linux/fs/ocfs2/ocfs2_fs.h
<<
>>
Prefs
   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#define OCFS2_XATTR_BLOCK_SIGNATURE     "XATTR01"
  68#define OCFS2_DIR_TRAILER_SIGNATURE     "DIRTRL1"
  69#define OCFS2_DX_ROOT_SIGNATURE         "DXDIR01"
  70#define OCFS2_DX_LEAF_SIGNATURE         "DXLEAF1"
  71#define OCFS2_REFCOUNT_BLOCK_SIGNATURE  "REFCNT1"
  72
  73/* Compatibility flags */
  74#define OCFS2_HAS_COMPAT_FEATURE(sb,mask)                       \
  75        ( OCFS2_SB(sb)->s_feature_compat & (mask) )
  76#define OCFS2_HAS_RO_COMPAT_FEATURE(sb,mask)                    \
  77        ( OCFS2_SB(sb)->s_feature_ro_compat & (mask) )
  78#define OCFS2_HAS_INCOMPAT_FEATURE(sb,mask)                     \
  79        ( OCFS2_SB(sb)->s_feature_incompat & (mask) )
  80#define OCFS2_SET_COMPAT_FEATURE(sb,mask)                       \
  81        OCFS2_SB(sb)->s_feature_compat |= (mask)
  82#define OCFS2_SET_RO_COMPAT_FEATURE(sb,mask)                    \
  83        OCFS2_SB(sb)->s_feature_ro_compat |= (mask)
  84#define OCFS2_SET_INCOMPAT_FEATURE(sb,mask)                     \
  85        OCFS2_SB(sb)->s_feature_incompat |= (mask)
  86#define OCFS2_CLEAR_COMPAT_FEATURE(sb,mask)                     \
  87        OCFS2_SB(sb)->s_feature_compat &= ~(mask)
  88#define OCFS2_CLEAR_RO_COMPAT_FEATURE(sb,mask)                  \
  89        OCFS2_SB(sb)->s_feature_ro_compat &= ~(mask)
  90#define OCFS2_CLEAR_INCOMPAT_FEATURE(sb,mask)                   \
  91        OCFS2_SB(sb)->s_feature_incompat &= ~(mask)
  92
  93#define OCFS2_FEATURE_COMPAT_SUPP       (OCFS2_FEATURE_COMPAT_BACKUP_SB \
  94                                         | OCFS2_FEATURE_COMPAT_JBD2_SB)
  95#define OCFS2_FEATURE_INCOMPAT_SUPP     (OCFS2_FEATURE_INCOMPAT_LOCAL_MOUNT \
  96                                         | OCFS2_FEATURE_INCOMPAT_SPARSE_ALLOC \
  97                                         | OCFS2_FEATURE_INCOMPAT_INLINE_DATA \
  98                                         | OCFS2_FEATURE_INCOMPAT_EXTENDED_SLOT_MAP \
  99                                         | OCFS2_FEATURE_INCOMPAT_USERSPACE_STACK \
 100                                         | OCFS2_FEATURE_INCOMPAT_XATTR \
 101                                         | OCFS2_FEATURE_INCOMPAT_META_ECC \
 102                                         | OCFS2_FEATURE_INCOMPAT_INDEXED_DIRS \
 103                                         | OCFS2_FEATURE_INCOMPAT_REFCOUNT_TREE)
 104#define OCFS2_FEATURE_RO_COMPAT_SUPP    (OCFS2_FEATURE_RO_COMPAT_UNWRITTEN \
 105                                         | OCFS2_FEATURE_RO_COMPAT_USRQUOTA \
 106                                         | OCFS2_FEATURE_RO_COMPAT_GRPQUOTA)
 107
 108/*
 109 * Heartbeat-only devices are missing journals and other files.  The
 110 * filesystem driver can't load them, but the library can.  Never put
 111 * this in OCFS2_FEATURE_INCOMPAT_SUPP, *ever*.
 112 */
 113#define OCFS2_FEATURE_INCOMPAT_HEARTBEAT_DEV    0x0002
 114
 115/*
 116 * tunefs sets this incompat flag before starting the resize and clears it
 117 * at the end. This flag protects users from inadvertently mounting the fs
 118 * after an aborted run without fsck-ing.
 119 */
 120#define OCFS2_FEATURE_INCOMPAT_RESIZE_INPROG    0x0004
 121
 122/* Used to denote a non-clustered volume */
 123#define OCFS2_FEATURE_INCOMPAT_LOCAL_MOUNT      0x0008
 124
 125/* Support for sparse allocation in b-trees */
 126#define OCFS2_FEATURE_INCOMPAT_SPARSE_ALLOC     0x0010
 127
 128/*
 129 * Tunefs sets this incompat flag before starting an operation which
 130 * would require cleanup on abort. This is done to protect users from
 131 * inadvertently mounting the fs after an aborted run without
 132 * fsck-ing.
 133 *
 134 * s_tunefs_flags on the super block describes precisely which
 135 * operations were in progress.
 136 */
 137#define OCFS2_FEATURE_INCOMPAT_TUNEFS_INPROG    0x0020
 138
 139/* Support for data packed into inode blocks */
 140#define OCFS2_FEATURE_INCOMPAT_INLINE_DATA      0x0040
 141
 142/*
 143 * Support for alternate, userspace cluster stacks.  If set, the superblock
 144 * field s_cluster_info contains a tag for the alternate stack in use as
 145 * well as the name of the cluster being joined.
 146 * mount.ocfs2 must pass in a matching stack name.
 147 *
 148 * If not set, the classic stack will be used.  This is compatbile with
 149 * all older versions.
 150 */
 151#define OCFS2_FEATURE_INCOMPAT_USERSPACE_STACK  0x0080
 152
 153/* Support for the extended slot map */
 154#define OCFS2_FEATURE_INCOMPAT_EXTENDED_SLOT_MAP 0x100
 155
 156/* Support for extended attributes */
 157#define OCFS2_FEATURE_INCOMPAT_XATTR            0x0200
 158
 159/* Support for indexed directores */
 160#define OCFS2_FEATURE_INCOMPAT_INDEXED_DIRS     0x0400
 161
 162/* Metadata checksum and error correction */
 163#define OCFS2_FEATURE_INCOMPAT_META_ECC         0x0800
 164
 165/* Refcount tree support */
 166#define OCFS2_FEATURE_INCOMPAT_REFCOUNT_TREE    0x1000
 167
 168/*
 169 * backup superblock flag is used to indicate that this volume
 170 * has backup superblocks.
 171 */
 172#define OCFS2_FEATURE_COMPAT_BACKUP_SB          0x0001
 173
 174/*
 175 * The filesystem will correctly handle journal feature bits.
 176 */
 177#define OCFS2_FEATURE_COMPAT_JBD2_SB            0x0002
 178
 179/*
 180 * Unwritten extents support.
 181 */
 182#define OCFS2_FEATURE_RO_COMPAT_UNWRITTEN       0x0001
 183
 184/*
 185 * Maintain quota information for this filesystem
 186 */
 187#define OCFS2_FEATURE_RO_COMPAT_USRQUOTA        0x0002
 188#define OCFS2_FEATURE_RO_COMPAT_GRPQUOTA        0x0004
 189
 190/* The byte offset of the first backup block will be 1G.
 191 * The following will be 4G, 16G, 64G, 256G and 1T.
 192 */
 193#define OCFS2_BACKUP_SB_START                   1 << 30
 194
 195/* the max backup superblock nums */
 196#define OCFS2_MAX_BACKUP_SUPERBLOCKS    6
 197
 198/*
 199 * Flags on ocfs2_super_block.s_tunefs_flags
 200 */
 201#define OCFS2_TUNEFS_INPROG_REMOVE_SLOT         0x0001  /* Removing slots */
 202
 203/*
 204 * Flags on ocfs2_dinode.i_flags
 205 */
 206#define OCFS2_VALID_FL          (0x00000001)    /* Inode is valid */
 207#define OCFS2_UNUSED2_FL        (0x00000002)
 208#define OCFS2_ORPHANED_FL       (0x00000004)    /* On the orphan list */
 209#define OCFS2_UNUSED3_FL        (0x00000008)
 210/* System inode flags */
 211#define OCFS2_SYSTEM_FL         (0x00000010)    /* System inode */
 212#define OCFS2_SUPER_BLOCK_FL    (0x00000020)    /* Super block */
 213#define OCFS2_LOCAL_ALLOC_FL    (0x00000040)    /* Slot local alloc bitmap */
 214#define OCFS2_BITMAP_FL         (0x00000080)    /* Allocation bitmap */
 215#define OCFS2_JOURNAL_FL        (0x00000100)    /* Slot local journal */
 216#define OCFS2_HEARTBEAT_FL      (0x00000200)    /* Heartbeat area */
 217#define OCFS2_CHAIN_FL          (0x00000400)    /* Chain allocator */
 218#define OCFS2_DEALLOC_FL        (0x00000800)    /* Truncate log */
 219#define OCFS2_QUOTA_FL          (0x00001000)    /* Quota file */
 220
 221/*
 222 * Flags on ocfs2_dinode.i_dyn_features
 223 *
 224 * These can change much more often than i_flags. When adding flags,
 225 * keep in mind that i_dyn_features is only 16 bits wide.
