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 "tree-log.h"
  30 #include "hash.h"
  31
  32 /* magic values for the inode_only field in btrfs_log_inode:
  33  *
  34  * LOG_INODE_ALL means to log everything
  35  * LOG_INODE_EXISTS means to log just enough to recreate the inode
  36  * during log replay
  37  */
  38 #define LOG_INODE_ALL 0
  39 #define LOG_INODE_EXISTS 1
  40
  41 /*
  42  * directory trouble cases
  43  *
  44  * 1) on rename or unlink, if the inode being unlinked isn't in the fsync
  45  * log, we must force a full commit before doing an fsync of the directory
  46  * where the unlink was done.
  47  * ---> record transid of last unlink/rename per directory
  48  *
  49  * mkdir foo/some_dir
  50  * normal commit
  51  * rename foo/some_dir foo2/some_dir
  52  * mkdir foo/some_dir
  53  * fsync foo/some_dir/some_file
  54  *
  55  * The fsync above will unlink the original some_dir without recording
  56  * it in its new location (foo2).  After a crash, some_dir will be gone
  57  * unless the fsync of some_file forces a full commit
  58  *
  59  * 2) we must log any new names for any file or dir that is in the fsync
  60  * log. ---> check inode while renaming/linking.
  61  *
  62  * 2a) we must log any new names for any file or dir during rename
  63  * when the directory they are being removed from was logged.
  64  * ---> check inode and old parent dir during rename
  65  *
  66  *  2a is actually the more important variant.  With the extra logging
  67  *  a crash might unlink the old name without recreating the new one
  68  *
  69  * 3) after a crash, we must go through any directories with a link count
  70  * of zero and redo the rm -rf
  71  *
  72  * mkdir f1/foo
  73  * normal commit
  74  * rm -rf f1/foo
  75  * fsync(f1)
  76  *
  77  * The directory f1 was fully removed from the FS, but fsync was never
  78  * called on f1, only its parent dir.  After a crash the rm -rf must
  79  * be replayed.  This must be able to recurse down the entire
  80  * directory tree.  The inode link count fixup code takes care of the
  81  * ugly details.
  82  */
  83
  84 /*
  85  * stages for the tree walking.  The first
  86  * stage (0) is to only pin down the blocks we find
  87  * the second stage (1) is to make sure that all the inodes
  88  * we find in the log are created in the subvolume.
  89  *
  90  * The last stage is to deal with directories and links and extents
  91  * and all the other fun semantics
  92  */
  93 #define LOG_WALK_PIN_ONLY 0
  94 #define LOG_WALK_REPLAY_INODES 1
  95 #define LOG_WALK_REPLAY_DIR_INDEX 2
  96 #define LOG_WALK_REPLAY_ALL 3
  97
  98 static int btrfs_log_inode(struct btrfs_trans_handle *trans,
  99                              struct btrfs_root *root, struct inode *inode,
 100                              int inode_only);
 101 static int link_to_fixup_dir(struct btrfs_trans_handle *trans,
 102                              struct btrfs_root *root,
 103                              struct btrfs_path *path, u64 objectid);
 104 static noinline int replay_dir_deletes(struct btrfs_trans_handle *trans,
 105                                        struct btrfs_root *root,
 106                                        struct btrfs_root *log,
 107                                        struct btrfs_path *path,
 108                                        u64 dirid, int del_all);
 109
 110 /*
 111  * tree logging is a special write ahead log used to make sure that
 112  * fsyncs and O_SYNCs can happen without doing full tree commits.
 113  *
 114  * Full tree commits are expensive because they require commonly
 115  * modified blocks to be recowed, creating many dirty pages in the
 116  * extent tree an 4x-6x higher write load than ext3.
 117  *
 118  * Instead of doing a tree commit on every fsync, we use the
 119  * key ranges and transaction ids to find items for a given file or directory
 120  * that have changed in this transaction.  Those items are copied into
 121  * a special tree (one per subvolume root), that tree is written to disk
 122  * and then the fsync is considered complete.
 123  *
 124  * After a crash, items are copied out of the log-tree back into the
 125  * subvolume tree.  Any file data extents found are recorded in the extent
 126  * allocation tree, and the log-tree freed.
 127  *
 128  * The log tree is read three times, once to pin down all the extents it is
 129  * using in ram and once, once to create all the inodes logged in the tree
 130  * and once to do all the other items.
 131  */
 132
 133 /*
 134  * start a sub transaction and setup the log tree
 135  * this increments the log tree writer count to make the people
 136  * syncing the tree wait for us to finish
 137  */
 138 static int start_log_trans(struct btrfs_trans_handle *trans,
 139                            struct btrfs_root *root)
 140 {
 141         int ret;
 142         int err = 0;
 143
 144         mutex_lock(&root->log_mutex);
 145         if (root->log_root) {
 146                 if (!root->log_start_pid) {
 147                         root->log_start_pid = current->pid;
 148                         root->log_multiple_pids = false;
 149                 } else if (root->log_start_pid != current->pid) {
 150                         root->log_multiple_pids = true;
 151                 }
 152
 153                 atomic_inc(&root->log_batch);
 154                 atomic_inc(&root->log_writers);
 155                 mutex_unlock(&root->log_mutex);
 156                 return 0;
 157         }
 158         root->log_multiple_pids = false;
 159         root->log_start_pid = current->pid;
 160         mutex_lock(&root->fs_info->tree_log_mutex);
 161         if (!root->fs_info->log_root_tree) {
 162                 ret = btrfs_init_log_root_tree(trans, root->fs_info);
 163                 if (ret)
 164                         err = ret;
 165         }
 166         if (err == 0 && !root->log_root) {
 167                 ret = btrfs_add_log_tree(trans, root);
 168                 if (ret)
 169                         err = ret;
 170         }
 171         mutex_unlock(&root->fs_info->tree_log_mutex);
 172         atomic_inc(&root->log_batch);
 173         atomic_inc(&root->log_writers);
 174         mutex_unlock(&root->log_mutex);
 175         return err;
 176 }
 177
 178 /*
 179  * returns 0 if there was a log transaction running and we were able
 180  * to join, or returns -ENOENT if there were not transactions
 181  * in progress
 182  */
 183 static int join_running_log_trans(struct btrfs_root *root)
 184 {
 185         int ret = -ENOENT;
 186
 187         smp_mb();
 188         if (!root->log_root)
 189                 return -ENOENT;
 190
 191         mutex_lock(&root->log_mutex);
 192         if (root->log_root) {
 193                 ret = 0;
 194                 atomic_inc(&root->log_writers);
 195         }
 196         mutex_unlock(&root->log_mutex);
 197         return ret;
 198 }
 199
 200 /*
 201  * This either makes the current running log transaction wait
 202  * until you call btrfs_end_log_trans() or it makes any future
 203  * log transactions wait until you call btrfs_end_log_trans()
 204  */
 205 int btrfs_pin_log_trans(struct btrfs_root *root)
 206 {
 207         int ret = -ENOENT;
 208
 209         mutex_lock(&root->log_mutex);
 210         atomic_inc(&root->log_writers);
 211         mutex_unlock(&root->log_mutex);
 212         return ret;
 213 }
 214
 215 /*
 216  * indicate we're done making changes to the log tree
 217  * and wake up anyone waiting to do a sync
 218  */
 219 void btrfs_end_log_trans(struct btrfs_root *root)
 220 {
 221         if (atomic_dec_and_test(&root->log_writers)) {
 222                 smp_mb();
 223                 if (waitqueue_active(&root->log_writer_wait))
 224                         wake_up(&root->log_writer_wait);
 225         }
 226 }
 227
 228
 229 /*
 230  * the walk control struct is used to pass state down the chain when
 231  * processing the log tree.  The stage field tells us which part
 232  * of the log tree processing we are currently doing.  The others
 233  * are state fields used for that specific part
 234  */
 235 struct walk_control {
 236         /* should we free the extent on disk when done?  This is used
 237          * at transaction commit time while freeing a log tree
 238          */
 239         int free;
 240
 241         /* should we write out the extent buffer?  This is used
 242          * while flushing the log tree to disk during a sync
 243          */
 244         int write;
 245
 246         /* should we wait for the extent buffer io to finish?  Also used
 247          * while flushing the log tree to disk for a sync
 248          */
 249         int wait;
 250
 251         /* pin only walk, we record which extents on disk belong to the
 252          * log trees
 253          */
 254         int pin;
 255
 256         /* what stage of the replay code we're currently in */
 257         int stage;
 258
 259         /* the root we are currently replaying */
 260         struct btrfs_root *replay_dest;
 261
 262         /* the trans handle for the current replay */
 263         struct btrfs_trans_handle *trans;
 264
 265         /* the function that gets used to process blocks we find in the
 266          * tree.  Note the extent_buffer might not be up to date when it is
 267          * passed in, and it must be checked or read if you need the data
 268          * inside it
 269          */
 270         int (*process_func)(struct btrfs_root *log, struct extent_buffer *eb,
 271                             struct walk_control *wc, u64 gen);
 272 };
 273
 274 /*
 275  * process_func used to pin down extents, write them or wait on them
 276  */
 277 static int process_one_buffer(struct btrfs_root *log,
 278                               struct extent_buffer *eb,
 279                               struct walk_control *wc, u64 gen)
 280 {
 281         int ret = 0;
 282
 283         /*
 284          * If this fs is mixed then we need to be able to process the leaves to
 285          * pin down any logged extents, so we have to read the block.
 286          */
 287         if (btrfs_fs_incompat(log->fs_info, MIXED_GROUPS)) {
 288                 ret = btrfs_read_buffer(eb, gen);
 289                 if (ret)
 290                         return ret;
 291         }
 292
 293         if (wc->pin)
 294                 ret = btrfs_pin_extent_for_log_replay(log->fs_info->extent_root,
 295                                                       eb->start, eb->len);
 296
 297         if (!ret && btrfs_buffer_uptodate(eb, gen, 0)) {
 298                 if (wc->pin && btrfs_header_level(eb) == 0)
 299                         ret = btrfs_exclude_logged_extents(log, eb);
 300                 if (wc->write)
 301                         btrfs_write_tree_block(eb);
 302                 if (wc->wait)
 303                         btrfs_wait_tree_block_writeback(eb);
 304         }
 305         return ret;
 306 }
 307
 308 /*
 309  * Item overwrite used by replay and tree logging.  eb, slot and key all refer
 310  * to the src data we are copying out.
 311  *
 312  * root is the tree we are copying into, and path is a scratch
 313  * path for use in this function (it should be released on entry and
 314  * will be released on exit).
 315  *
 316  * If the key is already in the destination tree the existing item is
 317  * overwritten.  If the existing item isn't big enough, it is extended.
 318  * If it is too large, it is truncated.
 319  *
 320  * If the key isn't in the destination yet, a new item is inserted.
 321  */
 322 static noinline int overwrite_item(struct btrfs_trans_handle *trans,
 323                                    struct btrfs_root *root,
 324                                    struct btrfs_path *path,
 325                                    struct extent_buffer *eb, int slot,
 326                                    struct btrfs_key *key)
 327 {
 328         int ret;
 329         u32 item_size;
 330         u64 saved_i_size = 0;
 331         int save_old_i_size = 0;
 332         unsigned long src_ptr;
 333         unsigned long dst_ptr;
 334         int overwrite_root = 0;
 335         bool inode_item = key->type == BTRFS_INODE_ITEM_KEY;
 336
 337         if (root->root_key.objectid != BTRFS_TREE_LOG_OBJECTID)
 338                 overwrite_root = 1;
 339
 340         item_size = btrfs_item_size_nr(eb, slot);
 341         src_ptr = btrfs_item_ptr_offset(eb, slot);
 342
 343         /* look for the key in the destination tree */
 344         ret = btrfs_search_slot(NULL, root, key, path, 0, 0);
 345         if (ret < 0)
 346                 return ret;
 347
 348         if (ret == 0) {
 349                 char *src_copy;
 350                 char *dst_copy;
 351                 u32 dst_size = btrfs_item_size_nr(path->nodes[0],
 352                                                   path->slots[0]);
 353                 if (dst_size != item_size)
 354                         goto insert;
 355
 356                 if (item_size == 0) {
 357                         btrfs_release_path(path);
 358                         return 0;
 359                 }
 360                 dst_copy = kmalloc(item_size, GFP_NOFS);
 361                 src_copy = kmalloc(item_size, GFP_NOFS);
 362                 if (!dst_copy || !src_copy) {
 363                         btrfs_release_path(path);
 364                         kfree(dst_copy);
 365                         kfree(src_copy);
 366                         return -ENOMEM;
 367                 }
 368
 369                 read_extent_buffer(eb, src_copy, src_ptr, item_size);
 370
 371                 dst_ptr = btrfs_item_ptr_offset(path->nodes[0], path->slots[0]);
 372                 read_extent_buffer(path->nodes[0], dst_copy, dst_ptr,
 373                                    item_size);
 374                 ret = memcmp(dst_copy, src_copy, item_size);
 375
 376                 kfree(dst_copy);
 377                 kfree(src_copy);
 378                 /*
 379                  * they have the same contents, just return, this saves
 380                  * us from cowing blocks in the destination tree and doing
 381                  * extra writes that may not have been done by a previous
 382                  * sync
 383                  */
 384                 if (ret == 0) {
 385                         btrfs_release_path(path);
 386                         return 0;
 387                 }
 388
 389                 /*
 390                  * We need to load the old nbytes into the inode so when we
 391                  * replay the extents we've logged we get the right nbytes.
 392                  */
 393                 if (inode_item) {
 394                         struct btrfs_inode_item *item;
 395                         u64 nbytes;
 396                         u32 mode;
 397
 398                         item = btrfs_item_ptr(path->nodes[0], path->slots[0],
 399                                               struct btrfs_inode_item);
 400                         nbytes = btrfs_inode_nbytes(path->nodes[0], item);
 401                         item = btrfs_item_ptr(eb, slot,
 402                                               struct btrfs_inode_item);
 403                         btrfs_set_inode_nbytes(eb, item, nbytes);
 404
 405                         /*
 406                          * If this is a directory we need to reset the i_size to
 407                          * 0 so that we can set it up properly when replaying
 408                          * the rest of the items in this log.
 409                          */
 410                         mode = btrfs_inode_mode(eb, item);
 411                         if (S_ISDIR(mode))
 412                                 btrfs_set_inode_size(eb, item, 0);
 413                 }
 414         } else if (inode_item) {
 415                 struct btrfs_inode_item *item;
 416                 u32 mode;
 417
 418                 /*
 419                  * New inode, set nbytes to 0 so that the nbytes comes out
 420                  * properly when we replay the extents.
 421                  */
 422                 item = btrfs_item_ptr(eb, slot, struct btrfs_inode_item);
 423                 btrfs_set_inode_nbytes(eb, item, 0);
 424
 425                 /*
 426                  * If this is a directory we need to reset the i_size to 0 so
 427                  * that we can set it up properly when replaying the rest of
 428                  * the items in this log.
 429                  */
 430                 mode = btrfs_inode_mode(eb, item);
 431                 if (S_ISDIR(mode))
 432                         btrfs_set_inode_size(eb, item, 0);
 433         }
 434 insert:
 435         btrfs_release_path(path);
 436         /* try to insert the key into the destination tree */
 437         ret = btrfs_insert_empty_item(trans, root, path,
 438                                       key, item_size);
 439
 440         /* make sure any existing item is the correct size */
 441         if (ret == -EEXIST) {
 442                 u32 found_size;
 443                 found_size = btrfs_item_size_nr(path->nodes[0],
 444                                                 path->slots[0]);
 445                 if (found_size > item_size)
 446                         btrfs_truncate_item(root, path, item_size, 1);
 447                 else if (found_size < item_size)
 448                         btrfs_extend_item(root, path,
 449                                           item_size - found_size);
 450         } else if (ret) {
 451                 return ret;
 452         }
 453         dst_ptr = btrfs_item_ptr_offset(path->nodes[0],
 454                                         path->slots[0]);
 455
 456         /* don't overwrite an existing inode if the generation number
 457          * was logged as zero.  This is done when the tree logging code
 458          * is just logging an inode to make sure it exists after recovery.
 459          *
 460          * Also, don't overwrite i_size on directories during replay.
 461          * log replay inserts and removes directory items based on the
 462          * state of the tree found in the subvolume, and i_size is modified
 463          * as it goes
 464          */
 465         if (key->type == BTRFS_INODE_ITEM_KEY && ret == -EEXIST) {
 466                 struct btrfs_inode_item *src_item;
 467                 struct btrfs_inode_item *dst_item;
 468
 469                 src_item = (struct btrfs_inode_item *)src_ptr;
 470                 dst_item = (struct btrfs_inode_item *)dst_ptr;
 471
 472                 if (btrfs_inode_generation(eb, src_item) == 0)
 473                         goto no_copy;
 474
 475                 if (overwrite_root &&
 476                     S_ISDIR(btrfs_inode_mode(eb, src_item)) &&
 477                     S_ISDIR(btrfs_inode_mode(path->nodes[0], dst_item))) {
 478                         save_old_i_size = 1;
 479                         saved_i_size = btrfs_inode_size(path->nodes[0],
 480                                                         dst_item);
 481                 }
 482         }
 483
 484         copy_extent_buffer(path->nodes[0], eb, dst_ptr,
 485                            src_ptr, item_size);
 486
 487         if (save_old_i_size) {
 488                 struct btrfs_inode_item *dst_item;
 489                 dst_item = (struct btrfs_inode_item *)dst_ptr;
 490                 btrfs_set_inode_size(path->nodes[0], dst_item, saved_i_size);
 491         }
 492
 493         /* make sure the generation is filled in */
 494         if (key->type == BTRFS_INODE_ITEM_KEY) {
 495                 struct btrfs_inode_item *dst_item;
 496                 dst_item = (struct btrfs_inode_item *)dst_ptr;
 497                 if (btrfs_inode_generation(path->nodes[0], dst_item) == 0) {
 498                         btrfs_set_inode_generation(path->nodes[0], dst_item,
 499                                                    trans->transid);
 500                 }
 501         }
 502 no_copy:
 503         btrfs_mark_buffer_dirty(path->nodes[0]);
 504         btrfs_release_path(path);
 505         return 0;
 506 }
 507
 508 /*
 509  * simple helper to read an inode off the disk from a given root
 510  * This can only be called for subvolume roots and not for the log
 511  */
 512 static noinline struct inode *read_one_inode(struct btrfs_root *root,
 513                                              u64 objectid)
 514 {
 515         struct btrfs_key key;
 516         struct inode *inode;
 517
 518         key.objectid = objectid;
 519         key.type = BTRFS_INODE_ITEM_KEY;
 520         key.offset = 0;
 521         inode = btrfs_iget(root->fs_info->sb, &key, root, NULL);
 522         if (IS_ERR(inode)) {
 523                 inode = NULL;
 524         } else if (is_bad_inode(inode)) {
 525                 iput(inode);
 526                 inode = NULL;
 527         }
 528         return inode;
 529 }
 530
 531 /* replays a single extent in 'eb' at 'slot' with 'key' into the
 532  * subvolume 'root'.  path is released on entry and should be released
 533  * on exit.
 534  *
 535  * extents in the log tree have not been allocated out of the extent
 536  * tree yet.  So, this completes the allocation, taking a reference
 537  * as required if the extent already exists or creating a new extent
 538  * if it isn't in the extent allocation tree yet.
 539  *
 540  * The extent is inserted into the file, dropping any existing extents
 541  * from the file that overlap the new one.
