fs/ocfs2/ocfs2_fs.h

   1 /* -*- mode: c; c-basic-offset: 8; -*-
   2  * vim: noexpandtab sw=8 ts=8 sts=0:
   3  *
   4  * ocfs2_fs.h
   5  *
   6  * On-disk structures for OCFS2.
   7  *
   8  * Copyright (C) 2002, 2004 Oracle.  All rights reserved.
   9  *
  10  * This program is free software; you can redistribute it and/or
  11  * modify it under the terms of the GNU General Public
  12  * License, version 2,  as published by the Free Software Foundation.
  13  *
  14  * This program is distributed in the hope that it will be useful,
  15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  17  * General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU General Public
  20  * License along with this program; if not, write to the
  21  * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  22  * Boston, MA 021110-1307, USA.
  23  */
  24
  25 #ifndef _OCFS2_FS_H
  26 #define _OCFS2_FS_H
  27
  28 /* Version */
  29 #define OCFS2_MAJOR_REV_LEVEL           0
  30 #define OCFS2_MINOR_REV_LEVEL           90
  31
  32 /*
  33  * An OCFS2 volume starts this way:
  34  * Sector 0: Valid ocfs1_vol_disk_hdr that cleanly fails to mount OCFS.
  35  * Sector 1: Valid ocfs1_vol_label that cleanly fails to mount OCFS.
  36  * Block OCFS2_SUPER_BLOCK_BLKNO: OCFS2 superblock.
  37  *
  38  * All other structures are found from the superblock information.
  39  *
  40  * OCFS2_SUPER_BLOCK_BLKNO is in blocks, not sectors.  eg, for a
  41  * blocksize of 2K, it is 4096 bytes into disk.
  42  */
  43 #define OCFS2_SUPER_BLOCK_BLKNO         2
  44
  45 /*
  46  * Cluster size limits. The maximum is kept arbitrarily at 1 MB, and could
  47  * grow if needed.
  48  */
  49 #define OCFS2_MIN_CLUSTERSIZE           4096
  50 #define OCFS2_MAX_CLUSTERSIZE           1048576
  51
  52 /*
  53  * Blocks cannot be bigger than clusters, so the maximum blocksize is the
  54  * minimum cluster size.
  55  */
  56 #define OCFS2_MIN_BLOCKSIZE             512
  57 #define OCFS2_MAX_BLOCKSIZE             OCFS2_MIN_CLUSTERSIZE
  58
  59 /* Filesystem magic number */
  60 #define OCFS2_SUPER_MAGIC               0x7461636f
  61
  62 /* Object signatures */
  63 #define OCFS2_SUPER_BLOCK_SIGNATURE     "OCFSV2"
  64 #define OCFS2_INODE_SIGNATURE           "INODE01"
  65 #define OCFS2_EXTENT_BLOCK_SIGNATURE    "EXBLK01"
  66 #define OCFS2_GROUP_DESC_SIGNATURE      "GROUP01"
  67
  68 /* Compatibility flags */
  69 #define OCFS2_HAS_COMPAT_FEATURE(sb,mask)                       \
  70         ( OCFS2_SB(sb)->s_feature_compat & (mask) )
  71 #define OCFS2_HAS_RO_COMPAT_FEATURE(sb,mask)                    \
  72         ( OCFS2_SB(sb)->s_feature_ro_compat & (mask) )
  73 #define OCFS2_HAS_INCOMPAT_FEATURE(sb,mask)                     \
  74         ( OCFS2_SB(sb)->s_feature_incompat & (mask) )
  75 #define OCFS2_SET_COMPAT_FEATURE(sb,mask)                       \
  76         OCFS2_SB(sb)->s_feature_compat |= (mask)
  77 #define OCFS2_SET_RO_COMPAT_FEATURE(sb,mask)                    \
  78         OCFS2_SB(sb)->s_feature_ro_compat |= (mask)
  79 #define OCFS2_SET_INCOMPAT_FEATURE(sb,mask)                     \
  80         OCFS2_SB(sb)->s_feature_incompat |= (mask)
  81 #define OCFS2_CLEAR_COMPAT_FEATURE(sb,mask)                     \
  82         OCFS2_SB(sb)->s_feature_compat &= ~(mask)
  83 #define OCFS2_CLEAR_RO_COMPAT_FEATURE(sb,mask)                  \
  84         OCFS2_SB(sb)->s_feature_ro_compat &= ~(mask)
  85 #define OCFS2_CLEAR_INCOMPAT_FEATURE(sb,mask)                   \
  86         OCFS2_SB(sb)->s_feature_incompat &= ~(mask)
  87
  88 #define OCFS2_FEATURE_COMPAT_SUPP       OCFS2_FEATURE_COMPAT_BACKUP_SB
  89 #define OCFS2_FEATURE_INCOMPAT_SUPP     (OCFS2_FEATURE_INCOMPAT_LOCAL_MOUNT \
  90                                          | OCFS2_FEATURE_INCOMPAT_SPARSE_ALLOC \
  91                                          | OCFS2_FEATURE_INCOMPAT_INLINE_DATA \
  92                                          | OCFS2_FEATURE_INCOMPAT_EXTENDED_SLOT_MAP \
  93                                          | OCFS2_FEATURE_INCOMPAT_USERSPACE_STACK)
  94 #define OCFS2_FEATURE_RO_COMPAT_SUPP    OCFS2_FEATURE_RO_COMPAT_UNWRITTEN
  95
  96 /*
  97  * Heartbeat-only devices are missing journals and other files.  The
  98  * filesystem driver can't load them, but the library can.  Never put
  99  * this in OCFS2_FEATURE_INCOMPAT_SUPP, *ever*.