 226 */
 227#define OCFS2_INLINE_DATA_FL    (0x0001)        /* Data stored in inode block */
 228#define OCFS2_HAS_XATTR_FL      (0x0002)
 229#define OCFS2_INLINE_XATTR_FL   (0x0004)
 230#define OCFS2_INDEXED_DIR_FL    (0x0008)
 231#define OCFS2_HAS_REFCOUNT_FL   (0x0010)
 232
 233/* Inode attributes, keep in sync with EXT2 */
 234#define OCFS2_SECRM_FL          (0x00000001)    /* Secure deletion */
 235#define OCFS2_UNRM_FL           (0x00000002)    /* Undelete */
 236#define OCFS2_COMPR_FL          (0x00000004)    /* Compress file */
 237#define OCFS2_SYNC_FL           (0x00000008)    /* Synchronous updates */
 238#define OCFS2_IMMUTABLE_FL      (0x00000010)    /* Immutable file */
 239#define OCFS2_APPEND_FL         (0x00000020)    /* writes to file may only append */
 240#define OCFS2_NODUMP_FL         (0x00000040)    /* do not dump file */
 241#define OCFS2_NOATIME_FL        (0x00000080)    /* do not update atime */
 242#define OCFS2_DIRSYNC_FL        (0x00010000)    /* dirsync behaviour (directories only) */
 243
 244#define OCFS2_FL_VISIBLE        (0x000100FF)    /* User visible flags */
 245#define OCFS2_FL_MODIFIABLE     (0x000100FF)    /* User modifiable flags */
 246
 247/*
 248 * Extent record flags (e_node.leaf.flags)
 249 */
 250#define OCFS2_EXT_UNWRITTEN             (0x01)  /* Extent is allocated but
 251                                                 * unwritten */
 252#define OCFS2_EXT_REFCOUNTED            (0x02)  /* Extent is reference
 253                                                 * counted in an associated
 254                                                 * refcount tree */
 255
 256/*
 257 * ioctl commands
 258 */
 259#define OCFS2_IOC_GETFLAGS      _IOR('f', 1, long)
 260#define OCFS2_IOC_SETFLAGS      _IOW('f', 2, long)
 261#define OCFS2_IOC32_GETFLAGS    _IOR('f', 1, int)
 262#define OCFS2_IOC32_SETFLAGS    _IOW('f', 2, int)
 263
 264/*
 265 * Space reservation / allocation / free ioctls and argument structure
 266 * are designed to be compatible with XFS.
 267 *
 268 * ALLOCSP* and FREESP* are not and will never be supported, but are
 269 * included here for completeness.
 270 */
 271struct ocfs2_space_resv {
 272        __s16           l_type;
 273        __s16           l_whence;
 274        __s64           l_start;
 275        __s64           l_len;          /* len == 0 means until end of file */
 276        __s32           l_sysid;
 277        __u32           l_pid;
 278        __s32           l_pad[4];       /* reserve area                     */
 279};
 280
 281#define OCFS2_IOC_ALLOCSP               _IOW ('X', 10, struct ocfs2_space_resv)
 282#define OCFS2_IOC_FREESP                _IOW ('X', 11, struct ocfs2_space_resv)
 283#define OCFS2_IOC_RESVSP                _IOW ('X', 40, struct ocfs2_space_resv)
 284#define OCFS2_IOC_UNRESVSP      _IOW ('X', 41, struct ocfs2_space_resv)
 285#define OCFS2_IOC_ALLOCSP64     _IOW ('X', 36, struct ocfs2_space_resv)
 286#define OCFS2_IOC_FREESP64      _IOW ('X', 37, struct ocfs2_space_resv)
 287#define OCFS2_IOC_RESVSP64      _IOW ('X', 42, struct ocfs2_space_resv)
 288#define OCFS2_IOC_UNRESVSP64    _IOW ('X', 43, struct ocfs2_space_resv)
 289
 290/* Used to pass group descriptor data when online resize is done */
 291struct ocfs2_new_group_input {
 292        __u64 group;            /* Group descriptor's blkno. */
 293        __u32 clusters;         /* Total number of clusters in this group */
 294        __u32 frees;            /* Total free clusters in this group */
 295        __u16 chain;            /* Chain for this group */
 296        __u16 reserved1;
 297        __u32 reserved2;
 298};
 299
 300#define OCFS2_IOC_GROUP_EXTEND  _IOW('o', 1, int)
 301#define OCFS2_IOC_GROUP_ADD     _IOW('o', 2,struct ocfs2_new_group_input)
 302#define OCFS2_IOC_GROUP_ADD64   _IOW('o', 3,struct ocfs2_new_group_input)
 303
 304/* Used to pass 2 file names to reflink. */
 305struct reflink_arguments {
 306        __u64 old_path;
 307        __u64 new_path;
 308        __u64 preserve;
 309};
 310#define OCFS2_IOC_REFLINK       _IOW('o', 4, struct reflink_arguments)
 311
 312
 313/*
 314 * Journal Flags (ocfs2_dinode.id1.journal1.i_flags)
 315 */
 316#define OCFS2_JOURNAL_DIRTY_FL  (0x00000001)    /* Journal needs recovery */
 317
 318/*
 319 * superblock s_state flags
 320 */
 321#define OCFS2_ERROR_FS          (0x00000001)    /* FS saw errors */
 322
 323/* Limit of space in ocfs2_dir_entry */
 324#define OCFS2_MAX_FILENAME_LEN          255
 325
 326/* Maximum slots on an ocfs2 file system */
 327#define OCFS2_MAX_SLOTS                 255
 328
 329/* Slot map indicator for an empty slot */
 330#define OCFS2_INVALID_SLOT              -1
 331
 332#define OCFS2_VOL_UUID_LEN              16
 333#define OCFS2_MAX_VOL_LABEL_LEN         64
 334
 335/* The alternate, userspace stack fields */
 336#define OCFS2_STACK_LABEL_LEN           4
 337#define OCFS2_CLUSTER_NAME_LEN          16
 338
 339/* Journal limits (in bytes) */
 340#define OCFS2_MIN_JOURNAL_SIZE          (4 * 1024 * 1024)
 341
 342/*
 343 * Default local alloc size (in megabytes)
 344 *
 345 * The value chosen should be such that most allocations, including new
 346 * block groups, use local alloc.
 347 */
 348#define OCFS2_DEFAULT_LOCAL_ALLOC_SIZE  8
 349
 350/*
 351 * Inline extended attribute size (in bytes)
 352 * The value chosen should be aligned to 16 byte boundaries.
