fs/btrfs/btrfs_inode.h

   1 /*
   2  * Copyright (C) 2007 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 #ifndef __BTRFS_I__
  20 #define __BTRFS_I__
  21
  22 #include <linux/hash.h>
  23 #include "extent_map.h"
  24 #include "extent_io.h"
  25 #include "ordered-data.h"
  26 #include "delayed-inode.h"
  27
  28 /*
  29  * ordered_data_close is set by truncate when a file that used
  30  * to have good data has been truncated to zero.  When it is set
  31  * the btrfs file release call will add this inode to the
  32  * ordered operations list so that we make sure to flush out any
  33  * new data the application may have written before commit.
  34  */
  35 #define BTRFS_INODE_ORDERED_DATA_CLOSE          0
  36 #define BTRFS_INODE_ORPHAN_META_RESERVED        1
  37 #define BTRFS_INODE_DUMMY                       2
  38 #define BTRFS_INODE_IN_DEFRAG                   3
  39 #define BTRFS_INODE_DELALLOC_META_RESERVED      4
  40 #define BTRFS_INODE_HAS_ORPHAN_ITEM             5
  41 #define BTRFS_INODE_HAS_ASYNC_EXTENT            6
  42 #define BTRFS_INODE_NEEDS_FULL_SYNC             7
  43 #define BTRFS_INODE_COPY_EVERYTHING             8
  44 #define BTRFS_INODE_IN_DELALLOC_LIST            9
  45 #define BTRFS_INODE_READDIO_NEED_LOCK           10
  46 #define BTRFS_INODE_HAS_PROPS                   11
  47 /*
  48  * The following 3 bits are meant only for the btree inode.
  49  * When any of them is set, it means an error happened while writing an
  50  * extent buffer belonging to:
  51  * 1) a non-log btree
  52  * 2) a log btree and first log sub-transaction
  53  * 3) a log btree and second log sub-transaction
  54  */
  55 #define BTRFS_INODE_BTREE_ERR                   12
  56 #define BTRFS_INODE_BTREE_LOG1_ERR              13
  57 #define BTRFS_INODE_BTREE_LOG2_ERR              14
  58
  59 /* in memory btrfs inode */
  60 struct btrfs_inode {
  61         /* which subvolume this inode belongs to */
  62         struct btrfs_root *root;
  63
  64         /* key used to find this inode on disk.  This is used by the code
  65          * to read in roots of subvolumes
  66          */
  67         struct btrfs_key location;
  68
  69         /* Lock for counters */
  70         spinlock_t lock;
  71
  72         /* the extent_tree has caches of all the extent mappings to disk */
  73         struct extent_map_tree extent_tree;
  74
  75         /* the io_tree does range state (DIRTY, LOCKED etc) */
  76         struct extent_io_tree io_tree;
  77
  78         /* special utility tree used to record which mirrors have already been
  79          * tried when checksums fail for a given block
  80          */
  81         struct extent_io_tree io_failure_tree;
  82
  83         /* held while logging the inode in tree-log.c */
  84         struct mutex log_mutex;
  85
  86         /* held while doing delalloc reservations */
  87         struct mutex delalloc_mutex;
  88
  89         /* used to order data wrt metadata */
  90         struct btrfs_ordered_inode_tree ordered_tree;
  91
  92         /* list of all the delalloc inodes in the FS.  There are times we need
  93          * to write all the delalloc pages to disk, and this list is used
  94          * to walk them all.
  95          */
  96         struct list_head delalloc_inodes;
  97
  98         /* node for the red-black tree that links inodes in subvolume root */
  99         struct rb_node rb_node;
 100
 101         unsigned long runtime_flags;
 102
 103         /* Keep track of who's O_SYNC/fsyncing currently */
 104         atomic_t sync_writers;
 105
 106         /* full 64 bit generation number, struct vfs_inode doesn't have a big
 107          * enough field for this.
 108          */
 109         u64 generation;
 110
 111         /*
 112          * transid of the trans_handle that last modified this inode
 113          */
 114         u64 last_trans;
 115
 116         /*
 117          * transid that last logged this inode
 118          */
 119         u64 logged_trans;
 120
 121         /*
 122          * log transid when this inode was last modified
 123          */
 124         int last_sub_trans;
 125
 126         /* a local copy of root's last_log_commit */
 127         int last_log_commit;
 128
 129         /* total number of bytes pending delalloc, used by stat to calc the
 130          * real block usage of the file
 131          */
 132         u64 delalloc_bytes;
 133
 134         /*
 135          * total number of bytes pending defrag, used by stat to check whether
 136          * it needs COW.
 137          */
 138         u64 defrag_bytes;
 139
 140         /*
 141          * the size of the file stored in the metadata on disk.  data=ordered
 142          * means the in-memory i_size might be larger than the size on disk
 143          * because not all the blocks are written yet.
 144          */
 145         u64 disk_i_size;
 146
 147         /*
 148          * if this is a directory then index_cnt is the counter for the index
 149          * number for new files that are created
 150          */
 151         u64 index_cnt;
 152
 153         /* Cache the directory index number to speed the dir/file remove */
 154         u64 dir_index;
 155
 156         /* the fsync log has some corner cases that mean we have to check
 157          * directories to see if any unlinks have been done before
 158          * the directory was logged.  See tree-log.c for all the
 159          * details
 160          */
 161         u64 last_unlink_trans;
 162
 163         /*
 164          * Number of bytes outstanding that are going to need csums.  This is
 165          * used in ENOSPC accounting.
