996                 root->fs_info->last_trans_log_full_commit = trans->transid;
3997                 ret = 1;
3998         }
3999         btrfs_end_log_trans(root);
4000 end_no_trans:
4001         return ret;
4002 }
4003
4004 /*
4005  * it is not safe to log dentry if the chunk root has added new
4006  * chunks.  This returns 0 if the dentry was logged, and 1 otherwise.
4007  * If this returns 1, you must commit the transaction to safely get your
4008  * data on disk.
4009  */
4010 int btrfs_log_dentry_safe(struct btrfs_trans_handle *trans,
4011                           struct btrfs_root *root, struct dentry *dentry)
4012 {
4013         struct dentry *parent = dget_parent(dentry);
4014         int ret;
4015
4016         ret = btrfs_log_inode_parent(trans, root, dentry->d_inode, parent, 0);
4017         dput(parent);
4018
4019         return ret;
4020 }
4021
4022 /*
4023  * should be called during mount to recover any replay any log trees
4024  * from the FS
4025  */
4026 int btrfs_recover_log_trees(struct btrfs_root *log_root_tree)
4027 {
4028         int ret;
4029         struct btrfs_path *path;
4030         struct btrfs_trans_handle *trans;
4031         struct btrfs_key key;
4032         struct btrfs_key found_key;
4033         struct btrfs_key tmp_key;
4034         struct btrfs_root *log;
4035         struct btrfs_fs_info *fs_info = log_root_tree->fs_info;
4036         struct walk_control wc = {
4037                 .process_func = process_one_buffer,
4038                 .stage = 0,
4039         };
4040
4041         path = btrfs_alloc_path();
4042         if (!path)
4043                 return -ENOMEM;
4044
4045         fs_info->log_root_recovering = 1;
4046
4047         trans = btrfs_start_transaction(fs_info->tree_root, 0);
4048         if (IS_ERR(trans)) {
4049                 ret = PTR_ERR(trans);
4050                 goto error;
4051         }
4052
4053         wc.trans = trans;
4054         wc.pin = 1;
4055
4056         ret = walk_log_tree(trans, log_root_tree, &wc);
4057         if (ret) {
4058                 btrfs_error(fs_info, ret, "Failed to pin buffers while "
4059                             "recovering log root tree.");
4060                 goto error;
4061         }
4062
4063 again:
4064         key.objectid = BTRFS_TREE_LOG_OBJECTID;
4065         key.offset = (u64)-1;
4066         btrfs_set_key_type(&key, BTRFS_ROOT_ITEM_KEY);
4067
4068         while (1) {
4069                 ret = btrfs_search_slot(NULL, log_root_tree, &key, path, 0, 0);
4070
4071                 if (ret < 0) {
4072                         btrfs_error(fs_info, ret,
4073                                     "Couldn't find tree log root.");
4074                         goto error;
4075                 }
4076                 if (ret > 0) {
4077                         if (path->slots[0] == 0)
4078                                 break;
4079                         path->slots[0]--;
4080                 }
4081                 btrfs_item_key_to_cpu(path->nodes[0], &found_key,
4082                                       path->slots[0]);
4083                 btrfs_release_path(path);
4084                 if (found_key.objectid != BTRFS_TREE_LOG_OBJECTID)
4085                         break;
4086
4087                 log = btrfs_read_fs_root(log_root_tree, &found_key);
4088                 if (IS_ERR(log)) {
4089                         ret = PTR_ERR(log);
4090                         btrfs_error(fs_info, ret,
4091                                     "Couldn't read tree log root.");
4092                         goto error;
4093                 }
4094
4095                 tmp_key.objectid = found_key.offset;
4096                 tmp_key.type = BTRFS_ROOT_ITEM_KEY;
4097                 tmp_key.offset = (u64)-1;
4098
4099                 wc.replay_dest = btrfs_read_fs_root_no_name(fs_info, &tmp_key);
4100                 if (IS_ERR(wc.replay_dest)) {
4101                         ret = PTR_ERR(wc.replay_dest);
4102                         free_extent_buffer(log->node);
4103                         free_extent_buffer(log->commit_root);
4104                         kfree(log);
4105                         btrfs_error(fs_info, ret, "Couldn't read target root "
4106                                     "for tree log recovery.");
4107                         goto error;
4108                 }
4109
4110                 wc.replay_dest->log_root = log;
4111                 btrfs_record_root_in_trans(trans, wc.replay_dest);
4112                 ret = walk_log_tree(trans, log, &wc);
4113
4114                 if (!ret && wc.stage == LOG_WALK_REPLAY_ALL) {
4115                         ret = fixup_inode_link_counts(trans, wc.replay_dest,
4116                                                       path);
4117                 }
4118
4119                 key.offset = found_key.offset - 1;
4120                 wc.replay_dest->log_root = NULL;
4121                 free_extent_buffer(log->node);
4122                 free_extent_buffer(log->commit_root);
4123                 kfree(log);
4124