 353 */
 354#define OCFS2_MIN_XATTR_INLINE_SIZE     256
 355
 356struct ocfs2_system_inode_info {
 357        char    *si_name;
 358        int     si_iflags;
 359        int     si_mode;
 360};
 361
 362/* System file index */
 363enum {
 364        BAD_BLOCK_SYSTEM_INODE = 0,
 365        GLOBAL_INODE_ALLOC_SYSTEM_INODE,
 366        SLOT_MAP_SYSTEM_INODE,
 367#define OCFS2_FIRST_ONLINE_SYSTEM_INODE SLOT_MAP_SYSTEM_INODE
 368        HEARTBEAT_SYSTEM_INODE,
 369        GLOBAL_BITMAP_SYSTEM_INODE,
 370        USER_QUOTA_SYSTEM_INODE,
 371        GROUP_QUOTA_SYSTEM_INODE,
 372#define OCFS2_LAST_GLOBAL_SYSTEM_INODE GROUP_QUOTA_SYSTEM_INODE
 373        ORPHAN_DIR_SYSTEM_INODE,
 374        EXTENT_ALLOC_SYSTEM_INODE,
 375        INODE_ALLOC_SYSTEM_INODE,
 376        JOURNAL_SYSTEM_INODE,
 377        LOCAL_ALLOC_SYSTEM_INODE,
 378        TRUNCATE_LOG_SYSTEM_INODE,
 379        LOCAL_USER_QUOTA_SYSTEM_INODE,
 380        LOCAL_GROUP_QUOTA_SYSTEM_INODE,
 381        NUM_SYSTEM_INODES
 382};
 383
 384static struct ocfs2_system_inode_info ocfs2_system_inodes[NUM_SYSTEM_INODES] = {
 385        /* Global system inodes (single copy) */
 386        /* The first two are only used from userspace mfks/tunefs */
 387        [BAD_BLOCK_SYSTEM_INODE]                = { "bad_blocks", 0, S_IFREG | 0644 },
 388        [GLOBAL_INODE_ALLOC_SYSTEM_INODE]       = { "global_inode_alloc", OCFS2_BITMAP_FL | OCFS2_CHAIN_FL, S_IFREG | 0644 },
 389
 390        /* These are used by the running filesystem */
 391        [SLOT_MAP_SYSTEM_INODE]                 = { "slot_map", 0, S_IFREG | 0644 },
 392        [HEARTBEAT_SYSTEM_INODE]                = { "heartbeat", OCFS2_HEARTBEAT_FL, S_IFREG | 0644 },
 393        [GLOBAL_BITMAP_SYSTEM_INODE]            = { "global_bitmap", 0, S_IFREG | 0644 },
 394        [USER_QUOTA_SYSTEM_INODE]               = { "aquota.user", OCFS2_QUOTA_FL, S_IFREG | 0644 },
 395        [GROUP_QUOTA_SYSTEM_INODE]              = { "aquota.group", OCFS2_QUOTA_FL, S_IFREG | 0644 },
 396
 397        /* Slot-specific system inodes (one copy per slot) */
 398        [ORPHAN_DIR_SYSTEM_INODE]               = { "orphan_dir:%04d", 0, S_IFDIR | 0755 },
 399        [EXTENT_ALLOC_SYSTEM_INODE]             = { "extent_alloc:%04d", OCFS2_BITMAP_FL | OCFS2_CHAIN_FL, S_IFREG | 0644 },
 400        [INODE_ALLOC_SYSTEM_INODE]              = { "inode_alloc:%04d", OCFS2_BITMAP_FL | OCFS2_CHAIN_FL, S_IFREG | 0644 },
 401        [JOURNAL_SYSTEM_INODE]                  = { "journal:%04d", OCFS2_JOURNAL_FL, S_IFREG | 0644 },
 402        [LOCAL_ALLOC_SYSTEM_INODE]              = { "local_alloc:%04d", OCFS2_BITMAP_FL | OCFS2_LOCAL_ALLOC_FL, S_IFREG | 0644 },
 403        [TRUNCATE_LOG_SYSTEM_INODE]             = { "truncate_log:%04d", OCFS2_DEALLOC_FL, S_IFREG | 0644 },
 404        [LOCAL_USER_QUOTA_SYSTEM_INODE]         = { "aquota.user:%04d", OCFS2_QUOTA_FL, S_IFREG | 0644 },
 405        [LOCAL_GROUP_QUOTA_SYSTEM_INODE]        = { "aquota.group:%04d", OCFS2_QUOTA_FL, S_IFREG | 0644 },
 406};
 407
 408/* Parameter passed from mount.ocfs2 to module */
 409#define OCFS2_HB_NONE                   "heartbeat=none"
 410#define OCFS2_HB_LOCAL                  "heartbeat=local"
 411
 412/*
 413 * OCFS2 directory file types.  Only the low 3 bits are used.  The
 414 * other bits are reserved for now.
 415 */
 416#define OCFS2_FT_UNKNOWN        0
 417#define OCFS2_FT_REG_FILE       1
 418#define OCFS2_FT_DIR            2
 419#define OCFS2_FT_CHRDEV         3
 420#define OCFS2_FT_BLKDEV         4
 421#define OCFS2_FT_FIFO           5
 422#define OCFS2_FT_SOCK           6
 423#define OCFS2_FT_SYMLINK        7
 424
 425#define OCFS2_FT_MAX            8
 426
 427/*
 428 * OCFS2_DIR_PAD defines the directory entries boundaries
 429 *
 430 * NOTE: It must be a multiple of 4
 431 */
 432#define OCFS2_DIR_PAD                   4
 433#define OCFS2_DIR_ROUND                 (OCFS2_DIR_PAD - 1)
 434#define OCFS2_DIR_MEMBER_LEN            offsetof(struct ocfs2_dir_entry, name)
 435#define OCFS2_DIR_REC_LEN(name_len)     (((name_len) + OCFS2_DIR_MEMBER_LEN + \
 436                                          OCFS2_DIR_ROUND) & \
 437                                         ~OCFS2_DIR_ROUND)
 438#define OCFS2_DIR_MIN_REC_LEN   OCFS2_DIR_REC_LEN(1)
 439
 440#define OCFS2_LINK_MAX          32000
 441#define OCFS2_DX_LINK_MAX       ((1U << 31) - 1U)
 442#define OCFS2_LINKS_HI_SHIFT    16
 443#define OCFS2_DX_ENTRIES_MAX    (0xffffffffU)
 444
 445#define S_SHIFT                 12
 446static unsigned char ocfs2_type_by_mode[S_IFMT >> S_SHIFT] = {
 447        [S_IFREG >> S_SHIFT]  = OCFS2_FT_REG_FILE,
 448        [S_IFDIR >> S_SHIFT]  = OCFS2_FT_DIR,
 449        [S_IFCHR >> S_SHIFT]  = OCFS2_FT_CHRDEV,
 450        [S_IFBLK >> S_SHIFT]  = OCFS2_FT_BLKDEV,
 451        [S_IFIFO >> S_SHIFT]  = OCFS2_FT_FIFO,
 452        [S_IFSOCK >> S_SHIFT] = OCFS2_FT_SOCK,
 453        [S_IFLNK >> S_SHIFT]  = OCFS2_FT_SYMLINK,
 454};
 455
 456
 457/*
 458 * Convenience casts
 459 */
 460#define OCFS2_RAW_SB(dinode)            (&((dinode)->id2.i_super))
 461
 462/*
 463 * Block checking structure.  This is used in metadata to validate the
 464 * contents.  If OCFS2_FEATURE_INCOMPAT_META_ECC is not set, it is all
 465 * zeros.
 466 */
 467struct ocfs2_block_check {
 468/*00*/  __le32 bc_crc32e;       /* 802.3 Ethernet II CRC32 */
 469        __le16 bc_ecc;          /* Single-error-correction parity vector.
 470                                   This is a simple Hamming code dependant
 471                                   on the blocksize.  OCFS2's maximum
 472                                   blocksize, 4K, requires 16 parity bits,
 473                                   so we fit in __le16. */
 474        __le16 bc_reserved1;
 475/*08*/
 476};
 477
 478/*
 479 * On disk extent record for OCFS2
 480 * It describes a range of clusters on disk.
 481 *
 482 * Length fields are divided into interior and leaf node versions.
 483 * This leaves room for a flags field (OCFS2_EXT_*) in the leaf nodes.
 484 */
 485struct ocfs2_extent_rec {
 486/*00*/  __le32 e_cpos;          /* Offset into the file, in clusters */
 487        union {
 488                __le32 e_int_clusters; /* Clusters covered by all children */
 489                struct {
 490                        __le16 e_leaf_clusters; /* Clusters covered by this
 491                                                   extent */
 492                        __u8 e_reserved1;
 493                        __u8 e_flags; /* Extent flags */
 494                };
 495        };
 496        __le64 e_blkno;         /* Physical disk offset, in blocks */
 497/*10*/
 498};
 499
 500struct ocfs2_chain_rec {
 501        __le32 c_free;  /* Number of free bits in this chain. */
 502        __le32 c_total; /* Number of total bits in this chain */
 503        __le64 c_blkno; /* Physical disk offset (blocks) of 1st group */
 504};
 505
 506struct ocfs2_truncate_rec {
 507        __le32 t_start;         /* 1st cluster in this log */
 508        __le32 t_clusters;      /* Number of total clusters covered */
 509};
 510
 511/*
 512 * On disk extent list for OCFS2 (node in the tree).  Note that this
 513 * is contained inside ocfs2_dinode or ocfs2_extent_block, so the
 514 * offsets are relative to ocfs2_dinode.id2.i_list or
 515 * ocfs2_extent_block.h_list, respectively.
 516 */
 517struct ocfs2_extent_list {
 518/*00*/  __le16 l_tree_depth;            /* Extent tree depth from this
 519                                           point.  0 means data extents
 520                                           hang directly off this
 521                                           header (a leaf)
 522                                           NOTE: The high 8 bits cannot be
 523                                           used - tree_depth is never that big.