 100  */
 101 #define OCFS2_FEATURE_INCOMPAT_HEARTBEAT_DEV    0x0002
 102
 103 /*
 104  * tunefs sets this incompat flag before starting the resize and clears it
 105  * at the end. This flag protects users from inadvertently mounting the fs
 106  * after an aborted run without fsck-ing.
 107  */
 108 #define OCFS2_FEATURE_INCOMPAT_RESIZE_INPROG    0x0004
 109
 110 /* Used to denote a non-clustered volume */
 111 #define OCFS2_FEATURE_INCOMPAT_LOCAL_MOUNT      0x0008
 112
 113 /* Support for sparse allocation in b-trees */
 114 #define OCFS2_FEATURE_INCOMPAT_SPARSE_ALLOC     0x0010
 115
 116 /*
 117  * Tunefs sets this incompat flag before starting an operation which
 118  * would require cleanup on abort. This is done to protect users from
 119  * inadvertently mounting the fs after an aborted run without
 120  * fsck-ing.
 121  *
 122  * s_tunefs_flags on the super block describes precisely which
 123  * operations were in progress.
 124  */
 125 #define OCFS2_FEATURE_INCOMPAT_TUNEFS_INPROG    0x0020
 126
 127 /* Support for data packed into inode blocks */
 128 #define OCFS2_FEATURE_INCOMPAT_INLINE_DATA      0x0040
 129
 130 /* Support for the extended slot map */
 131 #define OCFS2_FEATURE_INCOMPAT_EXTENDED_SLOT_MAP 0x100
 132
 133
 134 /*
 135  * Support for alternate, userspace cluster stacks.  If set, the superblock
 136  * field s_cluster_info contains a tag for the alternate stack in use as
 137  * well as the name of the cluster being joined.
 138  * mount.ocfs2 must pass in a matching stack name.
 139  *
 140  * If not set, the classic stack will be used.  This is compatbile with
 141  * all older versions.
 142  */
 143 #define OCFS2_FEATURE_INCOMPAT_USERSPACE_STACK  0x0080
 144
 145 /*
 146  * backup superblock flag is used to indicate that this volume
 147  * has backup superblocks.
 148  */
 149 #define OCFS2_FEATURE_COMPAT_BACKUP_SB          0x0001
 150
 151 /*
 152  * Unwritten extents support.
 153  */
 154 #define OCFS2_FEATURE_RO_COMPAT_UNWRITTEN       0x0001
 155
 156 /* The byte offset of the first backup block will be 1G.
 157  * The following will be 4G, 16G, 64G, 256G and 1T.
 158  */
 159 #define OCFS2_BACKUP_SB_START                   1 << 30
 160
 161 /* the max backup superblock nums */
 162 #define OCFS2_MAX_BACKUP_SUPERBLOCKS    6
 163
 164 /*
 165  * Flags on ocfs2_super_block.s_tunefs_flags
 166  */
 167 #define OCFS2_TUNEFS_INPROG_REMOVE_SLOT         0x0001  /* Removing slots */
 168
 169 /*
 170  * Flags on ocfs2_dinode.i_flags
 171  */
 172 #define OCFS2_VALID_FL          (0x00000001)    /* Inode is valid */
 173 #define OCFS2_UNUSED2_FL        (0x00000002)
 174 #define OCFS2_ORPHANED_FL       (0x00000004)    /* On the orphan list */
 175 #define OCFS2_UNUSED3_FL        (0x00000008)
 176 /* System inode flags */
 177 #define OCFS2_SYSTEM_FL         (0x00000010)    /* System inode */
 178 #define OCFS2_SUPER_BLOCK_FL    (0x00000020)    /* Super block */
 179 #define OCFS2_LOCAL_ALLOC_FL    (0x00000040)    /* Slot local alloc bitmap */
 180 #define OCFS2_BITMAP_FL         (0x00000080)    /* Allocation bitmap */
 181 #define OCFS2_JOURNAL_FL        (0x00000100)    /* Slot local journal */
 182 #define OCFS2_HEARTBEAT_FL      (0x00000200)    /* Heartbeat area */
 183 #define OCFS2_CHAIN_FL          (0x00000400)    /* Chain allocator */
 184 #define OCFS2_DEALLOC_FL        (0x00000800)    /* Truncate log */
 185
 186 /*
 187  * Flags on ocfs2_dinode.i_dyn_features
 188  *
 189  * These can change much more often than i_flags. When adding flags,
 190  * keep in mind that i_dyn_features is only 16 bits wide.
