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