 524                                        */
 525        __le16 l_count;                 /* Number of extent records */
 526        __le16 l_next_free_rec;         /* Next unused extent slot */
 527        __le16 l_reserved1;
 528        __le64 l_reserved2;             /* Pad to
 529                                           sizeof(ocfs2_extent_rec) */
 530/*10*/  struct ocfs2_extent_rec l_recs[0];      /* Extent records */
 531};
 532
 533/*
 534 * On disk allocation chain list for OCFS2.  Note that this is
 535 * contained inside ocfs2_dinode, so the offsets are relative to
 536 * ocfs2_dinode.id2.i_chain.
 537 */
 538struct ocfs2_chain_list {
 539/*00*/  __le16 cl_cpg;                  /* Clusters per Block Group */
 540        __le16 cl_bpc;                  /* Bits per cluster */
 541        __le16 cl_count;                /* Total chains in this list */
 542        __le16 cl_next_free_rec;        /* Next unused chain slot */
 543        __le64 cl_reserved1;
 544/*10*/  struct ocfs2_chain_rec cl_recs[0];      /* Chain records */
 545};
 546
 547/*
 548 * On disk deallocation log for OCFS2.  Note that this is
 549 * contained inside ocfs2_dinode, so the offsets are relative to
 550 * ocfs2_dinode.id2.i_dealloc.
 551 */
 552struct ocfs2_truncate_log {
 553/*00*/  __le16 tl_count;                /* Total records in this log */
 554        __le16 tl_used;                 /* Number of records in use */
 555        __le32 tl_reserved1;
 556/*08*/  struct ocfs2_truncate_rec tl_recs[0];   /* Truncate records */
 557};
 558
 559/*
 560 * On disk extent block (indirect block) for OCFS2
 561 */
 562struct ocfs2_extent_block
 563{
 564/*00*/  __u8 h_signature[8];            /* Signature for verification */
 565        struct ocfs2_block_check h_check;       /* Error checking */
 566/*10*/  __le16 h_suballoc_slot;         /* Slot suballocator this
 567                                           extent_header belongs to */
 568        __le16 h_suballoc_bit;          /* Bit offset in suballocator
 569                                           block group */
 570        __le32 h_fs_generation;         /* Must match super block */
 571        __le64 h_blkno;                 /* Offset on disk, in blocks */
 572/*20*/  __le64 h_reserved3;
 573        __le64 h_next_leaf_blk;         /* Offset on disk, in blocks,
 574                                           of next leaf header pointing
 575                                           to data */
 576/*30*/  struct ocfs2_extent_list h_list;        /* Extent record list */
 577/* Actual on-disk size is one block */
 578};
 579
 580/*
 581 * On disk slot map for OCFS2.  This defines the contents of the "slot_map"
 582 * system file.  A slot is valid if it contains a node number >= 0.  The
 583 * value -1 (0xFFFF) is OCFS2_INVALID_SLOT.  This marks a slot empty.
 584 */
 585struct ocfs2_slot_map {
 586/*00*/  __le16 sm_slots[0];
 587/*
 588 * Actual on-disk size is one block.  OCFS2_MAX_SLOTS is 255,
 589 * 255 * sizeof(__le16) == 512B, within the 512B block minimum blocksize.
 590 */
 591};
 592
 593struct ocfs2_extended_slot {
 594/*00*/  __u8    es_valid;
 595        __u8    es_reserved1[3];
 596        __le32  es_node_num;
 597/*10*/
 598};
 599
 600/*
 601 * The extended slot map, used when OCFS2_FEATURE_INCOMPAT_EXTENDED_SLOT_MAP
 602 * is set.  It separates out the valid marker from the node number, and
 603 * has room to grow.  Unlike the old slot map, this format is defined by
 604 * i_size.
 605 */
 606struct ocfs2_slot_map_extended {
 607/*00*/  struct ocfs2_extended_slot se_slots[0];
 608/*
 609 * Actual size is i_size of the slot_map system file.  It should
 610 * match s_max_slots * sizeof(struct ocfs2_extended_slot)
 611 */
 612};
 613
 614struct ocfs2_cluster_info {
 615/*00*/  __u8   ci_stack[OCFS2_STACK_LABEL_LEN];
 616        __le32 ci_reserved;
 617/*08*/  __u8   ci_cluster[OCFS2_CLUSTER_NAME_LEN];
 618/*18*/
 619};
 620
 621/*
 622 * On disk superblock for OCFS2
 623 * Note that it is contained inside an ocfs2_dinode, so all offsets
 624 * are relative to the start of ocfs2_dinode.id2.
 625 */
 626struct ocfs2_super_block {
 627/*00*/  __le16 s_major_rev_level;
 628        __le16 s_minor_rev_level;
 629        __le16 s_mnt_count;
 630        __le16 s_max_mnt_count;
 631        __le16 s_state;                 /* File system state */
 632        __le16 s_errors;                        /* Behaviour when detecting errors */
 633        __le32 s_checkinterval;         /* Max time between checks */
 634/*10*/  __le64 s_lastcheck;             /* Time of last check */
 635        __le32 s_creator_os;            /* OS */
 636        __le32 s_feature_compat;                /* Compatible feature set */
 637/*20*/  __le32 s_feature_incompat;      /* Incompatible feature set */
 638        __le32 s_feature_ro_compat;     /* Readonly-compatible feature set */
 639        __le64 s_root_blkno;            /* Offset, in blocks, of root directory
 640                                           dinode */
 641/*30*/  __le64 s_system_dir_blkno;      /* Offset, in blocks, of system
 642                                           directory dinode */
 643        __le32 s_blocksize_bits;                /* Blocksize for this fs */
 644        __le32 s_clustersize_bits;      /* Clustersize for this fs */
 645/*40*/  __le16 s_max_slots;             /* Max number of simultaneous mounts
 646                                           before tunefs required */
 647        __le16 s_tunefs_flag;
 648        __le32 s_uuid_hash;             /* hash value of uuid */
 649        __le64 s_first_cluster_group;   /* Block offset of 1st cluster
 650                                         * group header */
 651/*50*/  __u8  s_label[OCFS2_MAX_VOL_LABEL_LEN]; /* Label for mounting, etc. */
 652/*90*/  __u8  s_uuid[OCFS2_VOL_UUID_LEN];       /* 128-bit uuid */
 653/*A0*/  struct ocfs2_cluster_info s_cluster_info; /* Selected userspace
 654                                                     stack.  Only valid
 655                                                     with INCOMPAT flag. */
 656/*B8*/  __le16 s_xattr_inline_size;     /* extended attribute inline size
 657                                           for this fs*/
 658        __le16 s_reserved0;
 659        __le32 s_dx_seed[3];            /* seed[0-2] for dx dir hash.
 660                                         * s_uuid_hash serves as seed[3]. */
 661/*C0*/  __le64 s_reserved2[15];         /* Fill out superblock */
 662/*140*/
 663
 664        /*
 665         * NOTE: As stated above, all offsets are relative to
 666         * ocfs2_dinode.id2, which is at 0xC0 in the inode.
 667         * 0xC0 + 0x140 = 0x200 or 512 bytes.  A superblock must fit within
 668         * our smallest blocksize, which is 512 bytes.  To ensure this,
 669         * we reserve the space in s_reserved2.  Anything past s_reserved2
 670         * will not be available on the smallest blocksize.
 671         */
 672};
 673
 674/*
 675 * Local allocation bitmap for OCFS2 slots
 676 * Note that it exists inside an ocfs2_dinode, so all offsets are
 677 * relative to the start of ocfs2_dinode.id2.
 678 */
 679struct ocfs2_local_alloc
 680{
 681/*00*/  __le32 la_bm_off;       /* Starting bit offset in main bitmap */
 682        __le16 la_size;         /* Size of included bitmap, in bytes */
 683        __le16 la_reserved1;
 684        __le64 la_reserved2;
 685/*10*/  __u8   la_bitmap[0];
 686};
 687
 688/*
 689 * Data-in-inode header. This is only used if i_dyn_features has
 690 * OCFS2_INLINE_DATA_FL set.