 542  */
 543 static noinline int replay_one_extent(struct btrfs_trans_handle *trans,
 544                                       struct btrfs_root *root,
 545                                       struct btrfs_path *path,
 546                                       struct extent_buffer *eb, int slot,
 547                                       struct btrfs_key *key)
 548 {
 549         int found_type;
 550         u64 extent_end;
 551         u64 start = key->offset;
 552         u64 nbytes = 0;
 553         struct btrfs_file_extent_item *item;
 554         struct inode *inode = NULL;
 555         unsigned long size;
 556         int ret = 0;
 557
 558         item = btrfs_item_ptr(eb, slot, struct btrfs_file_extent_item);
 559         found_type = btrfs_file_extent_type(eb, item);
 560
 561         if (found_type == BTRFS_FILE_EXTENT_REG ||
 562             found_type == BTRFS_FILE_EXTENT_PREALLOC) {
 563                 nbytes = btrfs_file_extent_num_bytes(eb, item);
 564                 extent_end = start + nbytes;
 565
 566                 /*
 567                  * We don't add to the inodes nbytes if we are prealloc or a
 568                  * hole.
 569                  */
 570                 if (btrfs_file_extent_disk_bytenr(eb, item) == 0)
 571                         nbytes = 0;
 572         } else if (found_type == BTRFS_FILE_EXTENT_INLINE) {
 573                 size = btrfs_file_extent_inline_len(eb, item);
 574                 nbytes = btrfs_file_extent_ram_bytes(eb, item);
 575                 extent_end = ALIGN(start + size, root->sectorsize);
 576         } else {
 577                 ret = 0;
 578                 goto out;
 579         }
 580
 581         inode = read_one_inode(root, key->objectid);
 582         if (!inode) {
 583                 ret = -EIO;
 584                 goto out;
 585         }
 586
 587         /*
 588          * first check to see if we already have this extent in the
 589          * file.  This must be done before the btrfs_drop_extents run
 590          * so we don't try to drop this extent.
 591          */
 592         ret = btrfs_lookup_file_extent(trans, root, path, btrfs_ino(inode),
 593                                        start, 0);
 594
 595         if (ret == 0 &&
 596             (found_type == BTRFS_FILE_EXTENT_REG ||
 597              found_type == BTRFS_FILE_EXTENT_PREALLOC)) {
 598                 struct btrfs_file_extent_item cmp1;
 599                 struct btrfs_file_extent_item cmp2;
 600                 struct btrfs_file_extent_item *existing;
 601                 struct extent_buffer *leaf;
 602
 603                 leaf = path->nodes[0];
 604                 existing = btrfs_item_ptr(leaf, path->slots[0],
 605                                           struct btrfs_file_extent_item);
 606
 607                 read_extent_buffer(eb, &cmp1, (unsigned long)item,
 608                                    sizeof(cmp1));
 609                 read_extent_buffer(leaf, &cmp2, (unsigned long)existing,
 610                                    sizeof(cmp2));
 611
 612                 /*
 613                  * we already have a pointer to this exact extent,
 614                  * we don't have to do anything
 615                  */
 616                 if (memcmp(&cmp1, &cmp2, sizeof(cmp1)) == 0) {
 617                         btrfs_release_path(path);
 618                         goto out;
 619                 }
 620         }
 621         btrfs_release_path(path);
 622
 623         /* drop any overlapping extents */
 624         ret = btrfs_drop_extents(trans, root, inode, start, extent_end, 1);
 625         if (ret)
 626                 goto out;
 627
 628         if (found_type == BTRFS_FILE_EXTENT_REG ||
 629             found_type == BTRFS_FILE_EXTENT_PREALLOC) {
 630                 u64 offset;
 631                 unsigned long dest_offset;
 632                 struct btrfs_key ins;
 633
 634                 ret = btrfs_insert_empty_item(trans, root, path, key,
 635                                               sizeof(*item));
 636                 if (ret)
 637                         goto out;
 638                 dest_offset = btrfs_item_ptr_offset(path->nodes[0],
 639                                                     path->slots[0]);
 640                 copy_extent_buffer(path->nodes[0], eb, dest_offset,
 641                                 (unsigned long)item,  sizeof(*item));
 642
 643                 ins.objectid = btrfs_file_extent_disk_bytenr(eb, item);
 644                 ins.offset = btrfs_file_extent_disk_num_bytes(eb, item);
 645                 ins.type = BTRFS_EXTENT_ITEM_KEY;
 646                 offset = key->offset - btrfs_file_extent_offset(eb, item);
 647
 648                 if (ins.objectid > 0) {
 649                         u64 csum_start;
 650                         u64 csum_end;
 651                         LIST_HEAD(ordered_sums);
 652                         /*
 653                          * is this extent already allocated in the extent
 654                          * allocation tree?  If so, just add a reference
 655                          */
 656                         ret = btrfs_lookup_extent(root, ins.objectid,
 657                                                 ins.offset);
 658                         if (ret == 0) {
 659                                 ret = btrfs_inc_extent_ref(trans, root,
 660                                                 ins.objectid, ins.offset,
 661                                                 0, root->root_key.objectid,
 662                                                 key->objectid, offset, 0);
 663                                 if (ret)
 664                                         goto out;
 665                         } else {
 666                                 /*
 667                                  * insert the extent pointer in the extent
 668                                  * allocation tree
 669                                  */
 670                                 ret = btrfs_alloc_logged_file_extent(trans,
 671                                                 root, root->root_key.objectid,
 672                                                 key->objectid, offset, &ins);
 673                                 if (ret)
 674                                         goto out;
 675                         }
 676                         btrfs_release_path(path);
 677
 678                         if (btrfs_file_extent_compression(eb, item)) {
 679                                 csum_start = ins.objectid;
 680                                 csum_end = csum_start + ins.offset;
 681                         } else {
 682                                 csum_start = ins.objectid +
 683                                         btrfs_file_extent_offset(eb, item);
 684                                 csum_end = csum_start +
 685                                         btrfs_file_extent_num_bytes(eb, item);
 686                         }
 687
 688                         ret = btrfs_lookup_csums_range(root->log_root,
 689                                                 csum_start, csum_end - 1,
 690                                                 &ordered_sums, 0);
 691                         if (ret)
 692                                 goto out;
 693                         while (!list_empty(&ordered_sums)) {
 694                                 struct btrfs_ordered_sum *sums;
 695                                 sums = list_entry(ordered_sums.next,
 696                                                 struct btrfs_ordered_sum,
 697                                                 list);
 698                                 if (!ret)
 699                                         ret = btrfs_csum_file_blocks(trans,
 700                                                 root->fs_info->csum_root,
 701                                                 sums);
 702                                 list_del(&sums->list);
 703                                 kfree(sums);
 704                         }
 705                         if (ret)
 706                                 goto out;
 707                 } else {
 708                         btrfs_release_path(path);
 709                 }
 710         } else if (found_type == BTRFS_FILE_EXTENT_INLINE) {
 711                 /* inline extents are easy, we just overwrite them */
 712                 ret = overwrite_item(trans, root, path, eb, slot, key);
 713                 if (ret)
 714                         goto out;
 715         }
 716
 717         inode_add_bytes(inode, nbytes);
 718         ret = btrfs_update_inode(trans, root, inode);
 719 out:
 720         if (inode)
 721                 iput(inode);
 722         return ret;
 723 }
 724
 725 /*
 726  * when cleaning up conflicts between the directory names in the
 727  * subvolume, directory names in the log and directory names in the
 728  * inode back references, we may have to unlink inodes from directories.
 729  *
 730  * This is a helper function to do the unlink of a specific directory
 731  * item
 732  */
 733 static noinline int drop_one_dir_item(struct btrfs_trans_handle *trans,
 734                                       struct btrfs_root *root,
 735                                       struct btrfs_path *path,
 736                                       struct inode *dir,
 737                                       struct btrfs_dir_item *di)
 738 {
 739         struct inode *inode;
 740         char *name;
 741         int name_len;
 742         struct extent_buffer *leaf;
 743         struct btrfs_key location;
 744         int ret;
 745
 746         leaf = path->nodes[0];
 747
 748         btrfs_dir_item_key_to_cpu(leaf, di, &location);
 749         name_len = btrfs_dir_name_len(leaf, di);
 750         name = kmalloc(name_len, GFP_NOFS);
 751         if (!name)
 752                 return -ENOMEM;
 753
 754         read_extent_buffer(leaf, name, (unsigned long)(di + 1), name_len);
 755         btrfs_release_path(path);
 756
 757         inode = read_one_inode(root, location.objectid);
 758         if (!inode) {
 759                 ret = -EIO;
 760                 goto out;
 761         }
 762
 763         ret = link_to_fixup_dir(trans, root, path, location.objectid);
 764         if (ret)
 765                 goto out;
 766
 767         ret = btrfs_unlink_inode(trans, root, dir, inode, name, name_len);
 768         if (ret)
 769                 goto out;
 770         else
 771                 ret = btrfs_run_delayed_items(trans, root);
 772 out:
 773         kfree(name);
 774         iput(inode);
 775         return ret;
 776 }
 777
 778 /*
 779  * helper function to see if a given name and sequence number found
 780  * in an inode back reference are already in a directory and correctly
 781  * point to this inode
 782  */
 783 static noinline int inode_in_dir(struct btrfs_root *root,
 784                                  struct btrfs_path *path,
 785                                  u64 dirid, u64 objectid, u64 index,
 786                                  const char *name, int name_len)
 787 {
 788         struct btrfs_dir_item *di;
 789         struct btrfs_key location;
 790         int match = 0;
 791
 792         di = btrfs_lookup_dir_index_item(NULL, root, path, dirid,
 793                                          index, name, name_len, 0);
 794         if (di && !IS_ERR(di)) {
 795                 btrfs_dir_item_key_to_cpu(path->nodes[0], di, &location);
 796                 if (location.objectid != objectid)
 797                         goto out;
 798         } else
 799                 goto out;
 800         btrfs_release_path(path);
 801
 802         di = btrfs_lookup_dir_item(NULL, root, path, dirid, name, name_len, 0);
 803         if (di && !IS_ERR(di)) {
 804                 btrfs_dir_item_key_to_cpu(path->nodes[0], di, &location);
 805                 if (location.objectid != objectid)
 806                         goto out;
 807         } else
 808                 goto out;
 809         match = 1;
 810 out:
 811         btrfs_release_path(path);
 812         return match;
 813 }
 814
 815 /*
 816  * helper function to check a log tree for a named back reference in
 817  * an inode.  This is used to decide if a back reference that is
 818  * found in the subvolume conflicts with what we find in the log.
 819  *
 820  * inode backreferences may have multiple refs in a single item,
 821  * during replay we process one reference at a time, and we don't
 822  * want to delete valid links to a file from the subvolume if that
 823  * link is also in the log.
 824  */
 825 static noinline int backref_in_log(struct btrfs_root *log,
 826                                    struct btrfs_key *key,
 827                                    u64 ref_objectid,
 828                                    char *name, int namelen)
 829 {
 830         struct btrfs_path *path;
 831         struct btrfs_inode_ref *ref;
 832         unsigned long ptr;
 833         unsigned long ptr_end;
 834         unsigned long name_ptr;
 835         int found_name_len;
 836         int item_size;
 837         int ret;
 838         int match = 0;
 839
 840         path = btrfs_alloc_path();
 841         if (!path)
 842                 return -ENOMEM;
 843
 844         ret = btrfs_search_slot(NULL, log, key, path, 0, 0);
 845         if (ret != 0)
 846                 goto out;
 847
 848         ptr = btrfs_item_ptr_offset(path->nodes[0], path->slots[0]);
 849
 850         if (key->type == BTRFS_INODE_EXTREF_KEY) {
 851                 if (btrfs_find_name_in_ext_backref(path, ref_objectid,
 852                                                    name, namelen, NULL))
 853                         match = 1;
 854
 855                 goto out;
 856         }
 857
 858         item_size = btrfs_item_size_nr(path->nodes[0], path->slots[0]);
 859         ptr_end = ptr + item_size;
 860         while (ptr < ptr_end) {
 861                 ref = (struct btrfs_inode_ref *)ptr;
 862                 found_name_len = btrfs_inode_ref_name_len(path->nodes[0], ref);
 863                 if (found_name_len == namelen) {
 864                         name_ptr = (unsigned long)(ref + 1);
 865                         ret = memcmp_extent_buffer(path->nodes[0], name,
 866                                                    name_ptr, namelen);
 867                         if (ret == 0) {
 868                                 match = 1;
 869                                 goto out;
 870                         }
 871                 }
 872                 ptr = (unsigned long)(ref + 1) + found_name_len;
 873         }
 874 out:
 875         btrfs_free_path(path);
 876         return match;
 877 }
 878
 879 static inline int __add_inode_ref(struct btrfs_trans_handle *trans,
 880                                   struct btrfs_root *root,
 881                                   struct btrfs_path *path,
 882                                   struct btrfs_root *log_root,
 883                                   struct inode *dir, struct inode *inode,
 884                                   struct extent_buffer *eb,
 885                                   u64 inode_objectid, u64 parent_objectid,
 886                                   u64 ref_index, char *name, int namelen,
 887                                   int *search_done)
 888 {
 889         int ret;
 890         char *victim_name;
 891         int victim_name_len;
 892         struct extent_buffer *leaf;
 893         struct btrfs_dir_item *di;
 894         struct btrfs_key search_key;
 895         struct btrfs_inode_extref *extref;
 896
 897 again:
 898         /* Search old style refs */
 899         search_key.objectid = inode_objectid;
 900         search_key.type = BTRFS_INODE_REF_KEY;
 901         search_key.offset = parent_objectid;
 902         ret = btrfs_search_slot(NULL, root, &search_key, path, 0, 0);
 903         if (ret == 0) {
 904                 struct btrfs_inode_ref *victim_ref;
 905                 unsigned long ptr;
 906                 unsigned long ptr_end;
 907
 908                 leaf = path->nodes[0];
 909
 910                 /* are we trying to overwrite a back ref for the root directory
 911                  * if so, just jump out, we're done
 912                  */
 913                 if (search_key.objectid == search_key.offset)
 914                         return 1;
 915
 916                 /* check all the names in this back reference to see
 917                  * if they are in the log.  if so, we allow them to stay
 918                  * otherwise they must be unlinked as a conflict
 919                  */
 920                 ptr = btrfs_item_ptr_offset(leaf, path->slots[0]);
 921                 ptr_end = ptr + btrfs_item_size_nr(leaf, path->slots[0]);
 922                 while (ptr < ptr_end) {
 923                         victim_ref = (struct btrfs_inode_ref *)ptr;
 924                         victim_name_len = btrfs_inode_ref_name_len(leaf,
 925                                                                    victim_ref);
 926                         victim_name = kmalloc(victim_name_len, GFP_NOFS);
 927                         if (!victim_name)
 928                                 return -ENOMEM;
 929
 930                         read_extent_buffer(leaf, victim_name,
 931                                            (unsigned long)(victim_ref + 1),
 932                                            victim_name_len);
 933
 934                         if (!backref_in_log(log_root, &search_key,
 935                                             parent_objectid,
 936                                             victim_name,
 937                                             victim_name_len)) {
 938                                 inc_nlink(inode);
 939                                 btrfs_release_path(path);
 940
 941                                 ret = btrfs_unlink_inode(trans, root, dir,
 942                                                          inode, victim_name,
 943                                                          victim_name_len);
 944                                 kfree(victim_name);
 945                                 if (ret)
 946                                         return ret;
 947                                 ret = btrfs_run_delayed_items(trans, root);
 948                                 if (ret)
 949                                         return ret;
 950                                 *search_done = 1;
 951                                 goto again;
 952                         }
 953                         kfree(victim_name);
 954
 955                         ptr = (unsigned long)(victim_ref + 1) + victim_name_len;
 956                 }
 957
 958                 /*
 959                  * NOTE: we have searched root tree and checked the
 960                  * coresponding ref, it does not need to check again.
 961                  */
 962                 *search_done = 1;
 963         }
 964         btrfs_release_path(path);
 965
 966         /* Same search but for extended refs */
 967         extref = btrfs_lookup_inode_extref(NULL, root, path, name, namelen,
 968                                            inode_objectid, parent_objectid, 0,
 969                                            0);
 970         if (!IS_ERR_OR_NULL(extref)) {
 971                 u32 item_size;
 972                 u32 cur_offset = 0;
 973                 unsigned long base;
 974                 struct inode *victim_parent;
 975
 976                 leaf = path->nodes[0];
 977
 978                 item_size = btrfs_item_size_nr(leaf, path->slots[0]);
 979                 base = btrfs_item_ptr_offset(leaf, path->slots[0]);
 980
 981                 while (cur_offset < item_size) {
 982                         extref = (struct btrfs_inode_extref *)base + cur_offset;
 983
 984                         victim_name_len = btrfs_inode_extref_name_len(leaf, extref);
 985
 986                         if (btrfs_inode_extref_parent(leaf, extref) != parent_objectid)
 987                                 goto next;
 988
 989                         victim_name = kmalloc(victim_name_len, GFP_NOFS);
 990                         if (!victim_name)
 991                                 return -ENOMEM;
 992                         read_extent_buffer(leaf, victim_name, (unsigned long)&extref->name,
 993                                            victim_name_len);
 994
 995                         search_key.objectid = inode_objectid;
 996                         search_key.type = BTRFS_INODE_EXTREF_KEY;
 997                         search_key.offset = btrfs_extref_hash(parent_objectid,
 998                                                               victim_name,
 999                                                               victim_name_len);
1000                         ret = 0;
1001                         if (!backref_in_log(log_root, &search_key,
1002                                             parent_objectid, victim_name,
1003                                             victim_name_len)) {
1004                                 ret = -ENOENT;
1005                                 victim_parent = read_one_inode(root,
1006                                                                parent_objectid);
1007                                 if (victim_parent) {
1008                                         inc_nlink(inode);
1009                                         btrfs_release_path(path);
1010
1011                                         ret = btrfs_unlink_inode(trans, root,
1012                                                                  victim_parent,
1013                                                                  inode,
1014                                                                  victim_name,
1015                                                                  victim_name_len);
1016                                         if (!ret)
1017                                                 ret = btrfs_run_delayed_items(
1018                                                                   trans, root);
1019                                 }
1020                                 iput(victim_parent);
1021                                 kfree(victim_name);
1022                                 if (ret)
1023                                         return ret;
1024                                 *search_done = 1;
1025                                 goto again;
1026                         }
1027                         kfree(victim_name);
1028                         if (ret)
1029                                 return ret;
1030 next:
1031                         cur_offset += victim_name_len + sizeof(*extref);
1032                 }
1033                 *search_done = 1;
1034         }
1035         btrfs_release_path(path);
1036
1037         /* look for a conflicting sequence number */
1038         di = btrfs_lookup_dir_index_item(trans, root, path, btrfs_ino(dir),
1039                                          ref_index, name, namelen, 0);
1040         if (di && !IS_ERR(di)) {
1041                 ret = drop_one_dir_item(trans, root, path, dir, di);
1042                 if (ret)
1043                         return ret;
1044         }
1045         btrfs_release_path(path);
1046
1047         /* look for a conflicing name */
1048         di = btrfs_lookup_dir_item(trans, root, path, btrfs_ino(dir),
1049                                    name, namelen, 0);
1050         if (di && !IS_ERR(di)) {
1051                 ret = drop_one_dir_item(trans, root, path, dir, di);
1052                 if (ret)
1053                         return ret;
1054         }
1055         btrfs_release_path(path);
1056
1057         return 0;
1058 }
1059
1060 static int extref_get_fields(struct extent_buffer *eb, unsigned long ref_ptr,
1061                              u32 *namelen, char **name, u64 *index,
1062                              u64 *parent_objectid)
1063 {
1064         struct btrfs_inode_extref *extref;
1065
1066         extref = (struct btrfs_inode_extref *)ref_ptr;
1067
1068         *namelen = btrfs_inode_extref_name_len(eb, extref);
1069         *name = kmalloc(*namelen, GFP_NOFS);
1070         if (*name == NULL)
1071                 return -ENOMEM;
1072
1073         read_extent_buffer(eb, *name, (unsigned long)&extref->name,
1074                            *namelen);
1075
1076         *index = btrfs_inode_extref_index(eb, extref);
1077         if (parent_objectid)
1078                 *parent_objectid = btrfs_inode_extref_parent(eb, extref);
1079
1080         return 0;
1081 }
1082
1083 static int ref_get_fields(struct extent_buffer *eb, unsigned long ref_ptr,
1084                           u32 *namelen, char **name, u64 *index)
1085 {
1086         struct btrfs_inode_ref *ref;
1087
1088         ref = (struct btrfs_inode_ref *)ref_ptr;
1089
1090         *namelen = btrfs_inode_ref_name_len(eb, ref);
1091         *name = kmalloc(*namelen, GFP_NOFS);
1092         if (*name == NULL)
1093                 return -ENOMEM;
1094
1095         read_extent_buffer(eb, *name, (unsigned long)(ref + 1), *namelen);
1096
1097         *index = btrfs_inode_ref_index(eb, ref);
1098
1099         return 0;
1100 }
1101
1102 /*
1103  * replay one inode back reference item found in the log tree.