 191  */
 192 #define OCFS2_INLINE_DATA_FL    (0x0001)        /* Data stored in inode block */
 193 #define OCFS2_HAS_XATTR_FL      (0x0002)
 194 #define OCFS2_INLINE_XATTR_FL   (0x0004)
 195 #define OCFS2_INDEXED_DIR_FL    (0x0008)
 196
 197 /* Inode attributes, keep in sync with EXT2 */
 198 #define OCFS2_SECRM_FL          (0x00000001)    /* Secure deletion */
 199 #define OCFS2_UNRM_FL           (0x00000002)    /* Undelete */
 200 #define OCFS2_COMPR_FL          (0x00000004)    /* Compress file */
 201 #define OCFS2_SYNC_FL           (0x00000008)    /* Synchronous updates */
 202 #define OCFS2_IMMUTABLE_FL      (0x00000010)    /* Immutable file */
 203 #define OCFS2_APPEND_FL         (0x00000020)    /* writes to file may only append */
 204 #define OCFS2_NODUMP_FL         (0x00000040)    /* do not dump file */
 205 #define OCFS2_NOATIME_FL        (0x00000080)    /* do not update atime */
 206 #define OCFS2_DIRSYNC_FL        (0x00010000)    /* dirsync behaviour (directories only) */
 207
 208 #define OCFS2_FL_VISIBLE        (0x000100FF)    /* User visible flags */
 209 #define OCFS2_FL_MODIFIABLE     (0x000100FF)    /* User modifiable flags */
 210
 211 /*
 212  * Extent record flags (e_node.leaf.flags)
 213  */
 214 #define OCFS2_EXT_UNWRITTEN     (0x01)  /* Extent is allocated but
 215                                          * unwritten */
 216
 217 /*
 218  * ioctl commands
 219  */
 220 #define OCFS2_IOC_GETFLAGS      _IOR('f', 1, long)
 221 #define OCFS2_IOC_SETFLAGS      _IOW('f', 2, long)
 222 #define OCFS2_IOC32_GETFLAGS    _IOR('f', 1, int)
 223 #define OCFS2_IOC32_SETFLAGS    _IOW('f', 2, int)
 224
 225 /*
 226  * Space reservation / allocation / free ioctls and argument structure
 227  * are designed to be compatible with XFS.
 228  *
 229  * ALLOCSP* and FREESP* are not and will never be supported, but are
 230  * included here for completeness.
 231  */
 232 struct ocfs2_space_resv {
 233         __s16           l_type;
 234         __s16           l_whence;
 235         __s64           l_start;
 236         __s64           l_len;          /* len == 0 means until end of file */
 237         __s32           l_sysid;
 238         __u32           l_pid;
 239         __s32           l_pad[4];       /* reserve area                     */
 240 };
 241
 242 #define OCFS2_IOC_ALLOCSP               _IOW ('X', 10, struct ocfs2_space_resv)
 243 #define OCFS2_IOC_FREESP                _IOW ('X', 11, struct ocfs2_space_resv)
 244 #define OCFS2_IOC_RESVSP                _IOW ('X', 40, struct ocfs2_space_resv)
 245 #define OCFS2_IOC_UNRESVSP      _IOW ('X', 41, struct ocfs2_space_resv)
 246 #define OCFS2_IOC_ALLOCSP64     _IOW ('X', 36, struct ocfs2_space_resv)
 247 #define OCFS2_IOC_FREESP64      _IOW ('X', 37, struct ocfs2_space_resv)
 248 #define OCFS2_IOC_RESVSP64      _IOW ('X', 42, struct ocfs2_space_resv)
 249 #define OCFS2_IOC_UNRESVSP64    _IOW ('X', 43, struct ocfs2_space_resv)
 250
 251 /* Used to pass group descriptor data when online resize is done */
 252 struct ocfs2_new_group_input {
 253         __u64 group;            /* Group descriptor's blkno. */
 254         __u32 clusters;         /* Total number of clusters in this group */
 255         __u32 frees;            /* Total free clusters in this group */
 256         __u16 chain;            /* Chain for this group */
 257         __u16 reserved1;
 258         __u32 reserved2;
 259 };
 260
 261 #define OCFS2_IOC_GROUP_EXTEND  _IOW('o', 1, int)
 262 #define OCFS2_IOC_GROUP_ADD     _IOW('o', 2,struct ocfs2_new_group_input)
 263 #define OCFS2_IOC_GROUP_ADD64   _IOW('o', 3,struct ocfs2_new_group_input)
 264
 265 /*
 266  * Journal Flags (ocfs2_dinode.id1.journal1.i_flags)
 267  */
 268 #define OCFS2_JOURNAL_DIRTY_FL  (0x00000001)    /* Journal needs recovery */
 269
 270 /*
 271  * superblock s_state flags
 272  */
 273 #define OCFS2_ERROR_FS          (0x00000001)    /* FS saw errors */
 274
 275 /* Limit of space in ocfs2_dir_entry */
 276 #define OCFS2_MAX_FILENAME_LEN          255
 277
 278 /* Maximum slots on an ocfs2 file system */
 279 #define OCFS2_MAX_SLOTS                 255
 280
 281 /* Slot map indicator for an empty slot */
 282 #define OCFS2_INVALID_SLOT              -1
 283
 284 #define OCFS2_VOL_UUID_LEN              16
 285 #define OCFS2_MAX_VOL_LABEL_LEN         64
 286
 287 /* The alternate, userspace stack fields */
 288 #define OCFS2_STACK_LABEL_LEN           4
 289 #define OCFS2_CLUSTER_NAME_LEN          16
 290
 291 /* Journal limits (in bytes) */
 292 #define OCFS2_MIN_JOURNAL_SIZE          (4 * 1024 * 1024)
 293
 294 /*
 295  * Default local alloc size (in megabytes)
 296  *
 297  * The value chosen should be such that most allocations, including new
 298  * block groups, use local alloc.