 691 */
 692struct ocfs2_inline_data
 693{
 694/*00*/  __le16  id_count;       /* Number of bytes that can be used
 695                                 * for data, starting at id_data */
 696        __le16  id_reserved0;
 697        __le32  id_reserved1;
 698        __u8    id_data[0];     /* Start of user data */
 699};
 700
 701/*
 702 * On disk inode for OCFS2
 703 */
 704struct ocfs2_dinode {
 705/*00*/  __u8 i_signature[8];            /* Signature for validation */
 706        __le32 i_generation;            /* Generation number */
 707        __le16 i_suballoc_slot;         /* Slot suballocator this inode
 708                                           belongs to */
 709        __le16 i_suballoc_bit;          /* Bit offset in suballocator
 710                                           block group */
 711/*10*/  __le16 i_links_count_hi;        /* High 16 bits of links count */
 712        __le16 i_xattr_inline_size;
 713        __le32 i_clusters;              /* Cluster count */
 714        __le32 i_uid;                   /* Owner UID */
 715        __le32 i_gid;                   /* Owning GID */
 716/*20*/  __le64 i_size;                  /* Size in bytes */
 717        __le16 i_mode;                  /* File mode */
 718        __le16 i_links_count;           /* Links count */
 719        __le32 i_flags;                 /* File flags */
 720/*30*/  __le64 i_atime;                 /* Access time */
 721        __le64 i_ctime;                 /* Creation time */
 722/*40*/  __le64 i_mtime;                 /* Modification time */
 723        __le64 i_dtime;                 /* Deletion time */
 724/*50*/  __le64 i_blkno;                 /* Offset on disk, in blocks */
 725        __le64 i_last_eb_blk;           /* Pointer to last extent
 726                                           block */
 727/*60*/  __le32 i_fs_generation;         /* Generation per fs-instance */
 728        __le32 i_atime_nsec;
 729        __le32 i_ctime_nsec;
 730        __le32 i_mtime_nsec;
 731/*70*/  __le32 i_attr;
 732        __le16 i_orphaned_slot;         /* Only valid when OCFS2_ORPHANED_FL
 733                                           was set in i_flags */
 734        __le16 i_dyn_features;
 735        __le64 i_xattr_loc;
 736/*80*/  struct ocfs2_block_check i_check;       /* Error checking */
 737/*88*/  __le64 i_dx_root;               /* Pointer to dir index root block */
 738/*90*/  __le64 i_refcount_loc;
 739        __le64 i_reserved2[4];
 740/*B8*/  union {
 741                __le64 i_pad1;          /* Generic way to refer to this
 742                                           64bit union */
 743                struct {
 744                        __le64 i_rdev;  /* Device number */
 745                } dev1;
 746                struct {                /* Info for bitmap system
 747                                           inodes */
 748                        __le32 i_used;  /* Bits (ie, clusters) used  */
 749                        __le32 i_total; /* Total bits (clusters)
 750                                           available */
 751                } bitmap1;
 752                struct {                /* Info for journal system
 753                                           inodes */
 754                        __le32 ij_flags;        /* Mounted, version, etc. */
 755                        __le32 ij_recovery_generation; /* Incremented when the
 756                                                          journal is recovered
 757                                                          after an unclean
 758                                                          shutdown */
 759                } journal1;
 760        } id1;                          /* Inode type dependant 1 */
 761/*C0*/  union {
 762                struct ocfs2_super_block        i_super;
 763                struct ocfs2_local_alloc        i_lab;
 764                struct ocfs2_chain_list         i_chain;
 765                struct ocfs2_extent_list        i_list;
 766                struct ocfs2_truncate_log       i_dealloc;
 767                struct ocfs2_inline_data        i_data;
 768                __u8                            i_symlink[0];
 769        } id2;
 770/* Actual on-disk size is one block */
 771};
 772
 773/*
 774 * On-disk directory entry structure for OCFS2
 775 *
 776 * Packed as this structure could be accessed unaligned on 64-bit platforms
 777 */
 778struct ocfs2_dir_entry {
 779/*00*/  __le64   inode;                  /* Inode number */
 780        __le16   rec_len;                /* Directory entry length */
 781        __u8    name_len;               /* Name length */
 782        __u8    file_type;
 783/*0C*/  char    name[OCFS2_MAX_FILENAME_LEN];   /* File name */
 784/* Actual on-disk length specified by rec_len */
 785} __attribute__ ((packed));
 786
 787/*
 788 * Per-block record for the unindexed directory btree. This is carefully
 789 * crafted so that the rec_len and name_len records of an ocfs2_dir_entry are
 790 * mirrored. That way, the directory manipulation code needs a minimal amount
 791 * of update.
 792 *
 793 * NOTE: Keep this structure aligned to a multiple of 4 bytes.
 794 */
 795struct ocfs2_dir_block_trailer {
 796/*00*/  __le64          db_compat_inode;        /* Always zero. Was inode */
 797
 798        __le16          db_compat_rec_len;      /* Backwards compatible with
 799                                                 * ocfs2_dir_entry. */
 800        __u8            db_compat_name_len;     /* Always zero. Was name_len */
 801        __u8            db_reserved0;
 802        __le16          db_reserved1;
 803        __le16          db_free_rec_len;        /* Size of largest empty hole
 804                                                 * in this block. (unused) */
 805/*10*/  __u8            db_signature[8];        /* Signature for verification */
 806        __le64          db_reserved2;
 807        __le64          db_free_next;           /* Next block in list (unused) */
 808/*20*/  __le64          db_blkno;               /* Offset on disk, in blocks */
 809        __le64          db_parent_dinode;       /* dinode which owns me, in
 810                                                   blocks */
 811/*30*/  struct ocfs2_block_check db_check;      /* Error checking */
 812/*40*/
 813};
 814
 815 /*
 816 * A directory entry in the indexed tree. We don't store the full name here,
 817 * but instead provide a pointer to the full dirent in the unindexed tree.
 818 *
 819 * We also store name_len here so as to reduce the number of leaf blocks we
 820 * need to search in case of collisions.
 821 */
 822struct ocfs2_dx_entry {
 823        __le32          dx_major_hash;  /* Used to find logical
 824                                         * cluster in index */
 825        __le32          dx_minor_hash;  /* Lower bits used to find
 826                                         * block in cluster */
 827        __le64          dx_dirent_blk;  /* Physical block in unindexed
 828                                         * tree holding this dirent. */
 829};
 830
 831struct ocfs2_dx_entry_list {
 832        __le32          de_reserved;
 833        __le16          de_count;       /* Maximum number of entries
 834                                         * possible in de_entries */
 835        __le16          de_num_used;    /* Current number of
 836                                         * de_entries entries */
 837        struct  ocfs2_dx_entry          de_entries[0];  /* Indexed dir entries
 838                                                         * in a packed array of
 839                                                         * length de_num_used */
 840};
 841
 842#define OCFS2_DX_FLAG_INLINE    0x01
 843
 844/*
 845 * A directory indexing block. Each indexed directory has one of these,
 846 * pointed to by ocfs2_dinode.
 847 *
 848 * This block stores an indexed btree root, and a set of free space
 849 * start-of-list pointers.
 850 */
 851struct ocfs2_dx_root_block {
 852        __u8            dr_signature[8];        /* Signature for verification */
 853        struct ocfs2_block_check dr_check;      /* Error checking */
 854        __le16          dr_suballoc_slot;       /* Slot suballocator this
 855                                                 * block belongs to. */
 856        __le16          dr_suballoc_bit;        /* Bit offset in suballocator
 857                                                 * block group */
 858        __le32          dr_fs_generation;       /* Must match super block */
 859        __le64          dr_blkno;               /* Offset on disk, in blocks */
 860        __le64          dr_last_eb_blk;         /* Pointer to last
 861                                                 * extent block */
 862        __le32          dr_clusters;            /* Clusters allocated
 863                                                 * to the indexed tree. */
 864        __u8            dr_flags;               /* OCFS2_DX_FLAG_* flags */
 865        __u8            dr_reserved0;
 866        __le16          dr_reserved1;
 867        __le64          dr_dir_blkno;           /* Pointer to parent inode */
 868        __le32          dr_num_entries;         /* Total number of
 869                                                 * names stored in
 870                                                 * this directory.*/
 871        __le32          dr_reserved2;
 872        __le64          dr_free_blk;            /* Pointer to head of free
 873                                                 * unindexed block list. */
 874        __le64          dr_reserved3[15];
 875        union {
 876                struct ocfs2_extent_list dr_list; /* Keep this aligned to 128
 877                                                   * bits for maximum space
 878                                                   * efficiency. */
 879                struct ocfs2_dx_entry_list dr_entries; /* In-root-block list of
 880                                                        * entries. We grow out
 881                                                        * to extents if this
 882                                                        * gets too big. */
 883        };
 884};
 885
 886/*
 887 * The header of a leaf block in the indexed tree.