1104  * eb, slot and key refer to the buffer and key found in the log tree.
1105  * root is the destination we are replaying into, and path is for temp
1106  * use by this function.  (it should be released on return).
1107  */
1108 static noinline int add_inode_ref(struct btrfs_trans_handle *trans,
1109                                   struct btrfs_root *root,
1110                                   struct btrfs_root *log,
1111                                   struct btrfs_path *path,
1112                                   struct extent_buffer *eb, int slot,
1113                                   struct btrfs_key *key)
1114 {
1115         struct inode *dir = NULL;
1116         struct inode *inode = NULL;
1117         unsigned long ref_ptr;
1118         unsigned long ref_end;
1119         char *name = NULL;
1120         int namelen;
1121         int ret;
1122         int search_done = 0;
1123         int log_ref_ver = 0;
1124         u64 parent_objectid;
1125         u64 inode_objectid;
1126         u64 ref_index = 0;
1127         int ref_struct_size;
1128
1129         ref_ptr = btrfs_item_ptr_offset(eb, slot);
1130         ref_end = ref_ptr + btrfs_item_size_nr(eb, slot);
1131
1132         if (key->type == BTRFS_INODE_EXTREF_KEY) {
1133                 struct btrfs_inode_extref *r;
1134
1135                 ref_struct_size = sizeof(struct btrfs_inode_extref);
1136                 log_ref_ver = 1;
1137                 r = (struct btrfs_inode_extref *)ref_ptr;
1138                 parent_objectid = btrfs_inode_extref_parent(eb, r);
1139         } else {
1140                 ref_struct_size = sizeof(struct btrfs_inode_ref);
1141                 parent_objectid = key->offset;
1142         }
1143         inode_objectid = key->objectid;
1144
1145         /*
1146          * it is possible that we didn't log all the parent directories
1147          * for a given inode.  If we don't find the dir, just don't
1148          * copy the back ref in.  The link count fixup code will take
1149          * care of the rest
1150          */
1151         dir = read_one_inode(root, parent_objectid);
1152         if (!dir) {
1153                 ret = -ENOENT;
1154                 goto out;
1155         }
1156
1157         inode = read_one_inode(root, inode_objectid);
1158         if (!inode) {
1159                 ret = -EIO;
1160                 goto out;
1161         }
1162
1163         while (ref_ptr < ref_end) {
1164                 if (log_ref_ver) {
1165                         ret = extref_get_fields(eb, ref_ptr, &namelen, &name,
1166                                                 &ref_index, &parent_objectid);
1167                         /*
1168                          * parent object can change from one array
1169                          * item to another.
1170                          */
1171                         if (!dir)
1172                                 dir = read_one_inode(root, parent_objectid);
1173                         if (!dir) {
1174                                 ret = -ENOENT;
1175                                 goto out;
1176                         }
1177                 } else {
1178                         ret = ref_get_fields(eb, ref_ptr, &namelen, &name,
1179                                              &ref_index);
1180                 }
1181                 if (ret)
1182                         goto out;
1183
1184                 /* if we already have a perfect match, we're done */
1185                 if (!inode_in_dir(root, path, btrfs_ino(dir), btrfs_ino(inode),
1186                                   ref_index, name, namelen)) {
1187                         /*
1188                          * look for a conflicting back reference in the
1189                          * metadata. if we find one we have to unlink that name
1190                          * of the file before we add our new link.  Later on, we
1191                          * overwrite any existing back reference, and we don't
1192                          * want to create dangling pointers in the directory.
1193                          */
1194
1195                         if (!search_done) {
1196                                 ret = __add_inode_ref(trans, root, path, log,
1197                                                       dir, inode, eb,
1198                                                       inode_objectid,
1199                                                       parent_objectid,
1200                                                       ref_index, name, namelen,
1201                                                       &search_done);
1202                                 if (ret) {
1203                                         if (ret == 1)
1204                                                 ret = 0;
1205                                         goto out;
1206                                 }
1207                         }
1208
1209                         /* insert our name */
1210                         ret = btrfs_add_link(trans, dir, inode, name, namelen,
1211                                              0, ref_index);
1212                         if (ret)
1213                                 goto out;
1214
1215                         btrfs_update_inode(trans, root, inode);
1216                 }
1217
1218                 ref_ptr = (unsigned long)(ref_ptr + ref_struct_size) + namelen;
1219                 kfree(name);
1220                 name = NULL;
1221                 if (log_ref_ver) {
1222                         iput(dir);
1223                         dir = NULL;
1224                 }
1225         }
1226
1227         /* finally write the back reference in the inode */
1228         ret = overwrite_item(trans, root, path, eb, slot, key);
1229 out:
1230         btrfs_release_path(path);
1231         kfree(name);
1232         iput(dir);
1233         iput(inode);
1234         return ret;
1235 }
1236
1237 static int insert_orphan_item(struct btrfs_trans_handle *trans,
1238                               struct btrfs_root *root, u64 offset)
1239 {
1240         int ret;
1241         ret = btrfs_find_orphan_item(root, offset);
1242         if (ret > 0)
1243                 ret = btrfs_insert_orphan_item(trans, root, offset);
1244         return ret;
1245 }
1246
1247 static int count_inode_extrefs(struct btrfs_root *root,
1248                                struct inode *inode, struct btrfs_path *path)
1249 {
1250         int ret = 0;
1251         int name_len;
1252         unsigned int nlink = 0;
1253         u32 item_size;
1254         u32 cur_offset = 0;
1255         u64 inode_objectid = btrfs_ino(inode);
1256         u64 offset = 0;
1257         unsigned long ptr;
1258         struct btrfs_inode_extref *extref;
1259         struct extent_buffer *leaf;
1260
1261         while (1) {
1262                 ret = btrfs_find_one_extref(root, inode_objectid, offset, path,
1263                                             &extref, &offset);
1264                 if (ret)
1265                         break;
1266
1267                 leaf = path->nodes[0];
1268                 item_size = btrfs_item_size_nr(leaf, path->slots[0]);
1269                 ptr = btrfs_item_ptr_offset(leaf, path->slots[0]);
1270
1271                 while (cur_offset < item_size) {
1272                         extref = (struct btrfs_inode_extref *) (ptr + cur_offset);
1273                         name_len = btrfs_inode_extref_name_len(leaf, extref);
1274
1275                         nlink++;
1276
1277                         cur_offset += name_len + sizeof(*extref);
1278                 }
1279
1280                 offset++;
1281                 btrfs_release_path(path);
1282         }
1283         btrfs_release_path(path);
1284
1285         if (ret < 0)
1286                 return ret;
1287         return nlink;
1288 }
1289
1290 static int count_inode_refs(struct btrfs_root *root,
1291                                struct inode *inode, struct btrfs_path *path)
1292 {
1293         int ret;
1294         struct btrfs_key key;
1295         unsigned int nlink = 0;
1296         unsigned long ptr;
1297         unsigned long ptr_end;
1298         int name_len;
1299         u64 ino = btrfs_ino(inode);
1300
1301         key.objectid = ino;
1302         key.type = BTRFS_INODE_REF_KEY;
1303         key.offset = (u64)-1;
1304
1305         while (1) {
1306                 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
1307                 if (ret < 0)
1308                         break;
1309                 if (ret > 0) {
1310                         if (path->slots[0] == 0)
1311                                 break;
1312                         path->slots[0]--;
1313                 }
1314 process_slot:
1315                 btrfs_item_key_to_cpu(path->nodes[0], &key,
1316                                       path->slots[0]);
1317                 if (key.objectid != ino ||
1318                     key.type != BTRFS_INODE_REF_KEY)
1319                         break;
1320                 ptr = btrfs_item_ptr_offset(path->nodes[0], path->slots[0]);
1321                 ptr_end = ptr + btrfs_item_size_nr(path->nodes[0],
1322                                                    path->slots[0]);
1323                 while (ptr < ptr_end) {
1324                         struct btrfs_inode_ref *ref;
1325
1326                         ref = (struct btrfs_inode_ref *)ptr;
1327                         name_len = btrfs_inode_ref_name_len(path->nodes[0],
1328                                                             ref);
1329                         ptr = (unsigned long)(ref + 1) + name_len;
1330                         nlink++;
1331                 }
1332
1333                 if (key.offset == 0)
1334                         break;
1335                 if (path->slots[0] > 0) {
1336                         path->slots[0]--;
1337                         goto process_slot;
1338                 }
1339                 key.offset--;
1340                 btrfs_release_path(path);
1341         }
1342         btrfs_release_path(path);
1343
1344         return nlink;
1345 }
1346
1347 /*
1348  * There are a few corners where the link count of the file can't
1349  * be properly maintained during replay.  So, instead of adding
1350  * lots of complexity to the log code, we just scan the backrefs
1351  * for any file that has been through replay.
1352  *
1353  * The scan will update the link count on the inode to reflect the
1354  * number of back refs found.  If it goes down to zero, the iput
1355  * will free the inode.
1356  */
1357 static noinline int fixup_inode_link_count(struct btrfs_trans_handle *trans,
1358                                            struct btrfs_root *root,
1359                                            struct inode *inode)
1360 {
1361         struct btrfs_path *path;
1362         int ret;
1363         u64 nlink = 0;
1364         u64 ino = btrfs_ino(inode);
1365
1366         path = btrfs_alloc_path();
1367         if (!path)
1368                 return -ENOMEM;
1369
1370         ret = count_inode_refs(root, inode, path);
1371         if (ret < 0)
1372                 goto out;
1373
1374         nlink = ret;
1375
1376         ret = count_inode_extrefs(root, inode, path);
1377         if (ret == -ENOENT)
1378                 ret = 0;
1379
1380         if (ret < 0)
1381                 goto out;
1382
1383         nlink += ret;
1384
1385         ret = 0;
1386
1387         if (nlink != inode->i_nlink) {
1388                 set_nlink(inode, nlink);
1389                 btrfs_update_inode(trans, root, inode);
1390         }
1391         BTRFS_I(inode)->index_cnt = (u64)-1;
1392
1393         if (inode->i_nlink == 0) {
1394                 if (S_ISDIR(inode->i_mode)) {
1395                         ret = replay_dir_deletes(trans, root, NULL, path,
1396                                                  ino, 1);
1397                         if (ret)
1398                                 goto out;
1399                 }
1400                 ret = insert_orphan_item(trans, root, ino);
1401         }
1402
1403 out:
1404         btrfs_free_path(path);
1405         return ret;
1406 }
1407
1408 static noinline int fixup_inode_link_counts(struct btrfs_trans_handle *trans,
1409                                             struct btrfs_root *root,
1410                                             struct btrfs_path *path)
1411 {
1412         int ret;
1413         struct btrfs_key key;
1414         struct inode *inode;
1415
1416         key.objectid = BTRFS_TREE_LOG_FIXUP_OBJECTID;
1417         key.type = BTRFS_ORPHAN_ITEM_KEY;
1418         key.offset = (u64)-1;
1419         while (1) {
1420                 ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
1421                 if (ret < 0)
1422                         break;
1423
1424                 if (ret == 1) {
1425                         if (path->slots[0] == 0)
1426                                 break;
1427                         path->slots[0]--;
1428                 }
1429
1430                 btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
1431                 if (key.objectid != BTRFS_TREE_LOG_FIXUP_OBJECTID ||
1432                     key.type != BTRFS_ORPHAN_ITEM_KEY)
1433                         break;
1434
1435                 ret = btrfs_del_item(trans, root, path);
1436                 if (ret)
1437                         goto out;
1438
1439                 btrfs_release_path(path);
1440                 inode = read_one_inode(root, key.offset);
1441                 if (!inode)
1442                         return -EIO;
1443
1444                 ret = fixup_inode_link_count(trans, root, inode);
1445                 iput(inode);
1446                 if (ret)
1447                         goto out;
1448
1449                 /*
1450                  * fixup on a directory may create new entries,
1451                  * make sure we always look for the highset possible
1452                  * offset
1453                  */
1454                 key.offset = (u64)-1;
1455         }
1456         ret = 0;
1457 out:
1458         btrfs_release_path(path);
1459         return ret;
1460 }
1461
1462
1463 /*
1464  * record a given inode in the fixup dir so we can check its link
1465  * count when replay is done.  The link count is incremented here
1466  * so the inode won't go away until we check it
1467  */
1468 static noinline int link_to_fixup_dir(struct btrfs_trans_handle *trans,
1469                                       struct btrfs_root *root,
1470                                       struct btrfs_path *path,
1471                                       u64 objectid)
1472 {
1473         struct btrfs_key key;
1474         int ret = 0;
1475         struct inode *inode;
1476
1477         inode = read_one_inode(root, objectid);
1478         if (!inode)
1479                 return -EIO;
1480
1481         key.objectid = BTRFS_TREE_LOG_FIXUP_OBJECTID;
1482         btrfs_set_key_type(&key, BTRFS_ORPHAN_ITEM_KEY);
1483         key.offset = objectid;
1484
1485         ret = btrfs_insert_empty_item(trans, root, path, &key, 0);
1486
1487         btrfs_release_path(path);
1488         if (ret == 0) {
1489                 if (!inode->i_nlink)
1490                         set_nlink(inode, 1);
1491                 else
1492                         inc_nlink(inode);
1493                 ret = btrfs_update_inode(trans, root, inode);
1494         } else if (ret == -EEXIST) {
1495                 ret = 0;
1496         } else {
1497                 BUG(); /* Logic Error */
1498         }
1499         iput(inode);
1500
1501         return ret;
1502 }
1503
1504 /*
1505  * when replaying the log for a directory, we only insert names
1506  * for inodes that actually exist.  This means an fsync on a directory
1507  * does not implicitly fsync all the new files in it
1508  */
1509 static noinline int insert_one_name(struct btrfs_trans_handle *trans,
1510                                     struct btrfs_root *root,
1511                                     struct btrfs_path *path,
1512                                     u64 dirid, u64 index,
1513                                     char *name, int name_len, u8 type,
1514                                     struct btrfs_key *location)
1515 {
1516         struct inode *inode;
1517         struct inode *dir;
1518         int ret;
1519
1520         inode = read_one_inode(root, location->objectid);
1521         if (!inode)
1522                 return -ENOENT;
1523
1524         dir = read_one_inode(root, dirid);
1525         if (!dir) {
1526                 iput(inode);
1527                 return -EIO;
1528         }
1529
1530         ret = btrfs_add_link(trans, dir, inode, name, name_len, 1, index);
1531
1532         /* FIXME, put inode into FIXUP list */
1533
1534         iput(inode);
1535         iput(dir);
1536         return ret;
1537 }
1538
1539 /*
1540  * take a single entry in a log directory item and replay it into
1541  * the subvolume.
1542  *
1543  * if a conflicting item exists in the subdirectory already,
1544  * the inode it points to is unlinked and put into the link count
1545  * fix up tree.
1546  *
1547  * If a name from the log points to a file or directory that does
1548  * not exist in the FS, it is skipped.  fsyncs on directories
1549  * do not force down inodes inside that directory, just changes to the
1550  * names or unlinks in a directory.
1551  */
1552 static noinline int replay_one_name(struct btrfs_trans_handle *trans,
1553                                     struct btrfs_root *root,
1554                                     struct btrfs_path *path,
1555                                     struct extent_buffer *eb,
1556                                     struct btrfs_dir_item *di,
1557                                     struct btrfs_key *key)
1558 {
1559         char *name;
1560         int name_len;
1561         struct btrfs_dir_item *dst_di;
1562         struct btrfs_key found_key;
1563         struct btrfs_key log_key;
1564         struct inode *dir;
1565         u8 log_type;
1566         int exists;
1567         int ret = 0;
1568         bool update_size = (key->type == BTRFS_DIR_INDEX_KEY);
1569
1570         dir = read_one_inode(root, key->objectid);
1571         if (!dir)
1572                 return -EIO;
1573
1574         name_len = btrfs_dir_name_len(eb, di);
1575         name = kmalloc(name_len, GFP_NOFS);
1576         if (!name) {
1577                 ret = -ENOMEM;
1578                 goto out;
1579         }
1580
1581         log_type = btrfs_dir_type(eb, di);
1582         read_extent_buffer(eb, name, (unsigned long)(di + 1),
1583                    name_len);
1584
1585         btrfs_dir_item_key_to_cpu(eb, di, &log_key);
1586         exists = btrfs_lookup_inode(trans, root, path, &log_key, 0);
1587         if (exists == 0)
1588                 exists = 1;
1589         else
1590                 exists = 0;
1591         btrfs_release_path(path);
1592
1593         if (key->type == BTRFS_DIR_ITEM_KEY) {
1594                 dst_di = btrfs_lookup_dir_item(trans, root, path, key->objectid,
1595                                        name, name_len, 1);
1596         } else if (key->type == BTRFS_DIR_INDEX_KEY) {
1597                 dst_di = btrfs_lookup_dir_index_item(trans, root, path,
1598                                                      key->objectid,
1599                                                      key->offset, name,
1600                                                      name_len, 1);
1601         } else {
1602                 /* Corruption */
1603                 ret = -EINVAL;
1604                 goto out;
1605         }
1606         if (IS_ERR_OR_NULL(dst_di)) {
1607                 /* we need a sequence number to insert, so we only
1608                  * do inserts for the BTRFS_DIR_INDEX_KEY types
1609                  */
1610                 if (key->type != BTRFS_DIR_INDEX_KEY)
1611                         goto out;
1612                 goto insert;
1613         }
1614
1615         btrfs_dir_item_key_to_cpu(path->nodes[0], dst_di, &found_key);
1616         /* the existing item matches the logged item */
1617         if (found_key.objectid == log_key.objectid &&
1618             found_key.type == log_key.type &&
1619             found_key.offset == log_key.offset &&
1620             btrfs_dir_type(path->nodes[0], dst_di) == log_type) {
1621                 goto out;
1622         }
1623
1624         /*
1625          * don't drop the conflicting directory entry if the inode
1626          * for the new entry doesn't exist
1627          */
1628         if (!exists)
1629                 goto out;
1630
1631         ret = drop_one_dir_item(trans, root, path, dir, dst_di);
1632         if (ret)
1633                 goto out;
1634
1635         if (key->type == BTRFS_DIR_INDEX_KEY)
1636                 goto insert;
1637 out:
1638         btrfs_release_path(path);
1639         if (!ret && update_size) {
1640                 btrfs_i_size_write(dir, dir->i_size + name_len * 2);
1641                 ret = btrfs_update_inode(trans, root, dir);
1642         }
1643         kfree(name);
1644         iput(dir);
1645         return ret;
1646
1647 insert:
1648         btrfs_release_path(path);
1649         ret = insert_one_name(trans, root, path, key->objectid, key->offset,
1650                               name, name_len, log_type, &log_key);
1651         if (ret && ret != -ENOENT)
1652                 goto out;
1653         update_size = false;
1654         ret = 0;
1655         goto out;
1656 }
1657
1658 /*
1659  * find all the names in a directory item and reconcile them into
1660  * the subvolume.  Only BTRFS_DIR_ITEM_KEY types will have more than
1661  * one name in a directory item, but the same code gets used for
1662  * both directory index types
1663  */
1664 static noinline int replay_one_dir_item(struct btrfs_trans_handle *trans,
1665                                         struct btrfs_root *root,
1666                                         struct btrfs_path *path,
1667                                         struct extent_buffer *eb, int slot,
1668                                         struct btrfs_key *key)
1669 {
1670         int ret;
1671         u32 item_size = btrfs_item_size_nr(eb, slot);
1672         struct btrfs_dir_item *di;
1673         int name_len;
1674         unsigned long ptr;
1675         unsigned long ptr_end;
1676
1677         ptr = btrfs_item_ptr_offset(eb, slot);
1678         ptr_end = ptr + item_size;
1679         while (ptr < ptr_end) {
1680                 di = (struct btrfs_dir_item *)ptr;
1681                 if (verify_dir_item(root, eb, di))
1682                         return -EIO;
1683                 name_len = btrfs_dir_name_len(eb, di);
1684                 ret = replay_one_name(trans, root, path, eb, di, key);
1685                 if (ret)
1686                         return ret;
1687                 ptr = (unsigned long)(di + 1);
1688                 ptr += name_len;
1689         }
1690         return 0;
1691 }
1692
1693 /*
1694  * directory replay has two parts.  There are the standard directory
1695  * items in the log copied from the subvolume, and range items
1696  * created in the log while the subvolume was logged.