 299  */
 300 #define OCFS2_DEFAULT_LOCAL_ALLOC_SIZE  8
 301
 302 struct ocfs2_system_inode_info {
 303         char    *si_name;
 304         int     si_iflags;
 305         int     si_mode;
 306 };
 307
 308 /* System file index */
 309 enum {
 310         BAD_BLOCK_SYSTEM_INODE = 0,
 311         GLOBAL_INODE_ALLOC_SYSTEM_INODE,
 312         SLOT_MAP_SYSTEM_INODE,
 313 #define OCFS2_FIRST_ONLINE_SYSTEM_INODE SLOT_MAP_SYSTEM_INODE
 314         HEARTBEAT_SYSTEM_INODE,
 315         GLOBAL_BITMAP_SYSTEM_INODE,
 316 #define OCFS2_LAST_GLOBAL_SYSTEM_INODE GLOBAL_BITMAP_SYSTEM_INODE
 317         ORPHAN_DIR_SYSTEM_INODE,
 318         EXTENT_ALLOC_SYSTEM_INODE,
 319         INODE_ALLOC_SYSTEM_INODE,
 320         JOURNAL_SYSTEM_INODE,
 321         LOCAL_ALLOC_SYSTEM_INODE,
 322         TRUNCATE_LOG_SYSTEM_INODE,
 323         NUM_SYSTEM_INODES
 324 };
 325
 326 static struct ocfs2_system_inode_info ocfs2_system_inodes[NUM_SYSTEM_INODES] = {
 327         /* Global system inodes (single copy) */
 328         /* The first two are only used from userspace mfks/tunefs */
 329         [BAD_BLOCK_SYSTEM_INODE]                = { "bad_blocks", 0, S_IFREG | 0644 },
 330         [GLOBAL_INODE_ALLOC_SYSTEM_INODE]       = { "global_inode_alloc", OCFS2_BITMAP_FL | OCFS2_CHAIN_FL, S_IFREG | 0644 },
 331
 332         /* These are used by the running filesystem */
 333         [SLOT_MAP_SYSTEM_INODE]                 = { "slot_map", 0, S_IFREG | 0644 },
 334         [HEARTBEAT_SYSTEM_INODE]                = { "heartbeat", OCFS2_HEARTBEAT_FL, S_IFREG | 0644 },
 335         [GLOBAL_BITMAP_SYSTEM_INODE]            = { "global_bitmap", 0, S_IFREG | 0644 },
 336
 337         /* Slot-specific system inodes (one copy per slot) */
 338         [ORPHAN_DIR_SYSTEM_INODE]               = { "orphan_dir:%04d", 0, S_IFDIR | 0755 },
 339         [EXTENT_ALLOC_SYSTEM_INODE]             = { "extent_alloc:%04d", OCFS2_BITMAP_FL | OCFS2_CHAIN_FL, S_IFREG | 0644 },
 340         [INODE_ALLOC_SYSTEM_INODE]              = { "inode_alloc:%04d", OCFS2_BITMAP_FL | OCFS2_CHAIN_FL, S_IFREG | 0644 },
 341         [JOURNAL_SYSTEM_INODE]                  = { "journal:%04d", OCFS2_JOURNAL_FL, S_IFREG | 0644 },
 342         [LOCAL_ALLOC_SYSTEM_INODE]              = { "local_alloc:%04d", OCFS2_BITMAP_FL | OCFS2_LOCAL_ALLOC_FL, S_IFREG | 0644 },
 343         [TRUNCATE_LOG_SYSTEM_INODE]             = { "truncate_log:%04d", OCFS2_DEALLOC_FL, S_IFREG | 0644 }
 344 };
 345
 346 /* Parameter passed from mount.ocfs2 to module */
 347 #define OCFS2_HB_NONE                   "heartbeat=none"
 348 #define OCFS2_HB_LOCAL                  "heartbeat=local"
 349
 350 /*
 351  * OCFS2 directory file types.  Only the low 3 bits are used.  The
 352  * other bits are reserved for now.
 353  */
 354 #define OCFS2_FT_UNKNOWN        0
 355 #define OCFS2_FT_REG_FILE       1
 356 #define OCFS2_FT_DIR            2
 357 #define OCFS2_FT_CHRDEV         3
 358 #define OCFS2_FT_BLKDEV         4
 359 #define OCFS2_FT_FIFO           5
 360 #define OCFS2_FT_SOCK           6
 361 #define OCFS2_FT_SYMLINK        7
 362
 363 #define OCFS2_FT_MAX            8
 364
 365 /*
 366  * OCFS2_DIR_PAD defines the directory entries boundaries
 367  *
 368  * NOTE: It must be a multiple of 4
 369  */
 370 #define OCFS2_DIR_PAD                   4
 371 #define OCFS2_DIR_ROUND                 (OCFS2_DIR_PAD - 1)
 372 #define OCFS2_DIR_MEMBER_LEN            offsetof(struct ocfs2_dir_entry, name)
 373 #define OCFS2_DIR_REC_LEN(name_len)     (((name_len) + OCFS2_DIR_MEMBER_LEN + \
 374                                           OCFS2_DIR_ROUND) & \
 375                                          ~OCFS2_DIR_ROUND)
 376
 377 #define OCFS2_LINK_MAX          32000
 378
 379 #define S_SHIFT                 12
 380 static unsigned char ocfs2_type_by_mode[S_IFMT >> S_SHIFT] = {
 381         [S_IFREG >> S_SHIFT]  = OCFS2_FT_REG_FILE,
 382         [S_IFDIR >> S_SHIFT]  = OCFS2_FT_DIR,
 383         [S_IFCHR >> S_SHIFT]  = OCFS2_FT_CHRDEV,
 384         [S_IFBLK >> S_SHIFT]  = OCFS2_FT_BLKDEV,
 385         [S_IFIFO >> S_SHIFT]  = OCFS2_FT_FIFO,
 386         [S_IFSOCK >> S_SHIFT] = OCFS2_FT_SOCK,
 387         [S_IFLNK >> S_SHIFT]  = OCFS2_FT_SYMLINK,
 388 };
 389
 390
 391 /*
 392  * Convenience casts
 393  */
 394 #define OCFS2_RAW_SB(dinode)            (&((dinode)->id2.i_super))
 395
 396 /*
 397  * On disk extent record for OCFS2
 398  * It describes a range of clusters on disk.