 888 */
 889struct ocfs2_dx_leaf {
 890        __u8            dl_signature[8];/* Signature for verification */
 891        struct ocfs2_block_check dl_check;      /* Error checking */
 892        __le64          dl_blkno;       /* Offset on disk, in blocks */
 893        __le32          dl_fs_generation;/* Must match super block */
 894        __le32          dl_reserved0;
 895        __le64          dl_reserved1;
 896        struct ocfs2_dx_entry_list      dl_list;
 897};
 898
 899/*
 900 * On disk allocator group structure for OCFS2
 901 */
 902struct ocfs2_group_desc
 903{
 904/*00*/  __u8    bg_signature[8];        /* Signature for validation */
 905        __le16   bg_size;                /* Size of included bitmap in
 906                                           bytes. */
 907        __le16   bg_bits;                /* Bits represented by this
 908                                           group. */
 909        __le16  bg_free_bits_count;     /* Free bits count */
 910        __le16   bg_chain;               /* What chain I am in. */
 911/*10*/  __le32   bg_generation;
 912        __le32  bg_reserved1;
 913        __le64   bg_next_group;          /* Next group in my list, in
 914                                           blocks */
 915/*20*/  __le64   bg_parent_dinode;       /* dinode which owns me, in
 916                                           blocks */
 917        __le64   bg_blkno;               /* Offset on disk, in blocks */
 918/*30*/  struct ocfs2_block_check bg_check;      /* Error checking */
 919        __le64   bg_reserved2;
 920/*40*/  __u8    bg_bitmap[0];
 921};
 922
 923struct ocfs2_refcount_rec {
 924/*00*/  __le64 r_cpos;          /* Physical offset, in clusters */
 925        __le32 r_clusters;      /* Clusters covered by this extent */
 926        __le32 r_refcount;      /* Reference count of this extent */
 927/*10*/
 928};
 929#define OCFS2_32BIT_POS_MASK            (0xffffffffULL)
 930
 931#define OCFS2_REFCOUNT_LEAF_FL          (0x00000001)
 932#define OCFS2_REFCOUNT_TREE_FL          (0x00000002)
 933
 934struct ocfs2_refcount_list {
 935/*00*/  __le16 rl_count;        /* Maximum number of entries possible
 936                                   in rl_records */
 937        __le16 rl_used;         /* Current number of used records */
 938        __le32 rl_reserved2;
 939        __le64 rl_reserved1;    /* Pad to sizeof(ocfs2_refcount_record) */
 940/*10*/  struct ocfs2_refcount_rec rl_recs[0];   /* Refcount records */
 941};
 942
 943
 944struct ocfs2_refcount_block {
 945/*00*/  __u8 rf_signature[8];           /* Signature for verification */
 946        __le16 rf_suballoc_slot;        /* Slot suballocator this block
 947                                           belongs to */
 948        __le16 rf_suballoc_bit;         /* Bit offset in suballocator
 949                                           block group */
 950        __le32 rf_fs_generation;        /* Must match superblock */
 951/*10*/  __le64 rf_blkno;                /* Offset on disk, in blocks */
 952        __le64 rf_parent;               /* Parent block, only valid if
 953                                           OCFS2_REFCOUNT_LEAF_FL is set in
 954                                           rf_flags */
 955/*20*/  struct ocfs2_block_check rf_check;      /* Error checking */
 956        __le64 rf_last_eb_blk;          /* Pointer to last extent block */
 957/*30*/  __le32 rf_count;                /* Number of inodes sharing this
 958                                           refcount tree */
 959        __le32 rf_flags;                /* See the flags above */
 960        __le32 rf_clusters;             /* clusters covered by refcount tree. */
 961        __le32 rf_cpos;                 /* cluster offset in refcount tree.*/
 962/*40*/  __le32 rf_generation;           /* generation number. all be the same
 963                                         * for the same refcount tree. */
 964        __le32 rf_reserved0;
 965        __le64 rf_reserved1[7];
 966/*80*/  union {
 967                struct ocfs2_refcount_list rf_records;  /* List of refcount
 968                                                          records */
 969                struct ocfs2_extent_list rf_list;       /* Extent record list,
 970                                                        only valid if
 971                                                        OCFS2_REFCOUNT_TREE_FL
 972                                                        is set in rf_flags */
 973        };
 974/* Actual on-disk size is one block */
 975};
 976
 977/*
 978 * On disk extended attribute structure for OCFS2.
 979 */
 980
 981/*
 982 * ocfs2_xattr_entry indicates one extend attribute.
 983 *
 984 * Note that it can be stored in inode, one block or one xattr bucket.
 985 */
 986struct ocfs2_xattr_entry {
 987        __le32  xe_name_hash;    /* hash value of xattr prefix+suffix. */
 988        __le16  xe_name_offset;  /* byte offset from the 1st entry in the
 989                                    local xattr storage(inode, xattr block or
 990                                    xattr bucket). */
 991        __u8    xe_name_len;     /* xattr name len, does't include prefix. */
 992        __u8    xe_type;         /* the low 7 bits indicate the name prefix
 993                                  * type and the highest bit indicates whether
 994                                  * the EA is stored in the local storage. */
 995        __le64  xe_value_size;   /* real xattr value length. */
 996};
 997
 998/*
 999 * On disk structure for xattr header.
1000 *
1001 * One ocfs2_xattr_header describes how many ocfs2_xattr_entry records in
1002 * the local xattr storage.
1003 */
1004struct ocfs2_xattr_header {
1005        __le16  xh_count;                       /* contains the count of how
1006                                                   many records are in the
1007                                                   local xattr storage. */
1008        __le16  xh_free_start;                  /* current offset for storing
1009                                                   xattr. */
1010        __le16  xh_name_value_len;              /* total length of name/value
1011                                                   length in this bucket. */
1012        __le16  xh_num_buckets;                 /* Number of xattr buckets
1013                                                   in this extent record,
1014                                                   only valid in the first
1015                                                   bucket. */
1016        struct ocfs2_block_check xh_check;      /* Error checking
1017                                                   (Note, this is only
1018                                                    used for xattr
1019                                                    buckets.  A block uses
1020                                                    xb_check and sets
1021                                                    this field to zero.) */
1022        struct ocfs2_xattr_entry xh_entries[0]; /* xattr entry list. */
1023};
1024
1025/*
1026 * On disk structure for xattr value root.
1027 *
1028 * When an xattr's value is large enough, it is stored in an external
1029 * b-tree like file data.  The xattr value root points to this structure.
1030 */
1031struct ocfs2_xattr_value_root {
1032/*00*/  __le32  xr_clusters;              /* clusters covered by xattr value. */
1033        __le32  xr_reserved0;
1034        __le64  xr_last_eb_blk;           /* Pointer to last extent block */
1035/*10*/  struct ocfs2_extent_list xr_list; /* Extent record list */
1036};
1037
1038/*
1039 * On disk structure for xattr tree root.
1040 *
1041 * It is used when there are too many extended attributes for one file. These
1042 * attributes will be organized and stored in an indexed-btree.
1043 */
1044struct ocfs2_xattr_tree_root {
1045/*00*/  __le32  xt_clusters;              /* clusters covered by xattr. */
1046        __le32  xt_reserved0;
1047        __le64  xt_last_eb_blk;           /* Pointer to last extent block */
1048/*10*/  struct ocfs2_extent_list xt_list; /* Extent record list */
1049};
1050
1051#define OCFS2_XATTR_INDEXED     0x1
1052#define OCFS2_HASH_SHIFT        5
1053#define OCFS2_XATTR_ROUND       3
1054#define OCFS2_XATTR_SIZE(size)  (((size) + OCFS2_XATTR_ROUND) & \
1055                                ~(OCFS2_XATTR_ROUND))
1056
1057#define OCFS2_XATTR_BUCKET_SIZE                 4096
1058#define OCFS2_XATTR_MAX_BLOCKS_PER_BUCKET       (OCFS2_XATTR_BUCKET_SIZE \
1059                                                 / OCFS2_MIN_BLOCKSIZE)
1060
1061/*
1062 * On disk structure for xattr block.