1697  *
1698  * The range items tell us which parts of the key space the log
1699  * is authoritative for.  During replay, if a key in the subvolume
1700  * directory is in a logged range item, but not actually in the log
1701  * that means it was deleted from the directory before the fsync
1702  * and should be removed.
1703  */
1704 static noinline int find_dir_range(struct btrfs_root *root,
1705                                    struct btrfs_path *path,
1706                                    u64 dirid, int key_type,
1707                                    u64 *start_ret, u64 *end_ret)
1708 {
1709         struct btrfs_key key;
1710         u64 found_end;
1711         struct btrfs_dir_log_item *item;
1712         int ret;
1713         int nritems;
1714
1715         if (*start_ret == (u64)-1)
1716                 return 1;
1717
1718         key.objectid = dirid;
1719         key.type = key_type;
1720         key.offset = *start_ret;
1721
1722         ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
1723         if (ret < 0)
1724                 goto out;
1725         if (ret > 0) {
1726                 if (path->slots[0] == 0)
1727                         goto out;
1728                 path->slots[0]--;
1729         }
1730         if (ret != 0)
1731                 btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
1732
1733         if (key.type != key_type || key.objectid != dirid) {
1734                 ret = 1;
1735                 goto next;
1736         }
1737         item = btrfs_item_ptr(path->nodes[0], path->slots[0],
1738                               struct btrfs_dir_log_item);
1739         found_end = btrfs_dir_log_end(path->nodes[0], item);
1740
1741         if (*start_ret >= key.offset && *start_ret <= found_end) {
1742                 ret = 0;
1743                 *start_ret = key.offset;
1744                 *end_ret = found_end;
1745                 goto out;
1746         }
1747         ret = 1;
1748 next:
1749         /* check the next slot in the tree to see if it is a valid item */
1750         nritems = btrfs_header_nritems(path->nodes[0]);
1751         if (path->slots[0] >= nritems) {
1752                 ret = btrfs_next_leaf(root, path);
1753                 if (ret)
1754                         goto out;
1755         } else {
1756                 path->slots[0]++;
1757         }
1758
1759         btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
1760
1761         if (key.type != key_type || key.objectid != dirid) {
1762                 ret = 1;
1763                 goto out;
1764         }
1765         item = btrfs_item_ptr(path->nodes[0], path->slots[0],
1766                               struct btrfs_dir_log_item);
1767         found_end = btrfs_dir_log_end(path->nodes[0], item);
1768         *start_ret = key.offset;
1769         *end_ret = found_end;
1770         ret = 0;
1771 out:
1772         btrfs_release_path(path);
1773         return ret;
1774 }
1775
1776 /*
1777  * this looks for a given directory item in the log.  If the directory
1778  * item is not in the log, the item is removed and the inode it points
1779  * to is unlinked
1780  */
1781 static noinline int check_item_in_log(struct btrfs_trans_handle *trans,
1782                                       struct btrfs_root *root,
1783                                       struct btrfs_root *log,
1784                                       struct btrfs_path *path,
1785                                       struct btrfs_path *log_path,
1786                                       struct inode *dir,
1787                                       struct btrfs_key *dir_key)
1788 {
1789         int ret;
1790         struct extent_buffer *eb;
1791         int slot;
1792         u32 item_size;
1793         struct btrfs_dir_item *di;
1794         struct btrfs_dir_item *log_di;
1795         int name_len;
1796         unsigned long ptr;
1797         unsigned long ptr_end;
1798         char *name;
1799         struct inode *inode;
1800         struct btrfs_key location;
1801
1802 again:
1803         eb = path->nodes[0];
1804         slot = path->slots[0];
1805         item_size = btrfs_item_size_nr(eb, slot);
1806         ptr = btrfs_item_ptr_offset(eb, slot);
1807         ptr_end = ptr + item_size;
1808         while (ptr < ptr_end) {
1809                 di = (struct btrfs_dir_item *)ptr;
1810                 if (verify_dir_item(root, eb, di)) {
1811                         ret = -EIO;
1812                         goto out;
1813                 }
1814
1815                 name_len = btrfs_dir_name_len(eb, di);
1816                 name = kmalloc(name_len, GFP_NOFS);
1817                 if (!name) {
1818                         ret = -ENOMEM;
1819                         goto out;
1820                 }
1821                 read_extent_buffer(eb, name, (unsigned long)(di + 1),
1822                                   name_len);
1823                 log_di = NULL;
1824                 if (log && dir_key->type == BTRFS_DIR_ITEM_KEY) {
1825                         log_di = btrfs_lookup_dir_item(trans, log, log_path,
1826                                                        dir_key->objectid,
1827                                                        name, name_len, 0);
1828                 } else if (log && dir_key->type == BTRFS_DIR_INDEX_KEY) {
1829                         log_di = btrfs_lookup_dir_index_item(trans, log,
1830                                                      log_path,
1831                                                      dir_key->objectid,
1832                                                      dir_key->offset,
1833                                                      name, name_len, 0);
1834                 }
1835                 if (!log_di || (IS_ERR(log_di) && PTR_ERR(log_di) == -ENOENT)) {
1836                         btrfs_dir_item_key_to_cpu(eb, di, &location);
1837                         btrfs_release_path(path);
1838                         btrfs_release_path(log_path);
1839                         inode = read_one_inode(root, location.objectid);
1840                         if (!inode) {
1841                                 kfree(name);
1842                                 return -EIO;
1843                         }
1844
1845                         ret = link_to_fixup_dir(trans, root,
1846                                                 path, location.objectid);
1847                         if (ret) {
1848                                 kfree(name);
1849                                 iput(inode);
1850                                 goto out;
1851                         }
1852
1853                         inc_nlink(inode);
1854                         ret = btrfs_unlink_inode(trans, root, dir, inode,
1855                                                  name, name_len);
1856                         if (!ret)
1857                                 ret = btrfs_run_delayed_items(trans, root);
1858                         kfree(name);
1859                         iput(inode);
1860                         if (ret)
1861                                 goto out;
1862
1863                         /* there might still be more names under this key
1864                          * check and repeat if required
1865                          */
1866                         ret = btrfs_search_slot(NULL, root, dir_key, path,
1867                                                 0, 0);
1868                         if (ret == 0)
1869                                 goto again;
1870                         ret = 0;
1871                         goto out;
1872                 } else if (IS_ERR(log_di)) {
1873                         kfree(name);
1874                         return PTR_ERR(log_di);
1875                 }
1876                 btrfs_release_path(log_path);
1877                 kfree(name);
1878
1879                 ptr = (unsigned long)(di + 1);
1880                 ptr += name_len;
1881         }
1882         ret = 0;
1883 out:
1884         btrfs_release_path(path);
1885         btrfs_release_path(log_path);
1886         return ret;
1887 }
1888
1889 /*
1890  * deletion replay happens before we copy any new directory items
1891  * out of the log or out of backreferences from inodes.  It
1892  * scans the log to find ranges of keys that log is authoritative for,
1893  * and then scans the directory to find items in those ranges that are
1894  * not present in the log.
1895  *
1896  * Anything we don't find in the log is unlinked and removed from the
1897  * directory.
1898  */
1899 static noinline int replay_dir_deletes(struct btrfs_trans_handle *trans,
1900                                        struct btrfs_root *root,
1901                                        struct btrfs_root *log,
1902                                        struct btrfs_path *path,
1903                                        u64 dirid, int del_all)
1904 {
1905         u64 range_start;
1906         u64 range_end;
1907         int key_type = BTRFS_DIR_LOG_ITEM_KEY;
1908         int ret = 0;
1909         struct btrfs_key dir_key;
1910         struct btrfs_key found_key;
1911         struct btrfs_path *log_path;
1912         struct inode *dir;
1913
1914         dir_key.objectid = dirid;
1915         dir_key.type = BTRFS_DIR_ITEM_KEY;
1916         log_path = btrfs_alloc_path();
1917         if (!log_path)
1918                 return -ENOMEM;
1919
1920         dir = read_one_inode(root, dirid);
1921         /* it isn't an error if the inode isn't there, that can happen
1922          * because we replay the deletes before we copy in the inode item
1923          * from the log
1924          */
1925         if (!dir) {
1926                 btrfs_free_path(log_path);
1927                 return 0;
1928         }
1929 again:
1930         range_start = 0;
1931         range_end = 0;
1932         while (1) {
1933                 if (del_all)
1934                         range_end = (u64)-1;
1935                 else {
1936                         ret = find_dir_range(log, path, dirid, key_type,
1937                                              &range_start, &range_end);
1938                         if (ret != 0)
1939                                 break;
1940                 }
1941
1942                 dir_key.offset = range_start;
1943                 while (1) {
1944                         int nritems;
1945                         ret = btrfs_search_slot(NULL, root, &dir_key, path,
1946                                                 0, 0);
1947                         if (ret < 0)
1948                                 goto out;
1949
1950                         nritems = btrfs_header_nritems(path->nodes[0]);
1951                         if (path->slots[0] >= nritems) {
1952                                 ret = btrfs_next_leaf(root, path);
1953                                 if (ret)
1954                                         break;
1955                         }
1956                         btrfs_item_key_to_cpu(path->nodes[0], &found_key,
1957                                               path->slots[0]);
1958                         if (found_key.objectid != dirid ||
1959                             found_key.type != dir_key.type)
1960                                 goto next_type;
1961
1962                         if (found_key.offset > range_end)
1963                                 break;
1964
1965                         ret = check_item_in_log(trans, root, log, path,
1966                                                 log_path, dir,
1967                                                 &found_key);
1968                         if (ret)
1969                                 goto out;
1970                         if (found_key.offset == (u64)-1)
1971                                 break;
1972                         dir_key.offset = found_key.offset + 1;
1973                 }
1974                 btrfs_release_path(path);
1975                 if (range_end == (u64)-1)
1976                         break;
1977                 range_start = range_end + 1;
1978         }
1979
1980 next_type:
1981         ret = 0;
1982         if (key_type == BTRFS_DIR_LOG_ITEM_KEY) {
1983                 key_type = BTRFS_DIR_LOG_INDEX_KEY;
1984                 dir_key.type = BTRFS_DIR_INDEX_KEY;
1985                 btrfs_release_path(path);
1986                 goto again;
1987         }
1988 out:
1989         btrfs_release_path(path);
1990         btrfs_free_path(log_path);
1991         iput(dir);
1992         return ret;
1993 }
1994
1995 /*
1996  * the process_func used to replay items from the log tree.  This
1997  * gets called in two different stages.  The first stage just looks
1998  * for inodes and makes sure they are all copied into the subvolume.
1999  *
2000  * The second stage copies all the other item types from the log into
2001  * the subvolume.  The two stage approach is slower, but gets rid of
2002  * lots of complexity around inodes referencing other inodes that exist
2003  * only in the log (references come from either directory items or inode
2004  * back refs).
2005  */
2006 static int replay_one_buffer(struct btrfs_root *log, struct extent_buffer *eb,
2007                              struct walk_control *wc, u64 gen)
2008 {
2009         int nritems;
2010         struct btrfs_path *path;
2011         struct btrfs_root *root = wc->replay_dest;
2012         struct btrfs_key key;
2013         int level;
2014         int i;
2015         int ret;
2016
2017         ret = btrfs_read_buffer(eb, gen);
2018         if (ret)
2019                 return ret;
2020
2021         level = btrfs_header_level(eb);
2022
2023         if (level != 0)
2024                 return 0;
2025
2026         path = btrfs_alloc_path();
2027         if (!path)
2028                 return -ENOMEM;
2029
2030         nritems = btrfs_header_nritems(eb);
2031         for (i = 0; i < nritems; i++) {
2032                 btrfs_item_key_to_cpu(eb, &key, i);
2033
2034                 /* inode keys are done during the first stage */
2035                 if (key.type == BTRFS_INODE_ITEM_KEY &&
2036                     wc->stage == LOG_WALK_REPLAY_INODES) {
2037                         struct btrfs_inode_item *inode_item;
2038                         u32 mode;
2039
2040                         inode_item = btrfs_item_ptr(eb, i,
2041                                             struct btrfs_inode_item);
2042                         mode = btrfs_inode_mode(eb, inode_item);
2043                         if (S_ISDIR(mode)) {
2044                                 ret = replay_dir_deletes(wc->trans,
2045                                          root, log, path, key.objectid, 0);
2046                                 if (ret)
2047                                         break;
2048                         }
2049                         ret = overwrite_item(wc->trans, root, path,
2050                                              eb, i, &key);
2051                         if (ret)
2052                                 break;
2053
2054                         /* for regular files, make sure corresponding
2055                          * orhpan item exist. extents past the new EOF
2056                          * will be truncated later by orphan cleanup.
2057                          */
2058                         if (S_ISREG(mode)) {
2059                                 ret = insert_orphan_item(wc->trans, root,
2060                                                          key.objectid);
2061                                 if (ret)
2062                                         break;
2063                         }
2064
2065                         ret = link_to_fixup_dir(wc->trans, root,
2066                                                 path, key.objectid);
2067                         if (ret)
2068                                 break;
2069                 }
2070
2071                 if (key.type == BTRFS_DIR_INDEX_KEY &&
2072                     wc->stage == LOG_WALK_REPLAY_DIR_INDEX) {
2073                         ret = replay_one_dir_item(wc->trans, root, path,
2074                                                   eb, i, &key);
2075                         if (ret)
2076                                 break;
2077                 }
2078
2079                 if (wc->stage < LOG_WALK_REPLAY_ALL)
2080                         continue;
2081
2082                 /* these keys are simply copied */
2083                 if (key.type == BTRFS_XATTR_ITEM_KEY) {
2084                         ret = overwrite_item(wc->trans, root, path,
2085                                              eb, i, &key);
2086                         if (ret)
2087                                 break;
2088                 } else if (key.type == BTRFS_INODE_REF_KEY ||
2089                            key.type == BTRFS_INODE_EXTREF_KEY) {
2090                         ret = add_inode_ref(wc->trans, root, log, path,
2091                                             eb, i, &key);
2092                         if (ret && ret != -ENOENT)
2093                                 break;
2094                         ret = 0;
2095                 } else if (key.type == BTRFS_EXTENT_DATA_KEY) {
2096                         ret = replay_one_extent(wc->trans, root, path,
2097                                                 eb, i, &key);
2098                         if (ret)
2099                                 break;
2100                 } else if (key.type == BTRFS_DIR_ITEM_KEY) {
2101                         ret = replay_one_dir_item(wc->trans, root, path,
2102                                                   eb, i, &key);
2103                         if (ret)
2104                                 break;
2105                 }
2106         }
2107         btrfs_free_path(path);
2108         return ret;
2109 }
2110
2111 static noinline int walk_down_log_tree(struct btrfs_trans_handle *trans,
2112                                    struct btrfs_root *root,
2113                                    struct btrfs_path *path, int *level,
2114                                    struct walk_control *wc)
2115 {
2116         u64 root_owner;
2117         u64 bytenr;
2118         u64 ptr_gen;
2119         struct extent_buffer *next;
2120         struct extent_buffer *cur;
2121         struct extent_buffer *parent;
2122         u32 blocksize;
2123         int ret = 0;
2124
2125         WARN_ON(*level < 0);
2126         WARN_ON(*level >= BTRFS_MAX_LEVEL);
2127
2128         while (*level > 0) {
2129                 WARN_ON(*level < 0);
2130                 WARN_ON(*level >= BTRFS_MAX_LEVEL);
2131                 cur = path->nodes[*level];
2132
2133                 WARN_ON(btrfs_header_level(cur) != *level);
2134
2135                 if (path->slots[*level] >=
2136                     btrfs_header_nritems(cur))
2137                         break;
2138
2139                 bytenr = btrfs_node_blockptr(cur, path->slots[*level]);
2140                 ptr_gen = btrfs_node_ptr_generation(cur, path->slots[*level]);
2141                 blocksize = btrfs_level_size(root, *level - 1);
2142
2143                 parent = path->nodes[*level];
2144                 root_owner = btrfs_header_owner(parent);
2145
2146                 next = btrfs_find_create_tree_block(root, bytenr, blocksize);
2147                 if (!next)
2148                         return -ENOMEM;
2149
2150                 if (*level == 1) {
2151                         ret = wc->process_func(root, next, wc, ptr_gen);
2152                         if (ret) {
2153                                 free_extent_buffer(next);
2154                                 return ret;
2155                         }
2156
2157                         path->slots[*level]++;
2158                         if (wc->free) {
2159                                 ret = btrfs_read_buffer(next, ptr_gen);
2160                                 if (ret) {
2161                                         free_extent_buffer(next);
2162                                         return ret;
2163                                 }
2164
2165                                 if (trans) {
2166                                         btrfs_tree_lock(next);
2167                                         btrfs_set_lock_blocking(next);
2168                                         clean_tree_block(trans, root, next);
2169                                         btrfs_wait_tree_block_writeback(next);
2170                                         btrfs_tree_unlock(next);
2171                                 }
2172
2173                                 WARN_ON(root_owner !=
2174                                         BTRFS_TREE_LOG_OBJECTID);
2175                                 ret = btrfs_free_and_pin_reserved_extent(root,
2176                                                          bytenr, blocksize);
2177                                 if (ret) {
2178                                         free_extent_buffer(next);
2179                                         return ret;
2180                                 }
2181                         }
2182                         free_extent_buffer(next);
2183                         continue;
2184                 }
2185                 ret = btrfs_read_buffer(next, ptr_gen);
2186                 if (ret) {
2187                         free_extent_buffer(next);
2188                         return ret;
2189                 }
2190
2191                 WARN_ON(*level <= 0);
2192                 if (path->nodes[*level-1])
2193                         free_extent_buffer(path->nodes[*level-1]);
2194                 path->nodes[*level-1] = next;
2195                 *level = btrfs_header_level(next);
2196                 path->slots[*level] = 0;
2197                 cond_resched();
2198         }
2199         WARN_ON(*level < 0);
2200         WARN_ON(*level >= BTRFS_MAX_LEVEL);
2201
2202         path->slots[*level] = btrfs_header_nritems(path->nodes[*level]);
2203
2204         cond_resched();
2205         return 0;
2206 }
2207
2208 static noinline int walk_up_log_tree(struct btrfs_trans_handle *trans,
2209                                  struct btrfs_root *root,
2210                                  struct btrfs_path *path, int *level,
2211                                  struct walk_control *wc)
2212 {
2213         u64 root_owner;
2214         int i;
2215         int slot;
2216         int ret;
2217
2218         for (i = *level; i < BTRFS_MAX_LEVEL - 1 && path->nodes[i]; i++) {
2219                 slot = path->slots[i];
2220                 if (slot + 1 < btrfs_header_nritems(path->nodes[i])) {
2221                         path->slots[i]++;
2222                         *level = i;
2223                         WARN_ON(*level == 0);
2224                         return 0;
2225                 } else {
2226                         struct extent_buffer *parent;
2227                         if (path->nodes[*level] == root->node)
2228                                 parent = path->nodes[*level];
2229                         else
2230                                 parent = path->nodes[*level + 1];
2231
2232                         root_owner = btrfs_header_owner(parent);
2233                         ret = wc->process_func(root, path->nodes[*level], wc,
2234                                  btrfs_header_generation(path->nodes[*level]));
2235                         if (ret)
2236                                 return ret;
2237
2238                         if (wc->free) {
2239                                 struct extent_buffer *next;
2240
2241                                 next = path->nodes[*level];
2242
2243                                 if (trans) {
2244                                         btrfs_tree_lock(next);
2245                                         btrfs_set_lock_blocking(next);
2246                                         clean_tree_block(trans, root, next);
2247                                         btrfs_wait_tree_block_writeback(next);
2248                                         btrfs_tree_unlock(next);
2249                                 }
2250
2251                                 WARN_ON(root_owner != BTRFS_TREE_LOG_OBJECTID);
2252                                 ret = btrfs_free_and_pin_reserved_extent(root,
2253                                                 path->nodes[*level]->start,
2254                                                 path->nodes[*level]->len);
2255                                 if (ret)
2256                                         return ret;
2257                         }
2258                         free_extent_buffer(path->nodes[*level]);
2259                         path->nodes[*level] = NULL;
2260                         *level = i + 1;
2261                 }
2262         }
2263         return 1;
2264 }
2265
2266 /*
2267  * drop the reference count on the tree rooted at 'snap'.  This traverses
2268  * the tree freeing any blocks that have a ref count of zero after being
2269  * decremented.