 399  *
 400  * Length fields are divided into interior and leaf node versions.
 401  * This leaves room for a flags field (OCFS2_EXT_*) in the leaf nodes.
 402  */
 403 struct ocfs2_extent_rec {
 404 /*00*/  __le32 e_cpos;          /* Offset into the file, in clusters */
 405         union {
 406                 __le32 e_int_clusters; /* Clusters covered by all children */
 407                 struct {
 408                         __le16 e_leaf_clusters; /* Clusters covered by this
 409                                                    extent */
 410                         __u8 e_reserved1;
 411                         __u8 e_flags; /* Extent flags */
 412                 };
 413         };
 414         __le64 e_blkno;         /* Physical disk offset, in blocks */
 415 /*10*/
 416 };
 417
 418 struct ocfs2_chain_rec {
 419         __le32 c_free;  /* Number of free bits in this chain. */
 420         __le32 c_total; /* Number of total bits in this chain */
 421         __le64 c_blkno; /* Physical disk offset (blocks) of 1st group */
 422 };
 423
 424 struct ocfs2_truncate_rec {
 425         __le32 t_start;         /* 1st cluster in this log */
 426         __le32 t_clusters;      /* Number of total clusters covered */
 427 };
 428
 429 /*
 430  * On disk extent list for OCFS2 (node in the tree).  Note that this
 431  * is contained inside ocfs2_dinode or ocfs2_extent_block, so the
 432  * offsets are relative to ocfs2_dinode.id2.i_list or
 433  * ocfs2_extent_block.h_list, respectively.
 434  */
 435 struct ocfs2_extent_list {
 436 /*00*/  __le16 l_tree_depth;            /* Extent tree depth from this
 437                                            point.  0 means data extents
 438                                            hang directly off this
 439                                            header (a leaf)
 440                                            NOTE: The high 8 bits cannot be
 441                                            used - tree_depth is never that big.
 442                                         */
 443         __le16 l_count;                 /* Number of extent records */
 444         __le16 l_next_free_rec;         /* Next unused extent slot */
 445         __le16 l_reserved1;
 446         __le64 l_reserved2;             /* Pad to
 447                                            sizeof(ocfs2_extent_rec) */
 448 /*10*/  struct ocfs2_extent_rec l_recs[0];      /* Extent records */
 449 };
 450
 451 /*
 452  * On disk allocation chain list for OCFS2.  Note that this is
 453  * contained inside ocfs2_dinode, so the offsets are relative to
 454  * ocfs2_dinode.id2.i_chain.
 455  */
 456 struct ocfs2_chain_list {
 457 /*00*/  __le16 cl_cpg;                  /* Clusters per Block Group */
 458         __le16 cl_bpc;                  /* Bits per cluster */
 459         __le16 cl_count;                /* Total chains in this list */
 460         __le16 cl_next_free_rec;        /* Next unused chain slot */
 461         __le64 cl_reserved1;
 462 /*10*/  struct ocfs2_chain_rec cl_recs[0];      /* Chain records */
 463 };
 464
 465 /*
 466  * On disk deallocation log for OCFS2.  Note that this is
 467  * contained inside ocfs2_dinode, so the offsets are relative to
 468  * ocfs2_dinode.id2.i_dealloc.
 469  */
 470 struct ocfs2_truncate_log {
 471 /*00*/  __le16 tl_count;                /* Total records in this log */
 472         __le16 tl_used;                 /* Number of records in use */
 473         __le32 tl_reserved1;
 474 /*08*/  struct ocfs2_truncate_rec tl_recs[0];   /* Truncate records */
 475 };
 476
 477 /*
 478  * On disk extent block (indirect block) for OCFS2
 479  */
 480 struct ocfs2_extent_block
 481 {
 482 /*00*/  __u8 h_signature[8];            /* Signature for verification */
 483         __le64 h_reserved1;
 484 /*10*/  __le16 h_suballoc_slot;         /* Slot suballocator this
 485                                            extent_header belongs to */
 486         __le16 h_suballoc_bit;          /* Bit offset in suballocator
 487                                            block group */
 488         __le32 h_fs_generation;         /* Must match super block */
 489         __le64 h_blkno;                 /* Offset on disk, in blocks */
 490 /*20*/  __le64 h_reserved3;
 491         __le64 h_next_leaf_blk;         /* Offset on disk, in blocks,
 492                                            of next leaf header pointing
 493                                            to data */
 494 /*30*/  struct ocfs2_extent_list h_list;        /* Extent record list */
 495 /* Actual on-disk size is one block */
 496 };
 497
 498 /*
 499  * On disk slot map for OCFS2.  This defines the contents of the "slot_map"
 500  * system file.  A slot is valid if it contains a node number >= 0.  The
 501  * value -1 (0xFFFF) is OCFS2_INVALID_SLOT.  This marks a slot empty.
 502  */
 503 struct ocfs2_slot_map {
 504 /*00*/  __le16 sm_slots[0];
 505 /*
 506  * Actual on-disk size is one block.  OCFS2_MAX_SLOTS is 255,
 507  * 255 * sizeof(__le16) == 512B, within the 512B block minimum blocksize.