 166          */
 167         u64 csum_bytes;
 168
 169         /* flags field from the on disk inode */
 170         u32 flags;
 171
 172         /*
 173          * Counters to keep track of the number of extent item's we may use due
 174          * to delalloc and such.  outstanding_extents is the number of extent
 175          * items we think we'll end up using, and reserved_extents is the number
 176          * of extent items we've reserved metadata for.
 177          */
 178         unsigned outstanding_extents;
 179         unsigned reserved_extents;
 180
 181         /*
 182          * always compress this one file
 183          */
 184         unsigned force_compress;
 185
 186         struct btrfs_delayed_node *delayed_node;
 187
 188         struct inode vfs_inode;
 189 };
 190
 191 extern unsigned char btrfs_filetype_table[];
 192
 193 static inline struct btrfs_inode *BTRFS_I(struct inode *inode)
 194 {
 195         return container_of(inode, struct btrfs_inode, vfs_inode);
 196 }
 197
 198 static inline unsigned long btrfs_inode_hash(u64 objectid,
 199                                              const struct btrfs_root *root)
 200 {
 201         u64 h = objectid ^ (root->objectid * GOLDEN_RATIO_PRIME);
 202
 203 #if BITS_PER_LONG == 32
 204         h = (h >> 32) ^ (h & 0xffffffff);
 205 #endif
 206
 207         return (unsigned long)h;
 208 }
 209
 210 static inline void btrfs_insert_inode_hash(struct inode *inode)
 211 {
 212         unsigned long h = btrfs_inode_hash(inode->i_ino, BTRFS_I(inode)->root);
 213
 214         __insert_inode_hash(inode, h);
 215 }
 216
 217 static inline u64 btrfs_ino(struct inode *inode)
 218 {
 219         u64 ino = BTRFS_I(inode)->location.objectid;
 220
 221         /*
 222          * !ino: btree_inode
 223          * type == BTRFS_ROOT_ITEM_KEY: subvol dir
 224          */
 225         if (!ino || BTRFS_I(inode)->location.type == BTRFS_ROOT_ITEM_KEY)
 226                 ino = inode->i_ino;
 227         return ino;
 228 }
 229
 230 static inline void btrfs_i_size_write(struct inode *inode, u64 size)
 231 {
 232         i_size_write(inode, size);
 233         BTRFS_I(inode)->disk_i_size = size;
 234 }
 235
 236 static inline bool btrfs_is_free_space_inode(struct inode *inode)
 237 {
 238         struct btrfs_root *root = BTRFS_I(inode)->root;
 239
 240         if (root == root->fs_info->tree_root &&
 241             btrfs_ino(inode) != BTRFS_BTREE_INODE_OBJECTID)
 242                 return true;
 243         if (BTRFS_I(inode)->location.objectid == BTRFS_FREE_INO_OBJECTID)
 244                 return true;
 245         return false;
 246 }
 247
 248 static inline int btrfs_inode_in_log(struct inode *inode, u64 generation)
 249 {
 250         if (BTRFS_I(inode)->logged_trans == generation &&
 251             BTRFS_I(inode)->last_sub_trans <=
 252             BTRFS_I(inode)->last_log_commit &&
 253             BTRFS_I(inode)->last_sub_trans <=
 254             BTRFS_I(inode)->root->last_log_commit) {
 255                 /*
 256                  * After a ranged fsync we might have left some extent maps
 257                  * (that fall outside the fsync's range). So return false
 258                  * here if the list isn't empty, to make sure btrfs_log_inode()
 259                  * will be called and process those extent maps.
 260                  */
 261                 smp_mb();
 262                 if (list_empty(&BTRFS_I(inode)->extent_tree.modified_extents))
 263                         return 1;
 264         }
 265         return 0;
 266 }
 267
 268 #define BTRFS_DIO_ORIG_BIO_SUBMITTED    0x1
 269
 270 struct btrfs_dio_private {
 271         struct inode *inode;
 272         unsigned long flags;
 273         u64 logical_offset;
 274         u64 disk_bytenr;
 275         u64 bytes;
 276         void *private;
 277
 278         /* number of bios pending for this dio */
 279         atomic_t pending_bios;
 280
 281         /* IO errors */
 282         int errors;
 283
 284         /* orig_bio is our btrfs_io_bio */
 285         struct bio *orig_bio;
 286
 287         /* dio_bio came from fs/direct-io.c */
 288         struct bio *dio_bio;
 289
 290         /*
 291          * The original bio may be splited to several sub-bios, this is
 292          * done during endio of sub-bios
 293          */
 294         int (*subio_endio)(struct inode *, struct btrfs_io_bio *, int);
 295 };
 296
 297 /*
 298  * Disable DIO read nolock optimization, so new dio readers will be forced
 299  * to grab i_mutex. It is used to avoid the endless truncate due to
 300  * nonlocked dio read.
 301  */
 302 static inline void btrfs_inode_block_unlocked_dio(struct inode *inode)
 303 {
 304         set_bit(BTRFS_INODE_READDIO_NEED_LOCK, &BTRFS_I(inode)->runtime_flags);
 305         smp_mb();
 306 }
 307
 308 static inline void btrfs_inode_resume_unlocked_dio(struct inode *inode)
 309 {
 310         smp_mb__before_atomic();
 311         clear_bit(BTRFS_INODE_READDIO_NEED_LOCK,
 312                   &BTRFS_I(inode)->runtime_flags);
 313 }
 314
 315 bool btrfs_page_exists_in_range(struct inode *inode, loff_t start, loff_t end);
 316
 317 #endif