2270  */
2271 static int walk_log_tree(struct btrfs_trans_handle *trans,
2272                          struct btrfs_root *log, struct walk_control *wc)
2273 {
2274         int ret = 0;
2275         int wret;
2276         int level;
2277         struct btrfs_path *path;
2278         int orig_level;
2279
2280         path = btrfs_alloc_path();
2281         if (!path)
2282                 return -ENOMEM;
2283
2284         level = btrfs_header_level(log->node);
2285         orig_level = level;
2286         path->nodes[level] = log->node;
2287         extent_buffer_get(log->node);
2288         path->slots[level] = 0;
2289
2290         while (1) {
2291                 wret = walk_down_log_tree(trans, log, path, &level, wc);
2292                 if (wret > 0)
2293                         break;
2294                 if (wret < 0) {
2295                         ret = wret;
2296                         goto out;
2297                 }
2298
2299                 wret = walk_up_log_tree(trans, log, path, &level, wc);
2300                 if (wret > 0)
2301                         break;
2302                 if (wret < 0) {
2303                         ret = wret;
2304                         goto out;
2305                 }
2306         }
2307
2308         /* was the root node processed? if not, catch it here */
2309         if (path->nodes[orig_level]) {
2310                 ret = wc->process_func(log, path->nodes[orig_level], wc,
2311                          btrfs_header_generation(path->nodes[orig_level]));
2312                 if (ret)
2313                         goto out;
2314                 if (wc->free) {
2315                         struct extent_buffer *next;
2316
2317                         next = path->nodes[orig_level];
2318
2319                         if (trans) {
2320                                 btrfs_tree_lock(next);
2321                                 btrfs_set_lock_blocking(next);
2322                                 clean_tree_block(trans, log, next);
2323                                 btrfs_wait_tree_block_writeback(next);
2324                                 btrfs_tree_unlock(next);
2325                         }
2326
2327                         WARN_ON(log->root_key.objectid !=
2328                                 BTRFS_TREE_LOG_OBJECTID);
2329                         ret = btrfs_free_and_pin_reserved_extent(log, next->start,
2330                                                          next->len);
2331                         if (ret)
2332                                 goto out;
2333                 }
2334         }
2335
2336 out:
2337         btrfs_free_path(path);
2338         return ret;
2339 }
2340
2341 /*
2342  * helper function to update the item for a given subvolumes log root
2343  * in the tree of log roots
2344  */
2345 static int update_log_root(struct btrfs_trans_handle *trans,
2346                            struct btrfs_root *log)
2347 {
2348         int ret;
2349
2350         if (log->log_transid == 1) {
2351                 /* insert root item on the first sync */
2352                 ret = btrfs_insert_root(trans, log->fs_info->log_root_tree,
2353                                 &log->root_key, &log->root_item);
2354         } else {
2355                 ret = btrfs_update_root(trans, log->fs_info->log_root_tree,
2356                                 &log->root_key, &log->root_item);
2357         }
2358         return ret;
2359 }
2360
2361 static int wait_log_commit(struct btrfs_trans_handle *trans,
2362                            struct btrfs_root *root, unsigned long transid)
2363 {
2364         DEFINE_WAIT(wait);
2365         int index = transid % 2;
2366
2367         /*
2368          * we only allow two pending log transactions at a time,
2369          * so we know that if ours is more than 2 older than the
2370          * current transaction, we're done
2371          */
2372         do {
2373                 prepare_to_wait(&root->log_commit_wait[index],
2374                                 &wait, TASK_UNINTERRUPTIBLE);
2375                 mutex_unlock(&root->log_mutex);
2376
2377                 if (root->fs_info->last_trans_log_full_commit !=
2378                     trans->transid && root->log_transid < transid + 2 &&
2379                     atomic_read(&root->log_commit[index]))
2380                         schedule();
2381
2382                 finish_wait(&root->log_commit_wait[index], &wait);
2383                 mutex_lock(&root->log_mutex);
2384         } while (root->fs_info->last_trans_log_full_commit !=
2385                  trans->transid && root->log_transid < transid + 2 &&
2386                  atomic_read(&root->log_commit[index]));
2387         return 0;
2388 }
2389
2390 static void wait_for_writer(struct btrfs_trans_handle *trans,
2391                             struct btrfs_root *root)
2392 {
2393         DEFINE_WAIT(wait);
2394         while (root->fs_info->last_trans_log_full_commit !=
2395                trans->transid && atomic_read(&root->log_writers)) {
2396                 prepare_to_wait(&root->log_writer_wait,
2397                                 &wait, TASK_UNINTERRUPTIBLE);
2398                 mutex_unlock(&root->log_mutex);
2399                 if (root->fs_info->last_trans_log_full_commit !=
2400                     trans->transid && atomic_read(&root->log_writers))
2401                         schedule();
2402                 mutex_lock(&root->log_mutex);
2403                 finish_wait(&root->log_writer_wait, &wait);
2404         }
2405 }
2406
2407 /*
2408  * btrfs_sync_log does sends a given tree log down to the disk and
2409  * updates the super blocks to record it.  When this call is done,
2410  * you know that any inodes previously logged are safely on disk only
2411  * if it returns 0.
2412  *
2413  * Any other return value means you need to call btrfs_commit_transaction.
2414  * Some of the edge cases for fsyncing directories that have had unlinks
2415  * or renames done in the past mean that sometimes the only safe
2416  * fsync is to commit the whole FS.  When btrfs_sync_log returns -EAGAIN,
2417  * that has happened.
2418  */
2419 int btrfs_sync_log(struct btrfs_trans_handle *trans,
2420                    struct btrfs_root *root)
2421 {
2422         int index1;
2423         int index2;
2424         int mark;
2425         int ret;
2426         struct btrfs_root *log = root->log_root;
2427         struct btrfs_root *log_root_tree = root->fs_info->log_root_tree;
2428         unsigned long log_transid = 0;
2429         struct blk_plug plug;
2430
2431         mutex_lock(&root->log_mutex);
2432         log_transid = root->log_transid;
2433         index1 = root->log_transid % 2;
2434         if (atomic_read(&root->log_commit[index1])) {
2435                 wait_log_commit(trans, root, root->log_transid);
2436                 mutex_unlock(&root->log_mutex);
2437                 return 0;
2438         }
2439         atomic_set(&root->log_commit[index1], 1);
2440
2441         /* wait for previous tree log sync to complete */
2442         if (atomic_read(&root->log_commit[(index1 + 1) % 2]))
2443                 wait_log_commit(trans, root, root->log_transid - 1);
2444         while (1) {
2445                 int batch = atomic_read(&root->log_batch);
2446                 /* when we're on an ssd, just kick the log commit out */
2447                 if (!btrfs_test_opt(root, SSD) && root->log_multiple_pids) {
2448                         mutex_unlock(&root->log_mutex);
2449                         schedule_timeout_uninterruptible(1);
2450                         mutex_lock(&root->log_mutex);
2451                 }
2452                 wait_for_writer(trans, root);
2453                 if (batch == atomic_read(&root->log_batch))
2454                         break;
2455         }
2456
2457         /* bail out if we need to do a full commit */
2458         if (root->fs_info->last_trans_log_full_commit == trans->transid) {
2459                 ret = -EAGAIN;
2460                 btrfs_free_logged_extents(log, log_transid);
2461                 mutex_unlock(&root->log_mutex);
2462                 goto out;
2463         }
2464
2465         if (log_transid % 2 == 0)
2466                 mark = EXTENT_DIRTY;
2467         else
2468                 mark = EXTENT_NEW;
2469
2470         /* we start IO on  all the marked extents here, but we don't actually
2471          * wait for them until later.
2472          */
2473         blk_start_plug(&plug);
2474         ret = btrfs_write_marked_extents(log, &log->dirty_log_pages, mark);
2475         if (ret) {
2476                 blk_finish_plug(&plug);
2477                 btrfs_abort_transaction(trans, root, ret);
2478                 btrfs_free_logged_extents(log, log_transid);
2479                 mutex_unlock(&root->log_mutex);
2480                 goto out;
2481         }
2482
2483         btrfs_set_root_node(&log->root_item, log->node);
2484
2485         root->log_transid++;
2486         log->log_transid = root->log_transid;
2487         root->log_start_pid = 0;
2488         smp_mb();
2489         /*
2490          * IO has been started, blocks of the log tree have WRITTEN flag set
2491          * in their headers. new modifications of the log will be written to
2492          * new positions. so it's safe to allow log writers to go in.
2493          */
2494         mutex_unlock(&root->log_mutex);
2495
2496         mutex_lock(&log_root_tree->log_mutex);
2497         atomic_inc(&log_root_tree->log_batch);
2498         atomic_inc(&log_root_tree->log_writers);
2499         mutex_unlock(&log_root_tree->log_mutex);
2500
2501         ret = update_log_root(trans, log);
2502
2503         mutex_lock(&log_root_tree->log_mutex);
2504         if (atomic_dec_and_test(&log_root_tree->log_writers)) {
2505                 smp_mb();
2506                 if (waitqueue_active(&log_root_tree->log_writer_wait))
2507                         wake_up(&log_root_tree->log_writer_wait);
2508         }
2509
2510         if (ret) {
2511                 blk_finish_plug(&plug);
2512                 if (ret != -ENOSPC) {
2513                         btrfs_abort_transaction(trans, root, ret);
2514                         mutex_unlock(&log_root_tree->log_mutex);
2515                         goto out;
2516                 }
2517                 root->fs_info->last_trans_log_full_commit = trans->transid;
2518                 btrfs_wait_marked_extents(log, &log->dirty_log_pages, mark);
2519                 btrfs_free_logged_extents(log, log_transid);
2520                 mutex_unlock(&log_root_tree->log_mutex);
2521                 ret = -EAGAIN;
2522                 goto out;
2523         }
2524
2525         index2 = log_root_tree->log_transid % 2;
2526         if (atomic_read(&log_root_tree->log_commit[index2])) {
2527                 blk_finish_plug(&plug);
2528                 btrfs_wait_marked_extents(log, &log->dirty_log_pages, mark);
2529                 wait_log_commit(trans, log_root_tree,
2530                                 log_root_tree->log_transid);
2531                 btrfs_free_logged_extents(log, log_transid);
2532                 mutex_unlock(&log_root_tree->log_mutex);
2533                 ret = 0;
2534                 goto out;
2535         }
2536         atomic_set(&log_root_tree->log_commit[index2], 1);
2537
2538         if (atomic_read(&log_root_tree->log_commit[(index2 + 1) % 2])) {
2539                 wait_log_commit(trans, log_root_tree,
2540                                 log_root_tree->log_transid - 1);
2541         }
2542
2543         wait_for_writer(trans, log_root_tree);
2544
2545         /*
2546          * now that we've moved on to the tree of log tree roots,
2547          * check the full commit flag again
2548          */
2549         if (root->fs_info->last_trans_log_full_commit == trans->transid) {
2550                 blk_finish_plug(&plug);
2551                 btrfs_wait_marked_extents(log, &log->dirty_log_pages, mark);
2552                 btrfs_free_logged_extents(log, log_transid);
2553                 mutex_unlock(&log_root_tree->log_mutex);
2554                 ret = -EAGAIN;
2555                 goto out_wake_log_root;
2556         }
2557
2558         ret = btrfs_write_marked_extents(log_root_tree,
2559                                          &log_root_tree->dirty_log_pages,
2560                                          EXTENT_DIRTY | EXTENT_NEW);
2561         blk_finish_plug(&plug);
2562         if (ret) {
2563                 btrfs_abort_transaction(trans, root, ret);
2564                 btrfs_free_logged_extents(log, log_transid);
2565                 mutex_unlock(&log_root_tree->log_mutex);
2566                 goto out_wake_log_root;
2567         }
2568         btrfs_wait_marked_extents(log, &log->dirty_log_pages, mark);
2569         btrfs_wait_marked_extents(log_root_tree,
2570                                   &log_root_tree->dirty_log_pages,
2571                                   EXTENT_NEW | EXTENT_DIRTY);
2572         btrfs_wait_logged_extents(log, log_transid);
2573
2574         btrfs_set_super_log_root(root->fs_info->super_for_commit,
2575                                 log_root_tree->node->start);
2576         btrfs_set_super_log_root_level(root->fs_info->super_for_commit,
2577                                 btrfs_header_level(log_root_tree->node));
2578
2579         log_root_tree->log_transid++;
2580         smp_mb();
2581
2582         mutex_unlock(&log_root_tree->log_mutex);
2583
2584         /*
2585          * nobody else is going to jump in and write the the ctree
2586          * super here because the log_commit atomic below is protecting
2587          * us.  We must be called with a transaction handle pinning
2588          * the running transaction open, so a full commit can't hop
2589          * in and cause problems either.
2590          */
2591         ret = write_ctree_super(trans, root->fs_info->tree_root, 1);
2592         if (ret) {
2593                 btrfs_abort_transaction(trans, root, ret);
2594                 goto out_wake_log_root;
2595         }
2596
2597         mutex_lock(&root->log_mutex);
2598         if (root->last_log_commit < log_transid)
2599                 root->last_log_commit = log_transid;
2600         mutex_unlock(&root->log_mutex);
2601
2602 out_wake_log_root:
2603         atomic_set(&log_root_tree->log_commit[index2], 0);
2604         smp_mb();
2605         if (waitqueue_active(&log_root_tree->log_commit_wait[index2]))
2606                 wake_up(&log_root_tree->log_commit_wait[index2]);
2607 out:
2608         atomic_set(&root->log_commit[index1], 0);
2609         smp_mb();
2610         if (waitqueue_active(&root->log_commit_wait[index1]))
2611                 wake_up(&root->log_commit_wait[index1]);
2612         return ret;
2613 }
2614
2615 static void free_log_tree(struct btrfs_trans_handle *trans,
2616                           struct btrfs_root *log)
2617 {
2618         int ret;
2619         u64 start;
2620         u64 end;
2621         struct walk_control wc = {
2622                 .free = 1,
2623                 .process_func = process_one_buffer
2624         };
2625
2626         ret = walk_log_tree(trans, log, &wc);
2627         /* I don't think this can happen but just in case */
2628         if (ret)
2629                 btrfs_abort_transaction(trans, log, ret);
2630
2631         while (1) {
2632                 ret = find_first_extent_bit(&log->dirty_log_pages,
2633                                 0, &start, &end, EXTENT_DIRTY | EXTENT_NEW,
2634                                 NULL);
2635                 if (ret)
2636                         break;
2637
2638                 clear_extent_bits(&log->dirty_log_pages, start, end,
2639                                   EXTENT_DIRTY | EXTENT_NEW, GFP_NOFS);
2640         }
2641
2642         /*
2643          * We may have short-circuited the log tree with the full commit logic
2644          * and left ordered extents on our list, so clear these out to keep us
2645          * from leaking inodes and memory.
2646          */
2647         btrfs_free_logged_extents(log, 0);
2648         btrfs_free_logged_extents(log, 1);
2649
2650         free_extent_buffer(log->node);
2651         kfree(log);
2652 }
2653
2654 /*
2655  * free all the extents used by the tree log.  This should be called
2656  * at commit time of the full transaction
2657  */
2658 int btrfs_free_log(struct btrfs_trans_handle *trans, struct btrfs_root *root)
2659 {
2660         if (root->log_root) {
2661                 free_log_tree(trans, root->log_root);
2662                 root->log_root = NULL;
2663         }
2664         return 0;
2665 }
2666
2667 int btrfs_free_log_root_tree(struct btrfs_trans_handle *trans,
2668                              struct btrfs_fs_info *fs_info)
2669 {
2670         if (fs_info->log_root_tree) {
2671                 free_log_tree(trans, fs_info->log_root_tree);
2672                 fs_info->log_root_tree = NULL;
2673         }
2674         return 0;
2675 }
2676
2677 /*
2678  * If both a file and directory are logged, and unlinks or renames are
2679  * mixed in, we have a few interesting corners:
2680  *
2681  * create file X in dir Y
2682  * link file X to X.link in dir Y
2683  * fsync file X
2684  * unlink file X but leave X.link
2685  * fsync dir Y
2686  *
2687  * After a crash we would expect only X.link to exist.  But file X
2688  * didn't get fsync'd again so the log has back refs for X and X.link.
