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 /* DIO is ready to submit */
  48 #define BTRFS_INODE_DIO_READY                   12
  49 /*
  50  * The following 3 bits are meant only for the btree inode.
  51  * When any of them is set, it means an error happened while writing an
  52  * extent buffer belonging to:
  53  * 1) a non-log btree
  54  * 2) a log btree and first log sub-transaction
  55  * 3) a log btree and second log sub-transaction
  56  */
  57 #define BTRFS_INODE_BTREE_ERR                   12
  58 #define BTRFS_INODE_BTREE_LOG1_ERR              13
  59 #define BTRFS_INODE_BTREE_LOG2_ERR              14
  60
  61 /* in memory btrfs inode */
  62 struct btrfs_inode {
  63         /* which subvolume this inode belongs to */
  64         struct btrfs_root *root;
  65
  66         /* key used to find this inode on disk.  This is used by the code
  67          * to read in roots of subvolumes
  68          */
  69         struct btrfs_key location;
  70
  71         /*
  72          * Lock for counters and all fields used to determine if the inode is in
  73          * the log or not (last_trans, last_sub_trans, last_log_commit,
  74          * logged_trans).
  75          */
  76         spinlock_t lock;
  77
  78         /* the extent_tree has caches of all the extent mappings to disk */
  79         struct extent_map_tree extent_tree;
  80
  81         /* the io_tree does range state (DIRTY, LOCKED etc) */
  82         struct extent_io_tree io_tree;
  83
  84         /* special utility tree used to record which mirrors have already been
  85          * tried when checksums fail for a given block
  86          */
  87         struct extent_io_tree io_failure_tree;
  88
  89         /* held while logging the inode in tree-log.c */
  90         struct mutex log_mutex;
  91
  92         /* held while doing delalloc reservations */
  93         struct mutex delalloc_mutex;
  94
  95         /* used to order data wrt metadata */
  96         struct btrfs_ordered_inode_tree ordered_tree;
  97
  98         /* list of all the delalloc inodes in the FS.  There are times we need
  99          * to write all the delalloc pages to disk, and this list is used
 100          * to walk them all.
 101          */
 102         struct list_head delalloc_inodes;
 103
 104         /* node for the red-black tree that links inodes in subvolume root */
 105         struct rb_node rb_node;
 106
 107         unsigned long runtime_flags;
 108
 109         /* Keep track of who's O_SYNC/fsyncing currently */
 110         atomic_t sync_writers;
 111
 112         /* full 64 bit generation number, struct vfs_inode doesn't have a big
 113          * enough field for this.
 114          */
 115         u64 generation;
 116
 117         /*
 118          * transid of the trans_handle that last modified this inode
 119          */
 120         u64 last_trans;
 121
 122         /*
 123          * transid that last logged this inode
 124          */
 125         u64 logged_trans;
 126
 127         /*
 128          * log transid when this inode was last modified
 129          */
 130         int last_sub_trans;
 131
 132         /* a local copy of root's last_log_commit */
 133         int last_log_commit;
 134
 135         /* total number of bytes pending delalloc, used by stat to calc the
 136          * real block usage of the file
 137          */
 138         u64 delalloc_bytes;
 139
 140         /*
 141          * total number of bytes pending defrag, used by stat to check whether
 142          * it needs COW.
 143          */
 144         u64 defrag_bytes;
 145
 146         /*
 147          * the size of the file stored in the metadata on disk.  data=ordered
 148          * means the in-memory i_size might be larger than the size on disk
 149          * because not all the blocks are written yet.
 150          */
 151         u64 disk_i_size;
 152
 153         /*
 154          * if this is a directory then index_cnt is the counter for the index
 155          * number for new files that are created
 156          */
 157         u64 index_cnt;
 158
 159         /* Cache the directory index number to speed the dir/file remove */
 160         u64 dir_index;
 161
 162         /* the fsync log has some corner cases that mean we have to check
 163          * directories to see if any unlinks have been done before
 164          * the directory was logged.  See tree-log.c for all the
 165          * details
 166          */
 167         u64 last_unlink_trans;
 168
 169         /*
 170          * Number of bytes outstanding that are going to need csums.  This is
 171          * used in ENOSPC accounting.
 172          */
 173         u64 csum_bytes;
 174
 175         /* flags field from the on disk inode */
 176         u32 flags;
 177
 178         /*
 179          * Counters to keep track of the number of extent item's we may use due
 180          * to delalloc and such.  outstanding_extents is the number of extent
 181          * items we think we'll end up using, and reserved_extents is the number
 182          * of extent items we've reserved metadata for.