 508  */
 509 };
 510
 511 struct ocfs2_extended_slot {
 512 /*00*/  __u8    es_valid;
 513         __u8    es_reserved1[3];
 514         __le32  es_node_num;
 515 /*10*/
 516 };
 517
 518 /*
 519  * The extended slot map, used when OCFS2_FEATURE_INCOMPAT_EXTENDED_SLOT_MAP
 520  * is set.  It separates out the valid marker from the node number, and
 521  * has room to grow.  Unlike the old slot map, this format is defined by
 522  * i_size.
 523  */
 524 struct ocfs2_slot_map_extended {
 525 /*00*/  struct ocfs2_extended_slot se_slots[0];
 526 /*
 527  * Actual size is i_size of the slot_map system file.  It should
 528  * match s_max_slots * sizeof(struct ocfs2_extended_slot)
 529  */
 530 };
 531
 532 struct ocfs2_cluster_info {
 533 /*00*/  __u8   ci_stack[OCFS2_STACK_LABEL_LEN];
 534         __le32 ci_reserved;
 535 /*08*/  __u8   ci_cluster[OCFS2_CLUSTER_NAME_LEN];
 536 /*18*/
 537 };
 538
 539 /*
 540  * On disk superblock for OCFS2
 541  * Note that it is contained inside an ocfs2_dinode, so all offsets
 542  * are relative to the start of ocfs2_dinode.id2.
 543  */
 544 struct ocfs2_super_block {
 545 /*00*/  __le16 s_major_rev_level;
 546         __le16 s_minor_rev_level;
 547         __le16 s_mnt_count;
 548         __le16 s_max_mnt_count;
 549         __le16 s_state;                 /* File system state */
 550         __le16 s_errors;                        /* Behaviour when detecting errors */
 551         __le32 s_checkinterval;         /* Max time between checks */
 552 /*10*/  __le64 s_lastcheck;             /* Time of last check */
 553         __le32 s_creator_os;            /* OS */
 554         __le32 s_feature_compat;                /* Compatible feature set */
 555 /*20*/  __le32 s_feature_incompat;      /* Incompatible feature set */
 556         __le32 s_feature_ro_compat;     /* Readonly-compatible feature set */
 557         __le64 s_root_blkno;            /* Offset, in blocks, of root directory
 558                                            dinode */
 559 /*30*/  __le64 s_system_dir_blkno;      /* Offset, in blocks, of system
 560                                            directory dinode */
 561         __le32 s_blocksize_bits;                /* Blocksize for this fs */
 562         __le32 s_clustersize_bits;      /* Clustersize for this fs */
 563 /*40*/  __le16 s_max_slots;             /* Max number of simultaneous mounts
 564                                            before tunefs required */
 565         __le16 s_tunefs_flag;
 566         __le32 s_reserved1;
 567         __le64 s_first_cluster_group;   /* Block offset of 1st cluster
 568                                          * group header */
 569 /*50*/  __u8  s_label[OCFS2_MAX_VOL_LABEL_LEN]; /* Label for mounting, etc. */
 570 /*90*/  __u8  s_uuid[OCFS2_VOL_UUID_LEN];       /* 128-bit uuid */
 571 /*A0*/  struct ocfs2_cluster_info s_cluster_info; /* Selected userspace
 572                                                      stack.  Only valid
 573                                                      with INCOMPAT flag. */
 574 /*B8*/  __le64 s_reserved2[17];         /* Fill out superblock */
 575 /*140*/
 576
 577         /*
 578          * NOTE: As stated above, all offsets are relative to
 579          * ocfs2_dinode.id2, which is at 0xC0 in the inode.
 580          * 0xC0 + 0x140 = 0x200 or 512 bytes.  A superblock must fit within
 581          * our smallest blocksize, which is 512 bytes.  To ensure this,
 582          * we reserve the space in s_reserved2.  Anything past s_reserved2
 583          * will not be available on the smallest blocksize.
 584          */
 585 };
 586
 587 /*
 588  * Local allocation bitmap for OCFS2 slots
 589  * Note that it exists inside an ocfs2_dinode, so all offsets are
 590  * relative to the start of ocfs2_dinode.id2.
 591  */
 592 struct ocfs2_local_alloc
 593 {
 594 /*00*/  __le32 la_bm_off;       /* Starting bit offset in main bitmap */
 595         __le16 la_size;         /* Size of included bitmap, in bytes */
 596         __le16 la_reserved1;
 597         __le64 la_reserved2;
 598 /*10*/  __u8   la_bitmap[0];
 599 };
 600
 601 /*
 602  * Data-in-inode header. This is only used if i_dyn_features has
 603  * OCFS2_INLINE_DATA_FL set.