2689  *
2690  * We solve this by removing directory entries and inode backrefs from the
2691  * log when a file that was logged in the current transaction is
2692  * unlinked.  Any later fsync will include the updated log entries, and
2693  * we'll be able to reconstruct the proper directory items from backrefs.
2694  *
2695  * This optimizations allows us to avoid relogging the entire inode
2696  * or the entire directory.
2697  */
2698 int btrfs_del_dir_entries_in_log(struct btrfs_trans_handle *trans,
2699                                  struct btrfs_root *root,
2700                                  const char *name, int name_len,
2701                                  struct inode *dir, u64 index)
2702 {
2703         struct btrfs_root *log;
2704         struct btrfs_dir_item *di;
2705         struct btrfs_path *path;
2706         int ret;
2707         int err = 0;
2708         int bytes_del = 0;
2709         u64 dir_ino = btrfs_ino(dir);
2710
2711         if (BTRFS_I(dir)->logged_trans < trans->transid)
2712                 return 0;
2713
2714         ret = join_running_log_trans(root);
2715         if (ret)
2716                 return 0;
2717
2718         mutex_lock(&BTRFS_I(dir)->log_mutex);
2719
2720         log = root->log_root;
2721         path = btrfs_alloc_path();
2722         if (!path) {
2723                 err = -ENOMEM;
2724                 goto out_unlock;
2725         }
2726
2727         di = btrfs_lookup_dir_item(trans, log, path, dir_ino,
2728                                    name, name_len, -1);
2729         if (IS_ERR(di)) {
2730                 err = PTR_ERR(di);
2731                 goto fail;
2732         }
2733         if (di) {
2734                 ret = btrfs_delete_one_dir_name(trans, log, path, di);
2735                 bytes_del += name_len;
2736                 if (ret) {
2737                         err = ret;
2738                         goto fail;
2739                 }
2740         }
2741         btrfs_release_path(path);
2742         di = btrfs_lookup_dir_index_item(trans, log, path, dir_ino,
2743                                          index, name, name_len, -1);
2744         if (IS_ERR(di)) {
2745                 err = PTR_ERR(di);
2746                 goto fail;
2747         }
2748         if (di) {
2749                 ret = btrfs_delete_one_dir_name(trans, log, path, di);
2750                 bytes_del += name_len;
2751                 if (ret) {
2752                         err = ret;
2753                         goto fail;
2754                 }
2755         }
2756
2757         /* update the directory size in the log to reflect the names
2758          * we have removed
2759          */
2760         if (bytes_del) {
2761                 struct btrfs_key key;
2762
2763                 key.objectid = dir_ino;
2764                 key.offset = 0;
2765                 key.type = BTRFS_INODE_ITEM_KEY;
2766                 btrfs_release_path(path);
2767
2768                 ret = btrfs_search_slot(trans, log, &key, path, 0, 1);
2769                 if (ret < 0) {
2770                         err = ret;
2771                         goto fail;
2772                 }
2773                 if (ret == 0) {
2774                         struct btrfs_inode_item *item;
2775                         u64 i_size;
2776
2777                         item = btrfs_item_ptr(path->nodes[0], path->slots[0],
2778                                               struct btrfs_inode_item);
2779                         i_size = btrfs_inode_size(path->nodes[0], item);
2780                         if (i_size > bytes_del)
2781                                 i_size -= bytes_del;
2782                         else
2783                                 i_size = 0;
2784                         btrfs_set_inode_size(path->nodes[0], item, i_size);
2785                         btrfs_mark_buffer_dirty(path->nodes[0]);
2786                 } else
2787                         ret = 0;
2788                 btrfs_release_path(path);
2789         }
2790 fail:
2791         btrfs_free_path(path);
2792 out_unlock:
2793         mutex_unlock(&BTRFS_I(dir)->log_mutex);
2794         if (ret == -ENOSPC) {
2795                 root->fs_info->last_trans_log_full_commit = trans->transid;
2796                 ret = 0;
2797         } else if (ret < 0)
2798                 btrfs_abort_transaction(trans, root, ret);
2799
2800         btrfs_end_log_trans(root);
2801
2802         return err;
2803 }
2804
2805 /* see comments for btrfs_del_dir_entries_in_log */
2806 int btrfs_del_inode_ref_in_log(struct btrfs_trans_handle *trans,
2807                                struct btrfs_root *root,
2808                                const char *name, int name_len,
2809                                struct inode *inode, u64 dirid)
2810 {
2811         struct btrfs_root *log;
2812         u64 index;
2813         int ret;
2814
2815         if (BTRFS_I(inode)->logged_trans < trans->transid)
2816                 return 0;
2817
2818         ret = join_running_log_trans(root);
2819         if (ret)
2820                 return 0;
2821         log = root->log_root;
2822         mutex_lock(&BTRFS_I(inode)->log_mutex);
2823
2824         ret = btrfs_del_inode_ref(trans, log, name, name_len, btrfs_ino(inode),
2825                                   dirid, &index);
2826         mutex_unlock(&BTRFS_I(inode)->log_mutex);
2827         if (ret == -ENOSPC) {
2828                 root->fs_info->last_trans_log_full_commit = trans->transid;
2829                 ret = 0;
2830         } else if (ret < 0 && ret != -ENOENT)
2831                 btrfs_abort_transaction(trans, root, ret);
2832         btrfs_end_log_trans(root);
2833
2834         return ret;
2835 }
2836
2837 /*
2838  * creates a range item in the log for 'dirid'.  first_offset and
2839  * last_offset tell us which parts of the key space the log should
2840  * be considered authoritative for.
2841  */
2842 static noinline int insert_dir_log_key(struct btrfs_trans_handle *trans,
2843                                        struct btrfs_root *log,
2844                                        struct btrfs_path *path,
2845                                        int key_type, u64 dirid,
2846                                        u64 first_offset, u64 last_offset)
2847 {
2848         int ret;
2849         struct btrfs_key key;
2850         struct btrfs_dir_log_item *item;
2851
2852         key.objectid = dirid;
2853         key.offset = first_offset;
2854         if (key_type == BTRFS_DIR_ITEM_KEY)
2855                 key.type = BTRFS_DIR_LOG_ITEM_KEY;
2856         else
2857                 key.type = BTRFS_DIR_LOG_INDEX_KEY;
2858         ret = btrfs_insert_empty_item(trans, log, path, &key, sizeof(*item));
2859         if (ret)
2860                 return ret;
2861
2862         item = btrfs_item_ptr(path->nodes[0], path->slots[0],
2863                               struct btrfs_dir_log_item);
2864         btrfs_set_dir_log_end(path->nodes[0], item, last_offset);
2865         btrfs_mark_buffer_dirty(path->nodes[0]);
2866         btrfs_release_path(path);
2867         return 0;
2868 }
2869
2870 /*
2871  * log all the items included in the current transaction for a given
2872  * directory.  This also creates the range items in the log tree required
2873  * to replay anything deleted before the fsync
2874  */
2875 static noinline int log_dir_items(struct btrfs_trans_handle *trans,
2876                           struct btrfs_root *root, struct inode *inode,
2877                           struct btrfs_path *path,
2878                           struct btrfs_path *dst_path, int key_type,
2879                           u64 min_offset, u64 *last_offset_ret)
2880 {
2881         struct btrfs_key min_key;
2882         struct btrfs_root *log = root->log_root;
2883         struct extent_buffer *src;
2884         int err = 0;
2885         int ret;
2886         int i;
2887         int nritems;
2888         u64 first_offset = min_offset;
2889         u64 last_offset = (u64)-1;
2890         u64 ino = btrfs_ino(inode);
2891
2892         log = root->log_root;
2893
2894         min_key.objectid = ino;
2895         min_key.type = key_type;
2896         min_key.offset = min_offset;
2897
2898         path->keep_locks = 1;
2899
2900         ret = btrfs_search_forward(root, &min_key, path, trans->transid);
2901
2902         /*
2903          * we didn't find anything from this transaction, see if there
2904          * is anything at all
2905          */
2906         if (ret != 0 || min_key.objectid != ino || min_key.type != key_type) {
2907                 min_key.objectid = ino;
2908                 min_key.type = key_type;
2909                 min_key.offset = (u64)-1;
2910                 btrfs_release_path(path);
2911                 ret = btrfs_search_slot(NULL, root, &min_key, path, 0, 0);
2912                 if (ret < 0) {
2913                         btrfs_release_path(path);
2914                         return ret;
2915                 }
2916                 ret = btrfs_previous_item(root, path, ino, key_type);
2917
2918                 /* if ret == 0 there are items for this type,
2919                  * create a range to tell us the last key of this type.
2920                  * otherwise, there are no items in this directory after
2921                  * *min_offset, and we create a range to indicate that.
2922                  */
2923                 if (ret == 0) {
2924                         struct btrfs_key tmp;
2925                         btrfs_item_key_to_cpu(path->nodes[0], &tmp,
2926                                               path->slots[0]);
2927                         if (key_type == tmp.type)
2928                                 first_offset = max(min_offset, tmp.offset) + 1;
2929                 }
2930                 goto done;
2931         }
2932
2933         /* go backward to find any previous key */
2934         ret = btrfs_previous_item(root, path, ino, key_type);
2935         if (ret == 0) {
2936                 struct btrfs_key tmp;
2937                 btrfs_item_key_to_cpu(path->nodes[0], &tmp, path->slots[0]);
2938                 if (key_type == tmp.type) {
2939                         first_offset = tmp.offset;
2940                         ret = overwrite_item(trans, log, dst_path,
2941                                              path->nodes[0], path->slots[0],
2942                                              &tmp);
2943                         if (ret) {
2944                                 err = ret;
2945                                 goto done;
2946                         }
2947                 }
2948         }
2949         btrfs_release_path(path);
2950
2951         /* find the first key from this transaction again */
2952         ret = btrfs_search_slot(NULL, root, &min_key, path, 0, 0);
2953         if (WARN_ON(ret != 0))
2954                 goto done;
2955
2956         /*
2957          * we have a block from this transaction, log every item in it
2958          * from our directory
2959          */
2960         while (1) {
2961                 struct btrfs_key tmp;
2962                 src = path->nodes[0];
2963                 nritems = btrfs_header_nritems(src);
2964                 for (i = path->slots[0]; i < nritems; i++) {
2965                         btrfs_item_key_to_cpu(src, &min_key, i);
2966
2967                         if (min_key.objectid != ino || min_key.type != key_type)
2968                                 goto done;
2969                         ret = overwrite_item(trans, log, dst_path, src, i,
2970                                              &min_key);
2971                         if (ret) {
2972                                 err = ret;
2973                                 goto done;
2974                         }
2975                 }
2976                 path->slots[0] = nritems;
2977
2978                 /*
2979                  * look ahead to the next item and see if it is also
2980                  * from this directory and from this transaction
2981                  */
2982                 ret = btrfs_next_leaf(root, path);
2983                 if (ret == 1) {
2984                         last_offset = (u64)-1;
2985                         goto done;
2986                 }
2987                 btrfs_item_key_to_cpu(path->nodes[0], &tmp, path->slots[0]);
2988                 if (tmp.objectid != ino || tmp.type != key_type) {
2989                         last_offset = (u64)-1;
2990                         goto done;
2991                 }
2992                 if (btrfs_header_generation(path->nodes[0]) != trans->transid) {
2993                         ret = overwrite_item(trans, log, dst_path,
2994                                              path->nodes[0], path->slots[0],
2995                                              &tmp);
2996                         if (ret)
2997                                 err = ret;
2998                         else
2999                                 last_offset = tmp.offset;
3000                         goto done;
3001                 }
3002         }
3003 done:
3004         btrfs_release_path(path);
3005         btrfs_release_path(dst_path);
3006
3007         if (err == 0) {
3008                 *last_offset_ret = last_offset;
3009                 /*
3010                  * insert the log range keys to indicate where the log
3011                  * is valid
3012                  */
3013                 ret = insert_dir_log_key(trans, log, path, key_type,
3014                                          ino, first_offset, last_offset);
3015                 if (ret)
3016                         err = ret;
3017         }
3018         return err;
3019 }
3020
3021 /*
3022  * logging directories is very similar to logging inodes, We find all the items
3023  * from the current transaction and write them to the log.
3024  *
3025  * The recovery code scans the directory in the subvolume, and if it finds a
3026  * key in the range logged that is not present in the log tree, then it means
3027  * that dir entry was unlinked during the transaction.
3028  *
3029  * In order for that scan to work, we must include one key smaller than
3030  * the smallest logged by this transaction and one key larger than the largest
3031  * key logged by this transaction.
3032  */
3033 static noinline int log_directory_changes(struct btrfs_trans_handle *trans,
3034                           struct btrfs_root *root, struct inode *inode,
3035                           struct btrfs_path *path,
3036                           struct btrfs_path *dst_path)
3037 {
3038         u64 min_key;
3039         u64 max_key;
3040         int ret;
3041         int key_type = BTRFS_DIR_ITEM_KEY;
3042
3043 again:
3044         min_key = 0;
3045         max_key = 0;
3046         while (1) {
3047                 ret = log_dir_items(trans, root, inode, path,
3048                                     dst_path, key_type, min_key,
3049                                     &max_key);
3050                 if (ret)
3051                         return ret;
3052                 if (max_key == (u64)-1)
3053                         break;
3054                 min_key = max_key + 1;
3055         }
3056
3057         if (key_type == BTRFS_DIR_ITEM_KEY) {
3058                 key_type = BTRFS_DIR_INDEX_KEY;
3059                 goto again;
3060         }
3061         return 0;
3062 }
3063
3064 /*
3065  * a helper function to drop items from the log before we relog an
3066  * inode.  max_key_type indicates the highest item type to remove.
3067  * This cannot be run for file data extents because it does not
3068  * free the extents they point to.
3069  */
3070 static int drop_objectid_items(struct btrfs_trans_handle *trans,
3071                                   struct btrfs_root *log,
3072                                   struct btrfs_path *path,
3073                                   u64 objectid, int max_key_type)
3074 {
3075         int ret;
3076         struct btrfs_key key;
3077         struct btrfs_key found_key;
3078         int start_slot;
3079
3080         key.objectid = objectid;
3081         key.type = max_key_type;
3082         key.offset = (u64)-1;
3083
3084         while (1) {
3085                 ret = btrfs_search_slot(trans, log, &key, path, -1, 1);
3086                 BUG_ON(ret == 0); /* Logic error */
3087                 if (ret < 0)
3088                         break;
3089
3090                 if (path->slots[0] == 0)
3091                         break;
3092
3093                 path->slots[0]--;
3094                 btrfs_item_key_to_cpu(path->nodes[0], &found_key,
3095                                       path->slots[0]);
3096
3097                 if (found_key.objectid != objectid)
3098                         break;
3099
3100                 found_key.offset = 0;
3101                 found_key.type = 0;
3102                 ret = btrfs_bin_search(path->nodes[0], &found_key, 0,
3103                                        &start_slot);
3104
3105                 ret = btrfs_del_items(trans, log, path, start_slot,
3106                                       path->slots[0] - start_slot + 1);
3107                 /*
3108                  * If start slot isn't 0 then we don't need to re-search, we've
3109                  * found the last guy with the objectid in this tree.
3110                  */
3111                 if (ret || start_slot != 0)
3112                         break;
3113                 btrfs_release_path(path);
3114         }
3115         btrfs_release_path(path);
3116         if (ret > 0)
3117                 ret = 0;
3118         return ret;
3119 }
3120
3121 static void fill_inode_item(struct btrfs_trans_handle *trans,
3122                             struct extent_buffer *leaf,
3123                             struct btrfs_inode_item *item,
3124                             struct inode *inode, int log_inode_only)
3125 {
3126         struct btrfs_map_token token;
3127
3128         btrfs_init_map_token(&token);
3129
3130         if (log_inode_only) {
3131                 /* set the generation to zero so the recover code
3132                  * can tell the difference between an logging
3133                  * just to say 'this inode exists' and a logging
3134                  * to say 'update this inode with these values'
3135                  */
3136                 btrfs_set_token_inode_generation(leaf, item, 0, &token);
3137                 btrfs_set_token_inode_size(leaf, item, 0, &token);
3138         } else {
3139                 btrfs_set_token_inode_generation(leaf, item,
3140                                                  BTRFS_I(inode)->generation,
3141                                                  &token);
3142                 btrfs_set_token_inode_size(leaf, item, inode->i_size, &token);
3143         }
3144
3145         btrfs_set_token_inode_uid(leaf, item, i_uid_read(inode), &token);
3146         btrfs_set_token_inode_gid(leaf, item, i_gid_read(inode), &token);
3147         btrfs_set_token_inode_mode(leaf, item, inode->i_mode, &token);
3148         btrfs_set_token_inode_nlink(leaf, item, inode->i_nlink, &token);
3149
3150         btrfs_set_token_timespec_sec(leaf, btrfs_inode_atime(item),
3151                                      inode->i_atime.tv_sec, &token);
3152         btrfs_set_token_timespec_nsec(leaf, btrfs_inode_atime(item),
3153                                       inode->i_atime.tv_nsec, &token);
3154
3155         btrfs_set_token_timespec_sec(leaf, btrfs_inode_mtime(item),
3156                                      inode->i_mtime.tv_sec, &token);
3157         btrfs_set_token_timespec_nsec(leaf, btrfs_inode_mtime(item),
3158                                       inode->i_mtime.tv_nsec, &token);
3159
3160         btrfs_set_token_timespec_sec(leaf, btrfs_inode_ctime(item),
3161                                      inode->i_ctime.tv_sec, &token);
3162         btrfs_set_token_timespec_nsec(leaf, btrfs_inode_ctime(item),
3163                                       inode->i_ctime.tv_nsec, &token);
3164
3165         btrfs_set_token_inode_nbytes(leaf, item, inode_get_bytes(inode),
3166                                      &token);
3167
3168         btrfs_set_token_inode_sequence(leaf, item, inode->i_version, &token);
3169         btrfs_set_token_inode_transid(leaf, item, trans->transid, &token);
3170         btrfs_set_token_inode_rdev(leaf, item, inode->i_rdev, &token);
3171         btrfs_set_token_inode_flags(leaf, item, BTRFS_I(inode)->flags, &token);
3172         btrfs_set_token_inode_block_group(leaf, item, 0, &token);
3173 }
3174
3175 static int log_inode_item(struct btrfs_trans_handle *trans,
3176                           struct btrfs_root *log, struct btrfs_path *path,
3177                           struct inode *inode)
3178 {
3179         struct btrfs_inode_item *inode_item;
3180         int ret;
3181
3182         ret = btrfs_insert_empty_item(trans, log, path,
3183                                       &BTRFS_I(inode)->location,
3184                                       sizeof(*inode_item));
3185         if (ret && ret != -EEXIST)
3186                 return ret;
3187         inode_item = btrfs_item_ptr(path->nodes[0], path->slots[0],
3188                                     struct btrfs_inode_item);
3189         fill_inode_item(trans, path->nodes[0], inode_item, inode, 0);
3190         btrfs_release_path(path);
3191         return 0;
3192 }
3193
3194 static noinline int copy_items(struct btrfs_trans_handle *trans,
3195                                struct inode *inode,
3196                                struct btrfs_path *dst_path,
3197                                struct extent_buffer *src,
3198                                int start_slot, int nr, int inode_only)
3199 {
3200         unsigned long src_offset;
3201         unsigned long dst_offset;
3202         struct btrfs_root *log = BTRFS_I(inode)->root->log_root;
3203         struct btrfs_file_extent_item *extent;
3204         struct btrfs_inode_item *inode_item;
3205         int ret;
3206         struct btrfs_key *ins_keys;
3207         u32 *ins_sizes;
3208         char *ins_data;
3209         int i;
3210         struct list_head ordered_sums;
3211         int skip_csum = BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM;
3212
3213         INIT_LIST_HEAD(&ordered_sums);
3214
3215         ins_data = kmalloc(nr * sizeof(struct btrfs_key) +
3216                            nr * sizeof(u32), GFP_NOFS);
3217         if (!ins_data)
3218                 return -ENOMEM;
3219
3220         ins_sizes = (u32 *)ins_data;
3221         ins_keys = (struct btrfs_key *)(ins_data + nr * sizeof(u32));
3222
3223         for (i = 0; i < nr; i++) {
3224                 ins_sizes[i] = btrfs_item_size_nr(src, i + start_slot);
3225                 btrfs_item_key_to_cpu(src, ins_keys + i, i + start_slot);
3226         }
3227         ret = btrfs_insert_empty_items(trans, log, dst_path,
3228                                        ins_keys, ins_sizes, nr);
3229         if (ret) {
3230                 kfree(ins_data);
3231                 return ret;
3232         }
3233
3234         for (i = 0; i < nr; i++, dst_path->slots[0]++) {
3235                 dst_offset = btrfs_item_ptr_offset(dst_path->nodes[0],
3236                                                    dst_path->slots[0]);
3237
3238                 src_offset = btrfs_item_ptr_offset(src, start_slot + i);
3239
3240                 if (ins_keys[i].type == BTRFS_INODE_ITEM_KEY) {
3241                         inode_item = btrfs_item_ptr(dst_path->nodes[0],
3242                                                     dst_path->slots[0],
3243                                                     struct btrfs_inode_item);
3244                         fill_inode_item(trans, dst_path->nodes[0], inode_item,
3245                                         inode, inode_only == LOG_INODE_EXISTS);
3246                 } else {
3247                         copy_extent_buffer(dst_path->nodes[0], src, dst_offset,
3248                                            src_offset, ins_sizes[i]);
3249                 }
3250
3251                 /* take a reference on file data extents so that truncates
3252                  * or deletes of this inode don't have to relog the inode
3253                  * again
3254                  */
3255                 if (btrfs_key_type(ins_keys + i) == BTRFS_EXTENT_DATA_KEY &&
3256                     !skip_csum) {
3257                         int found_type;
3258                         extent = btrfs_item_ptr(src, start_slot + i,
3259                                                 struct btrfs_file_extent_item);
3260
3261                         if (btrfs_file_extent_generation(src, extent) < trans->transid)
3262                                 continue;
3263
3264                         found_type = btrfs_file_extent_type(src, extent);
3265                         if (found_type == BTRFS_FILE_EXTENT_REG) {
3266                                 u64 ds, dl, cs, cl;
3267                                 ds = btrfs_file_extent_disk_bytenr(src,
3268                                                                 extent);
3269                                 /* ds == 0 is a hole */
3270                                 if (ds == 0)
3271                                         continue;
3272
3273                                 dl = btrfs_file_extent_disk_num_bytes(src,
3274                                                                 extent);
3275                                 cs = btrfs_file_extent_offset(src, extent);
3276                                 cl = btrfs_file_extent_num_bytes(src,
3277                                                                 extent);
3278                                 if (btrfs_file_extent_compression(src,
3279                                                                   extent)) {
3280                                         cs = 0;
3281                                         cl = dl;
3282                                 }
3283
3284                                 ret = btrfs_lookup_csums_range(
3285                                                 log->fs_info->csum_root,
3286                                                 ds + cs, ds + cs + cl - 1,
3287                                                 &ordered_sums, 0);
3288                                 if (ret) {
3289                                         btrfs_release_path(dst_path);
3290                                         kfree(ins_data);
3291                                         return ret;
3292                                 }
3293                         }
3294                 }
3295         }
3296
3297         btrfs_mark_buffer_dirty(dst_path->nodes[0]);
3298         btrfs_release_path(dst_path);
3299         kfree(ins_data);
3300
3301         /*
3302          * we have to do this after the loop above to avoid changing the
3303          * log tree while trying to change the log tree.