 183          */
 184         unsigned outstanding_extents;
 185         unsigned reserved_extents;
 186
 187         /*
 188          * always compress this one file
 189          */
 190         unsigned force_compress;
 191
 192         struct btrfs_delayed_node *delayed_node;
 193
 194         /* File creation time. */
 195         struct timespec i_otime;
 196
 197         struct inode vfs_inode;
 198 };
 199
 200 extern unsigned char btrfs_filetype_table[];
 201
 202 static inline struct btrfs_inode *BTRFS_I(struct inode *inode)
 203 {
 204         return container_of(inode, struct btrfs_inode, vfs_inode);
 205 }
 206
 207 static inline unsigned long btrfs_inode_hash(u64 objectid,
 208                                              const struct btrfs_root *root)
 209 {
 210         u64 h = objectid ^ (root->objectid * GOLDEN_RATIO_PRIME);
 211
 212 #if BITS_PER_LONG == 32
 213         h = (h >> 32) ^ (h & 0xffffffff);
 214 #endif
 215
 216         return (unsigned long)h;
 217 }
 218
 219 static inline void btrfs_insert_inode_hash(struct inode *inode)
 220 {
 221         unsigned long h = btrfs_inode_hash(inode->i_ino, BTRFS_I(inode)->root);
 222
 223         __insert_inode_hash(inode, h);
 224 }
 225
 226 static inline u64 btrfs_ino(struct inode *inode)
 227 {
 228         u64 ino = BTRFS_I(inode)->location.objectid;
 229
 230         /*
 231          * !ino: btree_inode
 232          * type == BTRFS_ROOT_ITEM_KEY: subvol dir
 233          */
 234         if (!ino || BTRFS_I(inode)->location.type == BTRFS_ROOT_ITEM_KEY)
 235                 ino = inode->i_ino;
 236         return ino;
 237 }
 238
 239 static inline void btrfs_i_size_write(struct inode *inode, u64 size)
 240 {
 241         i_size_write(inode, size);
 242         BTRFS_I(inode)->disk_i_size = size;
 243 }
 244
 245 static inline bool btrfs_is_free_space_inode(struct inode *inode)
 246 {
 247         struct btrfs_root *root = BTRFS_I(inode)->root;
 248
 249         if (root == root->fs_info->tree_root &&
 250             btrfs_ino(inode) != BTRFS_BTREE_INODE_OBJECTID)
 251                 return true;
 252         if (BTRFS_I(inode)->location.objectid == BTRFS_FREE_INO_OBJECTID)
 253                 return true;
 254         return false;
 255 }
 256
 257 static inline int btrfs_inode_in_log(struct inode *inode, u64 generation)
 258 {
 259         int ret = 0;
 260
 261         spin_lock(&BTRFS_I(inode)->lock);
 262         if (BTRFS_I(inode)->logged_trans == generation &&
 263             BTRFS_I(inode)->last_sub_trans <=
 264             BTRFS_I(inode)->last_log_commit &&
 265             BTRFS_I(inode)->last_sub_trans <=
 266             BTRFS_I(inode)->root->last_log_commit) {
 267                 /*
 268                  * After a ranged fsync we might have left some extent maps
 269                  * (that fall outside the fsync's range). So return false
 270                  * here if the list isn't empty, to make sure btrfs_log_inode()
 271                  * will be called and process those extent maps.
 272                  */
 273                 smp_mb();
 274                 if (list_empty(&BTRFS_I(inode)->extent_tree.modified_extents))
 275                         ret = 1;
 276         }
 277         spin_unlock(&BTRFS_I(inode)->lock);
 278         return ret;
 279 }
 280
 281 #define BTRFS_DIO_ORIG_BIO_SUBMITTED    0x1
 282
 283 struct btrfs_dio_private {
 284         struct inode *inode;
 285         unsigned long flags;
 286         u64 logical_offset;
 287         u64 disk_bytenr;
 288         u64 bytes;
 289         void *private;
 290
 291         /* number of bios pending for this dio */
 292         atomic_t pending_bios;
 293
 294         /* IO errors */
 295         int errors;
 296
 297         /* orig_bio is our btrfs_io_bio */
 298         struct bio *orig_bio;
 299
 300         /* dio_bio came from fs/direct-io.c */
 301         struct bio *dio_bio;
 302
 303         /*
 304          * The original bio may be splited to several sub-bios, this is
 305          * done during endio of sub-bios
 306          */
 307         int (*subio_endio)(struct inode *, struct btrfs_io_bio *, int);
 308 };
 309
 310 /*
 311  * Disable DIO read nolock optimization, so new dio readers will be forced
 312  * to grab i_mutex. It is used to avoid the endless truncate due to
 313  * nonlocked dio read.
 314  */
 315 static inline void btrfs_inode_block_unlocked_dio(struct inode *inode)
 316 {
 317         set_bit(BTRFS_INODE_READDIO_NEED_LOCK, &BTRFS_I(inode)->runtime_flags);
 318         smp_mb();
 319 }
 320
 321 static inline void btrfs_inode_resume_unlocked_dio(struct inode *inode)
 322 {
 323         smp_mb__before_atomic();
 324         clear_bit(BTRFS_INODE_READDIO_NEED_LOCK,
 325                   &BTRFS_I(inode)->runtime_flags);
 326 }
 327
 328 bool btrfs_page_exists_in_range(struct inode *inode, loff_t start, loff_t end);
 329
 330 #endif