 604  */
 605 struct ocfs2_inline_data
 606 {
 607 /*00*/  __le16  id_count;       /* Number of bytes that can be used
 608                                  * for data, starting at id_data */
 609         __le16  id_reserved0;
 610         __le32  id_reserved1;
 611         __u8    id_data[0];     /* Start of user data */
 612 };
 613
 614 /*
 615  * On disk inode for OCFS2
 616  */
 617 struct ocfs2_dinode {
 618 /*00*/  __u8 i_signature[8];            /* Signature for validation */
 619         __le32 i_generation;            /* Generation number */
 620         __le16 i_suballoc_slot;         /* Slot suballocator this inode
 621                                            belongs to */
 622         __le16 i_suballoc_bit;          /* Bit offset in suballocator
 623                                            block group */
 624 /*10*/  __le32 i_reserved0;
 625         __le32 i_clusters;              /* Cluster count */
 626         __le32 i_uid;                   /* Owner UID */
 627         __le32 i_gid;                   /* Owning GID */
 628 /*20*/  __le64 i_size;                  /* Size in bytes */
 629         __le16 i_mode;                  /* File mode */
 630         __le16 i_links_count;           /* Links count */
 631         __le32 i_flags;                 /* File flags */
 632 /*30*/  __le64 i_atime;                 /* Access time */
 633         __le64 i_ctime;                 /* Creation time */
 634 /*40*/  __le64 i_mtime;                 /* Modification time */
 635         __le64 i_dtime;                 /* Deletion time */
 636 /*50*/  __le64 i_blkno;                 /* Offset on disk, in blocks */
 637         __le64 i_last_eb_blk;           /* Pointer to last extent
 638                                            block */
 639 /*60*/  __le32 i_fs_generation;         /* Generation per fs-instance */
 640         __le32 i_atime_nsec;
 641         __le32 i_ctime_nsec;
 642         __le32 i_mtime_nsec;
 643         __le32 i_attr;
 644         __le16 i_orphaned_slot;         /* Only valid when OCFS2_ORPHANED_FL
 645                                            was set in i_flags */
 646         __le16 i_dyn_features;
 647 /*70*/  __le64 i_reserved2[8];
 648 /*B8*/  union {
 649                 __le64 i_pad1;          /* Generic way to refer to this
 650                                            64bit union */
 651                 struct {
 652                         __le64 i_rdev;  /* Device number */
 653                 } dev1;
 654                 struct {                /* Info for bitmap system
 655                                            inodes */
 656                         __le32 i_used;  /* Bits (ie, clusters) used  */
 657                         __le32 i_total; /* Total bits (clusters)
 658                                            available */
 659                 } bitmap1;
 660                 struct {                /* Info for journal system
 661                                            inodes */
 662                         __le32 ij_flags;        /* Mounted, version, etc. */
 663                         __le32 ij_recovery_generation; /* Incremented when the
 664                                                           journal is recovered
 665                                                           after an unclean
 666                                                           shutdown */
 667                 } journal1;
 668         } id1;                          /* Inode type dependant 1 */
 669 /*C0*/  union {
 670                 struct ocfs2_super_block        i_super;
 671                 struct ocfs2_local_alloc        i_lab;
 672                 struct ocfs2_chain_list         i_chain;
 673                 struct ocfs2_extent_list        i_list;
 674                 struct ocfs2_truncate_log       i_dealloc;
 675                 struct ocfs2_inline_data        i_data;
 676                 __u8                            i_symlink[0];
 677         } id2;
 678 /* Actual on-disk size is one block */
 679 };
 680
 681 /*
 682  * On-disk directory entry structure for OCFS2
 683  *
 684  * Packed as this structure could be accessed unaligned on 64-bit platforms
 685  */
 686 struct ocfs2_dir_entry {
 687 /*00*/  __le64   inode;                  /* Inode number */
 688         __le16   rec_len;                /* Directory entry length */
 689         __u8    name_len;               /* Name length */
 690         __u8    file_type;
 691 /*0C*/  char    name[OCFS2_MAX_FILENAME_LEN];   /* File name */
 692 /* Actual on-disk length specified by rec_len */
 693 } __attribute__ ((packed));
 694
 695 /*
 696  * On disk allocator group structure for OCFS2
 697  */
 698 struct ocfs2_group_desc
 699 {
 700 /*00*/  __u8    bg_signature[8];        /* Signature for validation */
 701         __le16   bg_size;                /* Size of included bitmap in
 702                                            bytes. */
 703         __le16   bg_bits;                /* Bits represented by this
 704                                            group. */
 705         __le16  bg_free_bits_count;     /* Free bits count */
 706         __le16   bg_chain;               /* What chain I am in. */
 707 /*10*/  __le32   bg_generation;
 708         __le32  bg_reserved1;
 709         __le64   bg_next_group;          /* Next group in my list, in
 710                                            blocks */
 711 /*20*/  __le64   bg_parent_dinode;       /* dinode which owns me, in
 712                                            blocks */
 713         __le64   bg_blkno;               /* Offset on disk, in blocks */
 714 /*30*/  __le64   bg_reserved2[2];
 715 /*40*/  __u8    bg_bitmap[0];
 716 };
 717
 718 #ifdef __KERNEL__
 719 static inline int ocfs2_fast_symlink_chars(struct super_block *sb)
 720 {
 721         return  sb->s_blocksize -
 722                  offsetof(struct ocfs2_dinode, id2.i_symlink);
 723 }
 724