1063 */
1064struct ocfs2_xattr_block {
1065/*00*/  __u8    xb_signature[8];     /* Signature for verification */
1066        __le16  xb_suballoc_slot;    /* Slot suballocator this
1067                                        block belongs to. */
1068        __le16  xb_suballoc_bit;     /* Bit offset in suballocator
1069                                        block group */
1070        __le32  xb_fs_generation;    /* Must match super block */
1071/*10*/  __le64  xb_blkno;            /* Offset on disk, in blocks */
1072        struct ocfs2_block_check xb_check;      /* Error checking */
1073/*20*/  __le16  xb_flags;            /* Indicates whether this block contains
1074                                        real xattr or a xattr tree. */
1075        __le16  xb_reserved0;
1076        __le32  xb_reserved1;
1077        __le64  xb_reserved2;
1078/*30*/  union {
1079                struct ocfs2_xattr_header xb_header; /* xattr header if this
1080                                                        block contains xattr */
1081                struct ocfs2_xattr_tree_root xb_root;/* xattr tree root if this
1082                                                        block cotains xattr
1083                                                        tree. */
1084        } xb_attrs;
1085};
1086
1087#define OCFS2_XATTR_ENTRY_LOCAL         0x80
1088#define OCFS2_XATTR_TYPE_MASK           0x7F
1089static inline void ocfs2_xattr_set_local(struct ocfs2_xattr_entry *xe,
1090                                         int local)
1091{
1092        if (local)
1093                xe->xe_type |= OCFS2_XATTR_ENTRY_LOCAL;
1094        else
1095                xe->xe_type &= ~OCFS2_XATTR_ENTRY_LOCAL;
1096}
1097
1098static inline int ocfs2_xattr_is_local(struct ocfs2_xattr_entry *xe)
1099{
1100        return xe->xe_type & OCFS2_XATTR_ENTRY_LOCAL;
1101}
1102
1103static inline void ocfs2_xattr_set_type(struct ocfs2_xattr_entry *xe, int type)
1104{
1105        xe->xe_type |= type & OCFS2_XATTR_TYPE_MASK;
1106}
1107
1108static inline int ocfs2_xattr_get_type(struct ocfs2_xattr_entry *xe)
1109{
1110        return xe->xe_type & OCFS2_XATTR_TYPE_MASK;
1111}
1112
1113/*
1114 *  On disk structures for global quota file
1115 */
1116
1117/* Magic numbers and known versions for global quota files */
1118#define OCFS2_GLOBAL_QMAGICS {\
1119        0x0cf52470, /* USRQUOTA */ \
1120        0x0cf52471  /* GRPQUOTA */ \
1121}
1122
1123#define OCFS2_GLOBAL_QVERSIONS {\
1124        0, \
1125        0, \
1126}
1127
1128
1129/* Each block of each quota file has a certain fixed number of bytes reserved
1130 * for OCFS2 internal use at its end. OCFS2 can use it for things like
1131 * checksums, etc. */
1132#define OCFS2_QBLK_RESERVED_SPACE 8
1133
1134/* Generic header of all quota files */
1135struct ocfs2_disk_dqheader {
1136        __le32 dqh_magic;       /* Magic number identifying file */
1137        __le32 dqh_version;     /* Quota format version */
1138};
1139
1140#define OCFS2_GLOBAL_INFO_OFF (sizeof(struct ocfs2_disk_dqheader))
1141
1142/* Information header of global quota file (immediately follows the generic
1143 * header) */
1144struct ocfs2_global_disk_dqinfo {
1145/*00*/  __le32 dqi_bgrace;      /* Grace time for space softlimit excess */
1146        __le32 dqi_igrace;      /* Grace time for inode softlimit excess */
1147        __le32 dqi_syncms;      /* Time after which we sync local changes to
1148                                 * global quota file */
1149        __le32 dqi_blocks;      /* Number of blocks in quota file */
1150/*10*/  __le32 dqi_free_blk;    /* First free block in quota file */
1151        __le32 dqi_free_entry;  /* First block with free dquot entry in quota
1152                                 * file */
1153};
1154
1155/* Structure with global user / group information. We reserve some space
1156 * for future use. */
1157struct ocfs2_global_disk_dqblk {
1158/*00*/  __le32 dqb_id;          /* ID the structure belongs to */
1159        __le32 dqb_use_count;   /* Number of nodes having reference to this structure */
1160        __le64 dqb_ihardlimit;  /* absolute limit on allocated inodes */
1161/*10*/  __le64 dqb_isoftlimit;  /* preferred inode limit */
1162        __le64 dqb_curinodes;   /* current # allocated inodes */
1163/*20*/  __le64 dqb_bhardlimit;  /* absolute limit on disk space */
1164        __le64 dqb_bsoftlimit;  /* preferred limit on disk space */
1165/*30*/  __le64 dqb_curspace;    /* current space occupied */
1166        __le64 dqb_btime;       /* time limit for excessive disk use */
1167/*40*/  __le64 dqb_itime;       /* time limit for excessive inode use */
1168        __le64 dqb_pad1;
1169/*50*/  __le64 dqb_pad2;
1170};
1171
1172/*
1173 *  On-disk structures for local quota file
1174 */
1175
1176/* Magic numbers and known versions for local quota files */
1177#define OCFS2_LOCAL_QMAGICS {\
1178        0x0cf524c0, /* USRQUOTA */ \
1179        0x0cf524c1  /* GRPQUOTA */ \
1180}
1181
1182#define OCFS2_LOCAL_QVERSIONS {\
1183        0, \
1184        0, \
1185}
1186
1187/* Quota flags in dqinfo header */
1188#define OLQF_CLEAN      0x0001  /* Quota file is empty (this should be after\
1189                                 * quota has been cleanly turned off) */
1190
1191#define OCFS2_LOCAL_INFO_OFF (sizeof(struct ocfs2_disk_dqheader))
1192
1193/* Information header of local quota file (immediately follows the generic
1194 * header) */
1195struct ocfs2_local_disk_dqinfo {
1196        __le32 dqi_flags;       /* Flags for quota file */
1197        __le32 dqi_chunks;      /* Number of chunks of quota structures
1198                                 * with a bitmap */
1199        __le32 dqi_blocks;      /* Number of blocks allocated for quota file */
1200};
1201
1202/* Header of one chunk of a quota file */
1203struct ocfs2_local_disk_chunk {
1204        __le32 dqc_free;        /* Number of free entries in the bitmap */
1205        u8 dqc_bitmap[0];       /* Bitmap of entries in the corresponding
1206                                 * chunk of quota file */
1207};
1208
1209/* One entry in local quota file */
1210struct ocfs2_local_disk_dqblk {
1211/*00*/  __le64 dqb_id;          /* id this quota applies to */
1212        __le64 dqb_spacemod;    /* Change in the amount of used space */
1213/*10*/  __le64 dqb_inodemod;    /* Change in the amount of used inodes */
1214};
1215
1216
1217/*
1218 * The quota trailer lives at the end of each quota block.