3304          */
3305         ret = 0;
3306         while (!list_empty(&ordered_sums)) {
3307                 struct btrfs_ordered_sum *sums = list_entry(ordered_sums.next,
3308                                                    struct btrfs_ordered_sum,
3309                                                    list);
3310                 if (!ret)
3311                         ret = btrfs_csum_file_blocks(trans, log, sums);
3312                 list_del(&sums->list);
3313                 kfree(sums);
3314         }
3315         return ret;
3316 }
3317
3318 static int extent_cmp(void *priv, struct list_head *a, struct list_head *b)
3319 {
3320         struct extent_map *em1, *em2;
3321
3322         em1 = list_entry(a, struct extent_map, list);
3323         em2 = list_entry(b, struct extent_map, list);
3324
3325         if (em1->start < em2->start)
3326                 return -1;
3327         else if (em1->start > em2->start)
3328                 return 1;
3329         return 0;
3330 }
3331
3332 static int log_one_extent(struct btrfs_trans_handle *trans,
3333                           struct inode *inode, struct btrfs_root *root,
3334                           struct extent_map *em, struct btrfs_path *path)
3335 {
3336         struct btrfs_root *log = root->log_root;
3337         struct btrfs_file_extent_item *fi;
3338         struct extent_buffer *leaf;
3339         struct btrfs_ordered_extent *ordered;
3340         struct list_head ordered_sums;
3341         struct btrfs_map_token token;
3342         struct btrfs_key key;
3343         u64 mod_start = em->mod_start;
3344         u64 mod_len = em->mod_len;
3345         u64 csum_offset;
3346         u64 csum_len;
3347         u64 extent_offset = em->start - em->orig_start;
3348         u64 block_len;
3349         int ret;
3350         int index = log->log_transid % 2;
3351         bool skip_csum = BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM;
3352
3353         ret = __btrfs_drop_extents(trans, log, inode, path, em->start,
3354                                    em->start + em->len, NULL, 0);
3355         if (ret)
3356                 return ret;
3357
3358         INIT_LIST_HEAD(&ordered_sums);
3359         btrfs_init_map_token(&token);
3360         key.objectid = btrfs_ino(inode);
3361         key.type = BTRFS_EXTENT_DATA_KEY;
3362         key.offset = em->start;
3363
3364         ret = btrfs_insert_empty_item(trans, log, path, &key, sizeof(*fi));
3365         if (ret)
3366                 return ret;
3367         leaf = path->nodes[0];
3368         fi = btrfs_item_ptr(leaf, path->slots[0],
3369                             struct btrfs_file_extent_item);
3370
3371         btrfs_set_token_file_extent_generation(leaf, fi, em->generation,
3372                                                &token);
3373         if (test_bit(EXTENT_FLAG_PREALLOC, &em->flags)) {
3374                 skip_csum = true;
3375                 btrfs_set_token_file_extent_type(leaf, fi,
3376                                                  BTRFS_FILE_EXTENT_PREALLOC,
3377                                                  &token);
3378         } else {
3379                 btrfs_set_token_file_extent_type(leaf, fi,
3380                                                  BTRFS_FILE_EXTENT_REG,
3381                                                  &token);
3382                 if (em->block_start == EXTENT_MAP_HOLE)
3383                         skip_csum = true;
3384         }
3385
3386         block_len = max(em->block_len, em->orig_block_len);
3387         if (em->compress_type != BTRFS_COMPRESS_NONE) {
3388                 btrfs_set_token_file_extent_disk_bytenr(leaf, fi,
3389                                                         em->block_start,
3390                                                         &token);
3391                 btrfs_set_token_file_extent_disk_num_bytes(leaf, fi, block_len,
3392                                                            &token);
3393         } else if (em->block_start < EXTENT_MAP_LAST_BYTE) {
3394                 btrfs_set_token_file_extent_disk_bytenr(leaf, fi,
3395                                                         em->block_start -
3396                                                         extent_offset, &token);
3397                 btrfs_set_token_file_extent_disk_num_bytes(leaf, fi, block_len,
3398                                                            &token);
3399         } else {
3400                 btrfs_set_token_file_extent_disk_bytenr(leaf, fi, 0, &token);
3401                 btrfs_set_token_file_extent_disk_num_bytes(leaf, fi, 0,
3402                                                            &token);
3403         }
3404
3405         btrfs_set_token_file_extent_offset(leaf, fi,
3406                                            em->start - em->orig_start,
3407                                            &token);
3408         btrfs_set_token_file_extent_num_bytes(leaf, fi, em->len, &token);
3409         btrfs_set_token_file_extent_ram_bytes(leaf, fi, em->ram_bytes, &token);
3410         btrfs_set_token_file_extent_compression(leaf, fi, em->compress_type,
3411                                                 &token);
3412         btrfs_set_token_file_extent_encryption(leaf, fi, 0, &token);
3413         btrfs_set_token_file_extent_other_encoding(leaf, fi, 0, &token);
3414         btrfs_mark_buffer_dirty(leaf);
3415
3416         btrfs_release_path(path);
3417         if (ret) {
3418                 return ret;
3419         }
3420
3421         if (skip_csum)
3422                 return 0;
3423
3424         /*
3425          * First check and see if our csums are on our outstanding ordered
3426          * extents.
3427          */
3428 again:
3429         spin_lock_irq(&log->log_extents_lock[index]);
3430         list_for_each_entry(ordered, &log->logged_list[index], log_list) {
3431                 struct btrfs_ordered_sum *sum;
3432
3433                 if (!mod_len)
3434                         break;
3435
3436                 if (ordered->inode != inode)
3437                         continue;
3438
3439                 if (ordered->file_offset + ordered->len <= mod_start ||
3440                     mod_start + mod_len <= ordered->file_offset)
3441                         continue;
3442
3443                 /*
3444                  * We are going to copy all the csums on this ordered extent, so
3445                  * go ahead and adjust mod_start and mod_len in case this
3446                  * ordered extent has already been logged.
3447                  */
3448                 if (ordered->file_offset > mod_start) {
3449                         if (ordered->file_offset + ordered->len >=
3450                             mod_start + mod_len)
3451                                 mod_len = ordered->file_offset - mod_start;
3452                         /*
3453                          * If we have this case
3454                          *
3455                          * |--------- logged extent ---------|
3456                          *       |----- ordered extent ----|
3457                          *
3458                          * Just don't mess with mod_start and mod_len, we'll
3459                          * just end up logging more csums than we need and it
3460                          * will be ok.
3461                          */
3462                 } else {
3463                         if (ordered->file_offset + ordered->len <
3464                             mod_start + mod_len) {
3465                                 mod_len = (mod_start + mod_len) -
3466                                         (ordered->file_offset + ordered->len);
3467                                 mod_start = ordered->file_offset +
3468                                         ordered->len;
3469                         } else {
3470                                 mod_len = 0;
3471                         }
3472                 }
3473
3474                 /*
3475                  * To keep us from looping for the above case of an ordered
3476                  * extent that falls inside of the logged extent.
3477                  */
3478                 if (test_and_set_bit(BTRFS_ORDERED_LOGGED_CSUM,
3479                                      &ordered->flags))
3480                         continue;
3481                 atomic_inc(&ordered->refs);
3482                 spin_unlock_irq(&log->log_extents_lock[index]);
3483                 /*
3484                  * we've dropped the lock, we must either break or
3485                  * start over after this.
3486                  */
3487
3488                 wait_event(ordered->wait, ordered->csum_bytes_left == 0);
3489
3490                 list_for_each_entry(sum, &ordered->list, list) {
3491                         ret = btrfs_csum_file_blocks(trans, log, sum);
3492                         if (ret) {
3493                                 btrfs_put_ordered_extent(ordered);
3494                                 goto unlocked;
3495                         }
3496                 }
3497                 btrfs_put_ordered_extent(ordered);
3498                 goto again;
3499
3500         }
3501         spin_unlock_irq(&log->log_extents_lock[index]);
3502 unlocked:
3503
3504         if (!mod_len || ret)
3505                 return ret;
3506
3507         if (em->compress_type) {
3508                 csum_offset = 0;
3509                 csum_len = block_len;
3510         } else {
3511                 csum_offset = mod_start - em->start;
3512                 csum_len = mod_len;
3513         }
3514
3515         /* block start is already adjusted for the file extent offset. */
3516         ret = btrfs_lookup_csums_range(log->fs_info->csum_root,
3517                                        em->block_start + csum_offset,
3518                                        em->block_start + csum_offset +
3519                                        csum_len - 1, &ordered_sums, 0);
3520         if (ret)
3521                 return ret;
3522
3523         while (!list_empty(&ordered_sums)) {
3524                 struct btrfs_ordered_sum *sums = list_entry(ordered_sums.next,
3525                                                    struct btrfs_ordered_sum,
3526                                                    list);
3527                 if (!ret)
3528                         ret = btrfs_csum_file_blocks(trans, log, sums);
3529                 list_del(&sums->list);
3530                 kfree(sums);
3531         }
3532
3533         return ret;
3534 }
3535
3536 static int btrfs_log_changed_extents(struct btrfs_trans_handle *trans,
3537                                      struct btrfs_root *root,
3538                                      struct inode *inode,
3539                                      struct btrfs_path *path)
3540 {
3541         struct extent_map *em, *n;
3542         struct list_head extents;
3543         struct extent_map_tree *tree = &BTRFS_I(inode)->extent_tree;
3544         u64 test_gen;
3545         int ret = 0;
3546         int num = 0;
3547
3548         INIT_LIST_HEAD(&extents);
3549
3550         write_lock(&tree->lock);
3551         test_gen = root->fs_info->last_trans_committed;
3552
3553         list_for_each_entry_safe(em, n, &tree->modified_extents, list) {
3554                 list_del_init(&em->list);
3555
3556                 /*
3557                  * Just an arbitrary number, this can be really CPU intensive
3558                  * once we start getting a lot of extents, and really once we
3559                  * have a bunch of extents we just want to commit since it will
3560                  * be faster.
3561                  */
3562                 if (++num > 32768) {
3563                         list_del_init(&tree->modified_extents);
3564                         ret = -EFBIG;
3565                         goto process;
3566                 }
3567
3568                 if (em->generation <= test_gen)
3569                         continue;
3570                 /* Need a ref to keep it from getting evicted from cache */
3571                 atomic_inc(&em->refs);
3572                 set_bit(EXTENT_FLAG_LOGGING, &em->flags);
3573                 list_add_tail(&em->list, &extents);
3574                 num++;
3575         }
3576
3577         list_sort(NULL, &extents, extent_cmp);
3578
3579 process:
3580         while (!list_empty(&extents)) {
3581                 em = list_entry(extents.next, struct extent_map, list);
3582
3583                 list_del_init(&em->list);
3584
3585                 /*
3586                  * If we had an error we just need to delete everybody from our
3587                  * private list.
3588                  */
3589                 if (ret) {
3590                         clear_em_logging(tree, em);
3591                         free_extent_map(em);
3592                         continue;
3593                 }
3594
3595                 write_unlock(&tree->lock);
3596
3597                 ret = log_one_extent(trans, inode, root, em, path);
3598                 write_lock(&tree->lock);
3599                 clear_em_logging(tree, em);
3600                 free_extent_map(em);
3601         }
3602         WARN_ON(!list_empty(&extents));
3603         write_unlock(&tree->lock);
3604
3605         btrfs_release_path(path);
3606         return ret;
3607 }
3608
3609 /* log a single inode in the tree log.
3610  * At least one parent directory for this inode must exist in the tree
3611  * or be logged already.
3612  *
3613  * Any items from this inode changed by the current transaction are copied
3614  * to the log tree.  An extra reference is taken on any extents in this
3615  * file, allowing us to avoid a whole pile of corner cases around logging
3616  * blocks that have been removed from the tree.
3617  *
3618  * See LOG_INODE_ALL and related defines for a description of what inode_only
3619  * does.
3620  *
3621  * This handles both files and directories.
3622  */
3623 static int btrfs_log_inode(struct btrfs_trans_handle *trans,
3624                              struct btrfs_root *root, struct inode *inode,
3625                              int inode_only)
3626 {
3627         struct btrfs_path *path;
3628         struct btrfs_path *dst_path;
3629         struct btrfs_key min_key;
3630         struct btrfs_key max_key;
3631         struct btrfs_root *log = root->log_root;
3632         struct extent_buffer *src = NULL;
3633         int err = 0;
3634         int ret;
3635         int nritems;
3636         int ins_start_slot = 0;
3637         int ins_nr;
3638         bool fast_search = false;
3639         u64 ino = btrfs_ino(inode);
3640
3641         path = btrfs_alloc_path();
3642         if (!path)
3643                 return -ENOMEM;
3644         dst_path = btrfs_alloc_path();
3645         if (!dst_path) {
3646                 btrfs_free_path(path);
3647                 return -ENOMEM;
3648         }
3649
3650         min_key.objectid = ino;
3651         min_key.type = BTRFS_INODE_ITEM_KEY;
3652         min_key.offset = 0;
3653
3654         max_key.objectid = ino;
3655
3656
3657         /* today the code can only do partial logging of directories */
3658         if (S_ISDIR(inode->i_mode) ||
3659             (!test_bit(BTRFS_INODE_NEEDS_FULL_SYNC,
3660                        &BTRFS_I(inode)->runtime_flags) &&
3661              inode_only == LOG_INODE_EXISTS))
3662                 max_key.type = BTRFS_XATTR_ITEM_KEY;
3663         else
3664                 max_key.type = (u8)-1;
3665         max_key.offset = (u64)-1;
3666
3667         /* Only run delayed items if we are a dir or a new file */
3668         if (S_ISDIR(inode->i_mode) ||
3669             BTRFS_I(inode)->generation > root->fs_info->last_trans_committed) {
3670                 ret = btrfs_commit_inode_delayed_items(trans, inode);
3671                 if (ret) {
3672                         btrfs_free_path(path);
3673                         btrfs_free_path(dst_path);
3674                         return ret;
3675                 }
3676         }
3677
3678         mutex_lock(&BTRFS_I(inode)->log_mutex);
3679
3680         btrfs_get_logged_extents(log, inode);
3681
3682         /*
3683          * a brute force approach to making sure we get the most uptodate
3684          * copies of everything.