 725 static inline int ocfs2_max_inline_data(struct super_block *sb)
 726 {
 727         return sb->s_blocksize -
 728                 offsetof(struct ocfs2_dinode, id2.i_data.id_data);
 729 }
 730
 731 static inline int ocfs2_extent_recs_per_inode(struct super_block *sb)
 732 {
 733         int size;
 734
 735         size = sb->s_blocksize -
 736                 offsetof(struct ocfs2_dinode, id2.i_list.l_recs);
 737
 738         return size / sizeof(struct ocfs2_extent_rec);
 739 }
 740
 741 static inline int ocfs2_chain_recs_per_inode(struct super_block *sb)
 742 {
 743         int size;
 744
 745         size = sb->s_blocksize -
 746                 offsetof(struct ocfs2_dinode, id2.i_chain.cl_recs);
 747
 748         return size / sizeof(struct ocfs2_chain_rec);
 749 }
 750
 751 static inline u16 ocfs2_extent_recs_per_eb(struct super_block *sb)
 752 {
 753         int size;
 754
 755         size = sb->s_blocksize -
 756                 offsetof(struct ocfs2_extent_block, h_list.l_recs);
 757
 758         return size / sizeof(struct ocfs2_extent_rec);
 759 }
 760
 761 static inline u16 ocfs2_local_alloc_size(struct super_block *sb)
 762 {
 763         u16 size;
 764
 765         size = sb->s_blocksize -
 766                 offsetof(struct ocfs2_dinode, id2.i_lab.la_bitmap);
 767
 768         return size;
 769 }
 770
 771 static inline int ocfs2_group_bitmap_size(struct super_block *sb)
 772 {
 773         int size;
 774
 775         size = sb->s_blocksize -
 776                 offsetof(struct ocfs2_group_desc, bg_bitmap);
 777
 778         return size;
 779 }
 780
 781 static inline int ocfs2_truncate_recs_per_inode(struct super_block *sb)
 782 {
 783         int size;
 784
 785         size = sb->s_blocksize -
 786                 offsetof(struct ocfs2_dinode, id2.i_dealloc.tl_recs);
 787
 788         return size / sizeof(struct ocfs2_truncate_rec);
 789 }
 790
 791 static inline u64 ocfs2_backup_super_blkno(struct super_block *sb, int index)
 792 {
 793         u64 offset = OCFS2_BACKUP_SB_START;
 794
 795         if (index >= 0 && index < OCFS2_MAX_BACKUP_SUPERBLOCKS) {
 796                 offset <<= (2 * index);
 797                 offset >>= sb->s_blocksize_bits;
 798                 return offset;
 799         }
 800
 801         return 0;
 802
 803 }
 804 #else
 805 static inline int ocfs2_fast_symlink_chars(int blocksize)
 806 {
 807         return blocksize - offsetof(struct ocfs2_dinode, id2.i_symlink);
 808 }
 809
 810 static inline int ocfs2_max_inline_data(int blocksize)
 811 {
 812         return blocksize - offsetof(struct ocfs2_dinode, id2.i_data.id_data);
 813 }
 814
 815 static inline int ocfs2_extent_recs_per_inode(int blocksize)
 816 {
 817         int size;
 818
 819         size = blocksize -
 820                 offsetof(struct ocfs2_dinode, id2.i_list.l_recs);
 821
 822         return size / sizeof(struct ocfs2_extent_rec);
 823 }
 824
 825 static inline int ocfs2_chain_recs_per_inode(int blocksize)
 826 {
 827         int size;
 828
 829         size = blocksize -
 830                 offsetof(struct ocfs2_dinode, id2.i_chain.cl_recs);
 831
 832         return size / sizeof(struct ocfs2_chain_rec);
 833 }
 834
 835 static inline int ocfs2_extent_recs_per_eb(int blocksize)
 836 {
 837         int size;
 838
 839         size = blocksize -
 840                 offsetof(struct ocfs2_extent_block, h_list.l_recs);
 841
 842         return size / sizeof(struct ocfs2_extent_rec);
 843 }
 844
 845 static inline int ocfs2_local_alloc_size(int blocksize)
 846 {
 847         int size;
 848
 849         size = blocksize -
 850                 offsetof(struct ocfs2_dinode, id2.i_lab.la_bitmap);
 851
 852         return size;
 853 }
 854
 855 static inline int ocfs2_group_bitmap_size(int blocksize)
 856 {
 857         int size;
 858
 859         size = blocksize -
 860                 offsetof(struct ocfs2_group_desc, bg_bitmap);
 861
 862         return size;
 863 }
 864
 865 static inline int ocfs2_truncate_recs_per_inode(int blocksize)
 866 {
 867         int size;
 868
 869         size = blocksize -
 870                 offsetof(struct ocfs2_dinode, id2.i_dealloc.tl_recs);
 871
 872         return size / sizeof(struct ocfs2_truncate_rec);
 873 }
 874
 875 static inline uint64_t ocfs2_backup_super_blkno(int blocksize, int index)
 876 {
 877         uint64_t offset = OCFS2_BACKUP_SB_START;
 878
 879         if (index >= 0 && index < OCFS2_MAX_BACKUP_SUPERBLOCKS) {
 880                 offset <<= (2 * index);
 881                 offset /= blocksize;
 882                 return offset;
 883         }
 884
 885         return 0;
 886 }
 887 #endif  /* __KERNEL__ */
 888
 889
 890 static inline int ocfs2_system_inode_is_global(int type)
 891 {
 892         return ((type >= 0) &&
 893                 (type <= OCFS2_LAST_GLOBAL_SYSTEM_INODE));
 894 }
 895
 896 static inline int ocfs2_sprintf_system_inode_name(char *buf, int len,
 897                                                   int type, int slot)
 898 {
 899         int chars;
 900
 901         /*
 902          * Global system inodes can only have one copy.  Everything
 903          * after OCFS2_LAST_GLOBAL_SYSTEM_INODE in the system inode
 904          * list has a copy per slot.
 905          */
 906         if (type <= OCFS2_LAST_GLOBAL_SYSTEM_INODE)
 907                 chars = snprintf(buf, len, "%s",
 908                                  ocfs2_system_inodes[type].si_name);
 909         else
 910                 chars = snprintf(buf, len,
 911                                  ocfs2_system_inodes[type].si_name,
 912                                  slot);
 913
 914         return chars;
 915 }
 916
 917 static inline void ocfs2_set_de_type(struct ocfs2_dir_entry *de,
 918                                     umode_t mode)
 919 {
 920         de->file_type = ocfs2_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
 921 }
 922
 923 #endif  /* _OCFS2_FS_H */
 924