1219 */
1220
1221struct ocfs2_disk_dqtrailer {
1222/*00*/  struct ocfs2_block_check dq_check;      /* Error checking */
1223/*08*/  /* Cannot be larger than OCFS2_QBLK_RESERVED_SPACE */
1224};
1225
1226static inline struct ocfs2_disk_dqtrailer *ocfs2_block_dqtrailer(int blocksize,
1227                                                                 void *buf)
1228{
1229        char *ptr = buf;
1230        ptr += blocksize - OCFS2_QBLK_RESERVED_SPACE;
1231
1232        return (struct ocfs2_disk_dqtrailer *)ptr;
1233}
1234
1235#ifdef __KERNEL__
1236static inline int ocfs2_fast_symlink_chars(struct super_block *sb)
1237{
1238        return  sb->s_blocksize -
1239                 offsetof(struct ocfs2_dinode, id2.i_symlink);
1240}
1241
1242static inline int ocfs2_max_inline_data_with_xattr(struct super_block *sb,
1243                                                   struct ocfs2_dinode *di)
1244{
1245        unsigned int xattrsize = le16_to_cpu(di->i_xattr_inline_size);
1246
1247        if (le16_to_cpu(di->i_dyn_features) & OCFS2_INLINE_XATTR_FL)
1248                return sb->s_blocksize -
1249                        offsetof(struct ocfs2_dinode, id2.i_data.id_data) -
1250                        xattrsize;
1251        else
1252                return sb->s_blocksize -
1253                        offsetof(struct ocfs2_dinode, id2.i_data.id_data);
1254}
1255
1256static inline int ocfs2_extent_recs_per_inode(struct super_block *sb)
1257{
1258        int size;
1259
1260        size = sb->s_blocksize -
1261                offsetof(struct ocfs2_dinode, id2.i_list.l_recs);
1262
1263        return size / sizeof(struct ocfs2_extent_rec);
1264}
1265
1266static inline int ocfs2_extent_recs_per_inode_with_xattr(
1267                                                struct super_block *sb,
1268                                                struct ocfs2_dinode *di)
1269{
1270        int size;
1271        unsigned int xattrsize = le16_to_cpu(di->i_xattr_inline_size);
1272
1273        if (le16_to_cpu(di->i_dyn_features) & OCFS2_INLINE_XATTR_FL)
1274                size = sb->s_blocksize -
1275                        offsetof(struct ocfs2_dinode, id2.i_list.l_recs) -
1276                        xattrsize;
1277        else
1278                size = sb->s_blocksize -
1279                        offsetof(struct ocfs2_dinode, id2.i_list.l_recs);
1280
1281        return size / sizeof(struct ocfs2_extent_rec);
1282}
1283
1284static inline int ocfs2_extent_recs_per_dx_root(struct super_block *sb)
1285{
1286        int size;
1287
1288        size = sb->s_blocksize -
1289                offsetof(struct ocfs2_dx_root_block, dr_list.l_recs);
1290
1291        return size / sizeof(struct ocfs2_extent_rec);
1292}
1293
1294static inline int ocfs2_chain_recs_per_inode(struct super_block *sb)
1295{
1296        int size;
1297
1298        size = sb->s_blocksize -
1299                offsetof(struct ocfs2_dinode, id2.i_chain.cl_recs);
1300
1301        return size / sizeof(struct ocfs2_chain_rec);
1302}
1303
1304static inline u16 ocfs2_extent_recs_per_eb(struct super_block *sb)
1305{
1306        int size;
1307
1308        size = sb->s_blocksize -
1309                offsetof(struct ocfs2_extent_block, h_list.l_recs);
1310
1311        return size / sizeof(struct ocfs2_extent_rec);
1312}
1313
1314static inline int ocfs2_dx_entries_per_leaf(struct super_block *sb)
1315{
1316        int size;
1317
1318        size = sb->s_blocksize -
1319                offsetof(struct ocfs2_dx_leaf, dl_list.de_entries);
1320
1321        return size / sizeof(struct ocfs2_dx_entry);
1322}
1323
1324static inline int ocfs2_dx_entries_per_root(struct super_block *sb)
1325{
1326        int size;
1327
1328        size = sb->s_blocksize -
1329                offsetof(struct ocfs2_dx_root_block, dr_entries.de_entries);
1330
1331        return size / sizeof(struct ocfs2_dx_entry);
1332}
1333
1334static inline u16 ocfs2_local_alloc_size(struct super_block *sb)
1335{
1336        u16 size;
1337
1338        size = sb->s_blocksize -
1339                offsetof(struct ocfs2_dinode, id2.i_lab.la_bitmap);
1340
1341        return size;
1342}
1343
1344static inline int ocfs2_group_bitmap_size(struct super_block *sb)
1345{
1346        int size;
1347
1348        size = sb->s_blocksize -
1349                offsetof(struct ocfs2_group_desc, bg_bitmap);
1350
1351        return size;
1352}
1353
1354static inline int ocfs2_truncate_recs_per_inode(struct super_block *sb)
1355{
1356        int size;
1357
1358        size = sb->s_blocksize -
1359                offsetof(struct ocfs2_dinode, id2.i_dealloc.tl_recs);
1360
1361        return size / sizeof(struct ocfs2_truncate_rec);
1362}
1363
1364static inline u64 ocfs2_backup_super_blkno(struct super_block *sb, int index)
1365{
1366        u64 offset = OCFS2_BACKUP_SB_START;
1367
1368        if (index >= 0 && index < OCFS2_MAX_BACKUP_SUPERBLOCKS) {
1369                offset <<= (2 * index);
1370                offset >>= sb->s_blocksize_bits;
1371                return offset;
1372        }
1373
1374        return 0;
1375
1376}
1377
1378static inline u16 ocfs2_xattr_recs_per_xb(struct super_block *sb)
1379{
1380        int size;
1381
1382        size = sb->s_blocksize -
1383                offsetof(struct ocfs2_xattr_block,
1384                         xb_attrs.xb_root.xt_list.l_recs);
1385
1386        return size / sizeof(struct ocfs2_extent_rec);
1387}
1388
1389static inline u16 ocfs2_extent_recs_per_rb(struct super_block *sb)
1390{
1391        int size;
1392
1393        size = sb->s_blocksize -
1394                offsetof(struct ocfs2_refcount_block, rf_list.l_recs);
1395
1396        return size / sizeof(struct ocfs2_extent_rec);
1397}
1398
1399static inline u16 ocfs2_refcount_recs_per_rb(struct super_block *sb)
1400{
1401        int size;
1402
1403        size = sb->s_blocksize -
1404                offsetof(struct ocfs2_refcount_block, rf_records.rl_recs);
1405
1406        return size / sizeof(struct ocfs2_refcount_rec);
1407}
1408
1409static inline u32
1410ocfs2_get_ref_rec_low_cpos(const struct ocfs2_refcount_rec *rec)
1411{
1412        return le64_to_cpu(rec->r_cpos) & OCFS2_32BIT_POS_MASK;
1413}
1414#else
1415static inline int ocfs2_fast_symlink_chars(int blocksize)
1416{
1417        return blocksize - offsetof(struct ocfs2_dinode, id2.i_symlink);
1418}
1419
1420static inline int ocfs2_max_inline_data(int blocksize)
1421{
1422        return blocksize - offsetof(struct ocfs2_dinode, id2.i_data.id_data);
1423}
1424
1425static inline int ocfs2_extent_recs_per_inode(int blocksize)
1426{
1427        int size;
1428
1429        size = blocksize -
1430                offsetof(struct ocfs2_dinode, id2.i_list.l_recs);
1431
1432        return size / sizeof(struct ocfs2_extent_rec);
1433}
1434
1435static inline int ocfs2_chain_recs_per_inode(int blocksize)
1436{
1437        int size;
1438
1439        size = blocksize -
1440                offsetof(struct ocfs2_dinode, id2.i_chain.cl_recs);
1441
1442        return size / sizeof(struct ocfs2_chain_rec);
1443}
1444
1445static inline int ocfs2_extent_recs_per_eb(int blocksize)
1446{
1447        int size;
1448
1449        size = blocksize -
1450                offsetof(struct ocfs2_extent_block, h_list.l_recs);
1451
1452        return size / sizeof(struct ocfs2_extent_rec);
1453}
1454
1455static inline int ocfs2_local_alloc_size(int blocksize)
1456{
1457        int size;
1458
1459        size = blocksize -
1460                offsetof(struct ocfs2_dinode, id2.i_lab.la_bitmap);
1461
1462        return size;
1463}
1464
1465static inline int ocfs2_group_bitmap_size(int blocksize)
1466{
1467        int size;
1468
1469        size = blocksize -
1470                offsetof(struct ocfs2_group_desc, bg_bitmap);
1471
1472        return size;
1473}
1474
1475static inline int ocfs2_truncate_recs_per_inode(int blocksize)
1476{
1477        int size;
1478
1479        size = blocksize -
1480                offsetof(struct ocfs2_dinode, id2.i_dealloc.tl_recs);
1481
1482        return size / sizeof(struct ocfs2_truncate_rec);
1483}
1484
1485static inline uint64_t ocfs2_backup_super_blkno(int blocksize, int index)
1486{
1487        uint64_t offset = OCFS2_BACKUP_SB_START;
1488
1489        if (index >= 0 && index < OCFS2_MAX_BACKUP_SUPERBLOCKS) {
1490                offset <<= (2 * index);
1491                offset /= blocksize;
1492                return offset;
1493        }
1494
1495        return 0;
1496}
1497
1498static inline int ocfs2_xattr_recs_per_xb(int blocksize)
1499{
1500        int size;
1501
1502        size = blocksize -
1503                offsetof(struct ocfs2_xattr_block,
1504                         xb_attrs.xb_root.xt_list.l_recs);
1505
1506        return size / sizeof(struct ocfs2_extent_rec);
1507}
1508#endif  /* __KERNEL__ */
1509
1510
1511static inline int ocfs2_system_inode_is_global(int type)
1512{
1513        return ((type >= 0) &&
1514                (type <= OCFS2_LAST_GLOBAL_SYSTEM_INODE));
1515}
1516
1517static inline int ocfs2_sprintf_system_inode_name(char *buf, int len,
1518                                                  int type, int slot)
1519{
1520        int chars;
1521
1522        /*
1523         * Global system inodes can only have one copy.  Everything
1524         * after OCFS2_LAST_GLOBAL_SYSTEM_INODE in the system inode
1525         * list has a copy per slot.
1526         */
1527        if (type <= OCFS2_LAST_GLOBAL_SYSTEM_INODE)
1528                chars = snprintf(buf, len, "%s",
1529                                 ocfs2_system_inodes[type].si_name);
1530        else
1531                chars = snprintf(buf, len,
1532                                 ocfs2_system_inodes[type].si_name,
1533                                 slot);
1534
1535        return chars;
1536}
1537
1538static inline void ocfs2_set_de_type(struct ocfs2_dir_entry *de,
1539                                    umode_t mode)
1540{
1541        de->file_type = ocfs2_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
1542}
1543
1544#endif  /* _OCFS2_FS_H */
1545
1546