3685          */
3686         if (S_ISDIR(inode->i_mode)) {
3687                 int max_key_type = BTRFS_DIR_LOG_INDEX_KEY;
3688
3689                 if (inode_only == LOG_INODE_EXISTS)
3690                         max_key_type = BTRFS_XATTR_ITEM_KEY;
3691                 ret = drop_objectid_items(trans, log, path, ino, max_key_type);
3692         } else {
3693                 if (test_and_clear_bit(BTRFS_INODE_NEEDS_FULL_SYNC,
3694                                        &BTRFS_I(inode)->runtime_flags)) {
3695                         clear_bit(BTRFS_INODE_COPY_EVERYTHING,
3696                                   &BTRFS_I(inode)->runtime_flags);
3697                         ret = btrfs_truncate_inode_items(trans, log,
3698                                                          inode, 0, 0);
3699                 } else if (test_and_clear_bit(BTRFS_INODE_COPY_EVERYTHING,
3700                                               &BTRFS_I(inode)->runtime_flags) ||
3701                            inode_only == LOG_INODE_EXISTS) {
3702                         if (inode_only == LOG_INODE_ALL)
3703                                 fast_search = true;
3704                         max_key.type = BTRFS_XATTR_ITEM_KEY;
3705                         ret = drop_objectid_items(trans, log, path, ino,
3706                                                   max_key.type);
3707                 } else {
3708                         if (inode_only == LOG_INODE_ALL)
3709                                 fast_search = true;
3710                         ret = log_inode_item(trans, log, dst_path, inode);
3711                         if (ret) {
3712                                 err = ret;
3713                                 goto out_unlock;
3714                         }
3715                         goto log_extents;
3716                 }
3717
3718         }
3719         if (ret) {
3720                 err = ret;
3721                 goto out_unlock;
3722         }
3723         path->keep_locks = 1;
3724
3725         while (1) {
3726                 ins_nr = 0;
3727                 ret = btrfs_search_forward(root, &min_key,
3728                                            path, trans->transid);
3729                 if (ret != 0)
3730                         break;
3731 again:
3732                 /* note, ins_nr might be > 0 here, cleanup outside the loop */
3733                 if (min_key.objectid != ino)
3734                         break;
3735                 if (min_key.type > max_key.type)
3736                         break;
3737
3738                 src = path->nodes[0];
3739                 if (ins_nr && ins_start_slot + ins_nr == path->slots[0]) {
3740                         ins_nr++;
3741                         goto next_slot;
3742                 } else if (!ins_nr) {
3743                         ins_start_slot = path->slots[0];
3744                         ins_nr = 1;
3745                         goto next_slot;
3746                 }
3747
3748                 ret = copy_items(trans, inode, dst_path, src, ins_start_slot,
3749                                  ins_nr, inode_only);
3750                 if (ret) {
3751                         err = ret;
3752                         goto out_unlock;
3753                 }
3754                 ins_nr = 1;
3755                 ins_start_slot = path->slots[0];
3756 next_slot:
3757
3758                 nritems = btrfs_header_nritems(path->nodes[0]);
3759                 path->slots[0]++;
3760                 if (path->slots[0] < nritems) {
3761                         btrfs_item_key_to_cpu(path->nodes[0], &min_key,
3762                                               path->slots[0]);
3763                         goto again;
3764                 }
3765                 if (ins_nr) {
3766                         ret = copy_items(trans, inode, dst_path, src,
3767                                          ins_start_slot,
3768                                          ins_nr, inode_only);
3769                         if (ret) {
3770                                 err = ret;
3771                                 goto out_unlock;
3772                         }
3773                         ins_nr = 0;
3774                 }
3775                 btrfs_release_path(path);
3776
3777                 if (min_key.offset < (u64)-1) {
3778                         min_key.offset++;
3779                 } else if (min_key.type < max_key.type) {
3780                         min_key.type++;
3781                         min_key.offset = 0;
3782                 } else {
3783                         break;
3784                 }
3785         }
3786         if (ins_nr) {
3787                 ret = copy_items(trans, inode, dst_path, src, ins_start_slot,
3788                                  ins_nr, inode_only);
3789                 if (ret) {
3790                         err = ret;
3791                         goto out_unlock;
3792                 }
3793                 ins_nr = 0;
3794         }
3795
3796 log_extents:
3797         btrfs_release_path(path);
3798         btrfs_release_path(dst_path);
3799         if (fast_search) {
3800                 ret = btrfs_log_changed_extents(trans, root, inode, dst_path);
3801                 if (ret) {
3802                         err = ret;
3803                         goto out_unlock;
3804                 }
3805         } else {
3806                 struct extent_map_tree *tree = &BTRFS_I(inode)->extent_tree;
3807                 struct extent_map *em, *n;
3808
3809                 write_lock(&tree->lock);
3810                 list_for_each_entry_safe(em, n, &tree->modified_extents, list)
3811                         list_del_init(&em->list);
3812                 write_unlock(&tree->lock);
3813         }
3814
3815         if (inode_only == LOG_INODE_ALL && S_ISDIR(inode->i_mode)) {
3816                 ret = log_directory_changes(trans, root, inode, path, dst_path);
3817                 if (ret) {
3818                         err = ret;
3819                         goto out_unlock;
3820                 }
3821         }
3822         BTRFS_I(inode)->logged_trans = trans->transid;
3823         BTRFS_I(inode)->last_log_commit = BTRFS_I(inode)->last_sub_trans;
3824 out_unlock:
3825         if (err)
3826                 btrfs_free_logged_extents(log, log->log_transid);
3827         mutex_unlock(&BTRFS_I(inode)->log_mutex);
3828
3829         btrfs_free_path(path);
3830         btrfs_free_path(dst_path);
3831         return err;
3832 }
3833
3834 /*
3835  * follow the dentry parent pointers up the chain and see if any
3836  * of the directories in it require a full commit before they can
3837  * be logged.  Returns zero if nothing special needs to be done or 1 if
3838  * a full commit is required.
3839  */
3840 static noinline int check_parent_dirs_for_sync(struct btrfs_trans_handle *trans,
3841                                                struct inode *inode,
3842                                                struct dentry *parent,
3843                                                struct super_block *sb,
3844                                                u64 last_committed)
3845 {
3846         int ret = 0;
3847         struct btrfs_root *root;
3848         struct dentry *old_parent = NULL;
3849         struct inode *orig_inode = inode;
3850
3851         /*
3852          * for regular files, if its inode is already on disk, we don't
3853          * have to worry about the parents at all.  This is because
3854          * we can use the last_unlink_trans field to record renames
3855          * and other fun in this file.
3856          */
3857         if (S_ISREG(inode->i_mode) &&
3858             BTRFS_I(inode)->generation <= last_committed &&
3859             BTRFS_I(inode)->last_unlink_trans <= last_committed)
3860                         goto out;
3861
3862         if (!S_ISDIR(inode->i_mode)) {
3863                 if (!parent || !parent->d_inode || sb != parent->d_inode->i_sb)
3864                         goto out;
3865                 inode = parent->d_inode;
3866         }
3867
3868         while (1) {
3869                 /*
3870                  * If we are logging a directory then we start with our inode,
3871                  * not our parents inode, so we need to skipp setting the
3872                  * logged_trans so that further down in the log code we don't
3873                  * think this inode has already been logged.
3874                  */
3875                 if (inode != orig_inode)
3876                         BTRFS_I(inode)->logged_trans = trans->transid;
3877                 smp_mb();
3878
3879                 if (BTRFS_I(inode)->last_unlink_trans > last_committed) {
3880                         root = BTRFS_I(inode)->root;
3881
3882                         /*
3883                          * make sure any commits to the log are forced
3884                          * to be full commits
3885                          */
3886                         root->fs_info->last_trans_log_full_commit =
3887                                 trans->transid;
3888                         ret = 1;
3889                         break;
3890                 }
3891
3892                 if (!parent || !parent->d_inode || sb != parent->d_inode->i_sb)
3893                         break;
3894
3895                 if (IS_ROOT(parent))
3896                         break;
3897
3898                 parent = dget_parent(parent);
3899                 dput(old_parent);
3900                 old_parent = parent;
3901                 inode = parent->d_inode;
3902
3903         }
3904         dput(old_parent);
3905 out:
3906         return ret;
3907 }
3908
3909 /*
3910  * helper function around btrfs_log_inode to make sure newly created
3911  * parent directories also end up in the log.  A minimal inode and backref
3912  * only logging is done of any parent directories that are older than
3913  * the last committed transaction
3914  */
3915 static int btrfs_log_inode_parent(struct btrfs_trans_handle *trans,
3916                                   struct btrfs_root *root, struct inode *inode,
3917                                   struct dentry *parent, int exists_only)
3918 {
3919         int inode_only = exists_only ? LOG_INODE_EXISTS : LOG_INODE_ALL;
3920         struct super_block *sb;
3921         struct dentry *old_parent = NULL;
3922         int ret = 0;
3923         u64 last_committed = root->fs_info->last_trans_committed;
3924
3925         sb = inode->i_sb;
3926
3927         if (btrfs_test_opt(root, NOTREELOG)) {
3928                 ret = 1;
3929                 goto end_no_trans;
3930         }
3931
3932         if (root->fs_info->last_trans_log_full_commit >
3933             root->fs_info->last_trans_committed) {
3934                 ret = 1;
3935                 goto end_no_trans;
3936         }
3937
3938         if (root != BTRFS_I(inode)->root ||
3939             btrfs_root_refs(&root->root_item) == 0) {
3940                 ret = 1;
3941                 goto end_no_trans;
3942         }
3943
3944         ret = check_parent_dirs_for_sync(trans, inode, parent,
3945                                          sb, last_committed);
3946         if (ret)
3947                 goto end_no_trans;
3948
3949         if (btrfs_inode_in_log(inode, trans->transid)) {
3950                 ret = BTRFS_NO_LOG_SYNC;
3951                 goto end_no_trans;
3952         }
3953
3954         ret = start_log_trans(trans, root);
3955         if (ret)
3956                 goto end_trans;
3957
3958         ret = btrfs_log_inode(trans, root, inode, inode_only);
3959         if (ret)
3960                 goto end_trans;
3961
3962         /*
3963          * for regular files, if its inode is already on disk, we don't
3964          * have to worry about the parents at all.  This is because
3965          * we can use the last_unlink_trans field to record renames
3966          * and other fun in this file.
3967          */
3968         if (S_ISREG(inode->i_mode) &&
3969             BTRFS_I(inode)->generation <= last_committed &&
3970             BTRFS_I(inode)->last_unlink_trans <= last_committed) {
3971                 ret = 0;
3972                 goto end_trans;
3973         }
3974
3975         inode_only = LOG_INODE_EXISTS;
3976         while (1) {
3977                 if (!parent || !parent->d_inode || sb != parent->d_inode->i_sb)
3978                         break;
3979
3980                 inode = parent->d_inode;
3981                 if (root != BTRFS_I(inode)->root)
3982                         break;
3983
3984                 if (BTRFS_I(inode)->generation >
3985                     root->fs_info->last_trans_committed) {
3986                         ret = btrfs_log_inode(trans, root, inode, inode_only);
3987                         if (ret)
3988                                 goto end_trans;
3989                 }
3990                 if (IS_ROOT(parent))
3991                         break;
3992
3993                 parent = dget_parent(parent);
3994                 dput(old_parent);
3995                 old_parent = parent;
3996         }
3997         ret = 0;
3998 end_trans:
3999         dput(old_parent);
4000         if (ret < 0) {
4001                 root->fs_info->last_trans_log_full_commit = trans->transid;
4002                 ret = 1;
4003         }
4004         btrfs_end_log_trans(root);
4005 end_no_trans:
4006         return ret;
4007 }
4008
4009 /*
4010  * it is not safe to log dentry if the chunk root has added new
4011  * chunks.  This returns 0 if the dentry was logged, and 1 otherwise.
4012  * If this returns 1, you must commit the transaction to safely get your
4013  * data on disk.
4014  */
4015 int btrfs_log_dentry_safe(struct btrfs_trans_handle *trans,
4016                           struct btrfs_root *root, struct dentry *dentry)
4017 {
4018         struct dentry *parent = dget_parent(dentry);
4019         int ret;
4020
4021         ret = btrfs_log_inode_parent(trans, root, dentry->d_inode, parent, 0);
4022         dput(parent);
4023
4024         return ret;
4025 }
4026
4027 /*
4028  * should be called during mount to recover any replay any log trees
4029  * from the FS
4030  */
4031 int btrfs_recover_log_trees(struct btrfs_root *log_root_tree)
4032 {
4033         int ret;
4034         struct btrfs_path *path;
4035         struct btrfs_trans_handle *trans;
4036         struct btrfs_key key;
4037         struct btrfs_key found_key;
4038         struct btrfs_key tmp_key;
4039         struct btrfs_root *log;
4040         struct btrfs_fs_info *fs_info = log_root_tree->fs_info;
4041         struct walk_control wc = {
4042                 .process_func = process_one_buffer,
4043                 .stage = 0,
4044         };
4045
4046         path = btrfs_alloc_path();
4047         if (!path)
4048                 return -ENOMEM;
4049
4050         fs_info->log_root_recovering = 1;
4051
4052         trans = btrfs_start_transaction(fs_info->tree_root, 0);
4053         if (IS_ERR(trans)) {
4054                 ret = PTR_ERR(trans);
4055                 goto error;
4056         }
4057
4058         wc.trans = trans;
4059         wc.pin = 1;
4060
4061         ret = walk_log_tree(trans, log_root_tree, &wc);
4062         if (ret) {
4063                 btrfs_error(fs_info, ret, "Failed to pin buffers while "
4064                             "recovering log root tree.");
4065                 goto error;
4066         }
4067
4068 again:
4069         key.objectid = BTRFS_TREE_LOG_OBJECTID;
4070         key.offset = (u64)-1;
4071         btrfs_set_key_type(&key, BTRFS_ROOT_ITEM_KEY);
4072
4073         while (1) {
4074                 ret = btrfs_search_slot(NULL, log_root_tree, &key, path, 0, 0);
4075
4076                 if (ret < 0) {
4077                         btrfs_error(fs_info, ret,
4078                                     "Couldn't find tree log root.");
4079                         goto error;
4080                 }
4081                 if (ret > 0) {
4082                         if (path->slots[0] == 0)
4083                                 break;
4084                         path->slots[0]--;
4085                 }
4086                 btrfs_item_key_to_cpu(path->nodes[0], &found_key,
4087                                       path->slots[0]);
4088                 btrfs_release_path(path);
4089                 if (found_key.objectid != BTRFS_TREE_LOG_OBJECTID)
4090                         break;
4091
4092                 log = btrfs_read_fs_root(log_root_tree, &found_key);
4093                 if (IS_ERR(log)) {
4094                         ret = PTR_ERR(log);
4095                         btrfs_error(fs_info, ret,
4096                                     "Couldn't read tree log root.");
4097                         goto error;
4098                 }
4099
4100                 tmp_key.objectid = found_key.offset;
4101                 tmp_key.type = BTRFS_ROOT_ITEM_KEY;
4102                 tmp_key.offset = (u64)-1;
4103
4104                 wc.replay_dest = btrfs_read_fs_root_no_name(fs_info, &tmp_key);
4105                 if (IS_ERR(wc.replay_dest)) {
4106                         ret = PTR_ERR(wc.replay_dest);
4107                         free_extent_buffer(log->node);
4108                         free_extent_buffer(log->commit_root);
4109                         kfree(log);
4110                         btrfs_error(fs_info, ret, "Couldn't read target root "
4111                                     "for tree log recovery.");
4112                         goto error;
4113                 }
4114
4115                 wc.replay_dest->log_root = log;
4116                 btrfs_record_root_in_trans(trans, wc.replay_dest);
4117                 ret = walk_log_tree(trans, log, &wc);
4118
4119                 if (!ret && wc.stage == LOG_WALK_REPLAY_ALL) {
4120                         ret = fixup_inode_link_counts(trans, wc.replay_dest,
4121                                                       path);
4122                 }
4123
4124                 key.offset = found_key.offset - 1;
4125                 wc.replay_dest->log_root = NULL;
4126                 free_extent_buffer(log->node);
4127                 free_extent_buffer(log->commit_root);
4128                 kfree(log);
4129
4130                 if (ret)
4131                         goto error;
4132
4133                 if (found_key.offset == 0)
4134                         break;
4135         }
4136         btrfs_release_path(path);
4137
4138         /* step one is to pin it all, step two is to replay just inodes */
4139         if (wc.pin) {
4140                 wc.pin = 0;
4141                 wc.process_func = replay_one_buffer;
4142                 wc.stage = LOG_WALK_REPLAY_INODES;
4143                 goto again;
4144         }
4145         /* step three is to replay everything */
4146         if (wc.stage < LOG_WALK_REPLAY_ALL) {
4147                 wc.stage++;
4148                 goto again;
4149         }
4150
4151         btrfs_free_path(path);
4152
4153         /* step 4: commit the transaction, which also unpins the blocks */
4154         ret = btrfs_commit_transaction(trans, fs_info->tree_root);
4155         if (ret)
4156                 return ret;
4157
4158         free_extent_buffer(log_root_tree->node);
4159         log_root_tree->log_root = NULL;
4160         fs_info->log_root_recovering = 0;
4161         kfree(log_root_tree);
4162
4163         return 0;
4164 error:
4165         if (wc.trans)
4166                 btrfs_end_transaction(wc.trans, fs_info->tree_root);
4167         btrfs_free_path(path);
4168         return ret;
4169 }
4170
4171 /*
4172  * there are some corner cases where we want to force a full
4173  * commit instead of allowing a directory to be logged.
4174  *
4175  * They revolve around files there were unlinked from the directory, and
4176  * this function updates the parent directory so that a full commit is
4177  * properly done if it is fsync'd later after the unlinks are done.
4178  */
4179 void btrfs_record_unlink_dir(struct btrfs_trans_handle *trans,
4180                              struct inode *dir, struct inode *inode,
4181                              int for_rename)
4182 {
4183         /*
4184          * when we're logging a file, if it hasn't been renamed
4185          * or unlinked, and its inode is fully committed on disk,
4186          * we don't have to worry about walking up the directory chain
4187          * to log its parents.
4188          *
4189          * So, we use the last_unlink_trans field to put this transid
4190          * into the file.  When the file is logged we check it and
4191          * don't log the parents if the file is fully on disk.
4192          */
4193         if (S_ISREG(inode->i_mode))
4194                 BTRFS_I(inode)->last_unlink_trans = trans->transid;
4195
4196         /*
4197          * if this directory was already logged any new
4198          * names for this file/dir will get recorded
4199          */
4200         smp_mb();
4201         if (BTRFS_I(dir)->logged_trans == trans->transid)
4202                 return;
4203
4204         /*
4205          * if the inode we're about to unlink was logged,
4206          * the log will be properly updated for any new names
4207          */
4208         if (BTRFS_I(inode)->logged_trans == trans->transid)
4209                 return;
4210
4211         /*
4212          * when renaming files across directories, if the directory
4213          * there we're unlinking from gets fsync'd later on, there's
4214          * no way to find the destination directory later and fsync it
4215          * properly.  So, we have to be conservative and force commits
4216          * so the new name gets discovered.
4217          */
4218         if (for_rename)
4219                 goto record;
4220
4221         /* we can safely do the unlink without any special recording */
4222         return;
4223
4224 record:
4225         BTRFS_I(dir)->last_unlink_trans = trans->transid;
4226 }
4227
4228 /*
4229  * Call this after adding a new name for a file and it will properly
4230  * update the log to reflect the new name.
4231  *
4232  * It will return zero if all goes well, and it will return 1 if a
4233  * full transaction commit is required.
4234  */
4235 int btrfs_log_new_name(struct btrfs_trans_handle *trans,
4236                         struct inode *inode, struct inode *old_dir,
4237                         struct dentry *parent)
4238 {
4239         struct btrfs_root * root = BTRFS_I(inode)->root;
4240
4241         /*
4242          * this will force the logging code to walk the dentry chain
4243          * up for the file
4244          */
4245         if (S_ISREG(inode->i_mode))
4246                 BTRFS_I(inode)->last_unlink_trans = trans->transid;
4247
4248         /*
4249          * if this inode hasn't been logged and directory we're renaming it
4250          * from hasn't been logged, we don't need to log it
4251          */
4252         if (BTRFS_I(inode)->logged_trans <=
4253             root->fs_info->last_trans_committed &&
4254             (!old_dir || BTRFS_I(old_dir)->logged_trans <=
4255                     root->fs_info->last_trans_committed))
4256                 return 0;
4257
4258         return btrfs_log_inode_parent(trans, root, inode, parent, 1);
4259 }
4260