4125                 if (ret)
4126                         goto error;
4127
4128                 if (found_key.offset == 0)
4129                         break;
4130         }
4131         btrfs_release_path(path);
4132
4133         /* step one is to pin it all, step two is to replay just inodes */
4134         if (wc.pin) {
4135                 wc.pin = 0;
4136                 wc.process_func = replay_one_buffer;
4137                 wc.stage = LOG_WALK_REPLAY_INODES;
4138                 goto again;
4139         }
4140         /* step three is to replay everything */
4141         if (wc.stage < LOG_WALK_REPLAY_ALL) {
4142                 wc.stage++;
4143                 goto again;
4144         }
4145
4146         btrfs_free_path(path);
4147
4148         /* step 4: commit the transaction, which also unpins the blocks */
4149         ret = btrfs_commit_transaction(trans, fs_info->tree_root);
4150         if (ret)
4151                 return ret;
4152
4153         free_extent_buffer(log_root_tree->node);
4154         log_root_tree->log_root = NULL;
4155         fs_info->log_root_recovering = 0;
4156         kfree(log_root_tree);
4157
4158         return 0;
4159 error:
4160         if (wc.trans)
4161                 btrfs_end_transaction(wc.trans, fs_info->tree_root);
4162         btrfs_free_path(path);
4163         return ret;
4164 }
4165
4166 /*
4167  * there are some corner cases where we want to force a full
4168  * commit instead of allowing a directory to be logged.
4169  *
4170  * They revolve around files there were unlinked from the directory, and
4171  * this function updates the parent directory so that a full commit is
4172  * properly done if it is fsync'd later after the unlinks are done.
4173  */
4174 void btrfs_record_unlink_dir(struct btrfs_trans_handle *trans,
4175                              struct inode *dir, struct inode *inode,
4176                              int for_rename)
4177 {
4178         /*
4179          * when we're logging a file, if it hasn't been renamed
4180          * or unlinked, and its inode is fully committed on disk,
4181          * we don't have to worry about walking up the directory chain
4182          * to log its parents.
4183          *
4184          * So, we use the last_unlink_trans field to put this transid
4185          * into the file.  When the file is logged we check it and
4186          * don't log the parents if the file is fully on disk.
4187          */
4188         if (S_ISREG(inode->i_mode))
4189                 BTRFS_I(inode)->last_unlink_trans = trans->transid;
4190
4191         /*
4192          * if this directory was already logged any new
4193          * names for this file/dir will get recorded
4194          */
4195         smp_mb();
4196         if (BTRFS_I(dir)->logged_trans == trans->transid)
4197                 return;
4198
4199         /*
4200          * if the inode we're about to unlink was logged,
4201          * the log will be properly updated for any new names
4202          */
4203         if (BTRFS_I(inode)->logged_trans == trans->transid)
4204                 return;
4205
4206         /*
4207          * when renaming files across directories, if the directory
4208          * there we're unlinking from gets fsync'd later on, there's
4209          * no way to find the destination directory later and fsync it
4210          * properly.  So, we have to be conservative and force commits
4211          * so the new name gets discovered.
4212          */
4213         if (for_rename)
4214                 goto record;
4215
4216         /* we can safely do the unlink without any special recording */
4217         return;
4218
4219 record:
4220         BTRFS_I(dir)->last_unlink_trans = trans->transid;
4221 }
4222
4223 /*
4224  * Call this after adding a new name for a file and it will properly
4225  * update the log to reflect the new name.
4226  *
4227  * It will return zero if all goes well, and it will return 1 if a
4228  * full transaction commit is required.
4229  */
4230 int btrfs_log_new_name(struct btrfs_trans_handle *trans,
4231                         struct inode *inode, struct inode *old_dir,
4232                         struct dentry *parent)
4233 {
4234         struct btrfs_root * root = BTRFS_I(inode)->root;
4235
4236         /*
4237          * this will force the logging code to walk the dentry chain
4238          * up for the file
4239          */
4240         if (S_ISREG(inode->i_mode))
4241                 BTRFS_I(inode)->last_unlink_trans = trans->transid;
4242
4243         /*
4244          * if this inode hasn't been logged and directory we're renaming it
4245          * from hasn't been logged, we don't need to log it
4246          */
4247         if (BTRFS_I(inode)->logged_trans <=
4248             root->fs_info->last_trans_committed &&
4249             (!old_dir || BTRFS_I(old_dir)->logged_trans <=
4250                     root->fs_info->last_trans_committed))
4251                 return 0;
4252
4253         return btrfs_log_inode_parent(trans, root, inode, parent, 1);
4254 }
4255