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 #include <linux/magic.h> 29 30 /* Version */ 31 #define OCFS2_MAJOR_REV_LEVEL 0 32 #define OCFS2_MINOR_REV_LEVEL 90 33 34 /* 35 * An OCFS2 volume starts this way: 36 * Sector 0: Valid ocfs1_vol_disk_hdr that cleanly fails to mount OCFS. 37 * Sector 1: Valid ocfs1_vol_label that cleanly fails to mount OCFS. 38 * Block OCFS2_SUPER_BLOCK_BLKNO: OCFS2 superblock. 39 * 40 * All other structures are found from the superblock information. 41 * 42 * OCFS2_SUPER_BLOCK_BLKNO is in blocks, not sectors. eg, for a 43 * blocksize of 2K, it is 4096 bytes into disk. 44 */ 45 #define OCFS2_SUPER_BLOCK_BLKNO 2 46 47 /* 48 * Cluster size limits. The maximum is kept arbitrarily at 1 MB, and could 49 * grow if needed. 50 */ 51 #define OCFS2_MIN_CLUSTERSIZE 4096 52 #define OCFS2_MAX_CLUSTERSIZE 1048576 53 54 /* 55 * Blocks cannot be bigger than clusters, so the maximum blocksize is the 56 * minimum cluster size. 57 */ 58 #define OCFS2_MIN_BLOCKSIZE 512 59 #define OCFS2_MAX_BLOCKSIZE OCFS2_MIN_CLUSTERSIZE 60 61 /* Object signatures */ 62 #define OCFS2_SUPER_BLOCK_SIGNATURE "OCFSV2" 63 #define OCFS2_INODE_SIGNATURE "INODE01" 64 #define OCFS2_EXTENT_BLOCK_SIGNATURE "EXBLK01" 65 #define OCFS2_GROUP_DESC_SIGNATURE "GROUP01" 66 #define OCFS2_XATTR_BLOCK_SIGNATURE "XATTR01" 67 #define OCFS2_DIR_TRAILER_SIGNATURE "DIRTRL1" 68 #define OCFS2_DX_ROOT_SIGNATURE "DXDIR01" 69 #define OCFS2_DX_LEAF_SIGNATURE "DXLEAF1" 70 #define OCFS2_REFCOUNT_BLOCK_SIGNATURE "REFCNT1" 71 72 /* Compatibility flags */ 73 #define OCFS2_HAS_COMPAT_FEATURE(sb,mask) \ 74 ( OCFS2_SB(sb)->s_feature_compat & (mask) ) 75 #define OCFS2_HAS_RO_COMPAT_FEATURE(sb,mask) \ 76 ( OCFS2_SB(sb)->s_feature_ro_compat & (mask) ) 77 #define OCFS2_HAS_INCOMPAT_FEATURE(sb,mask) \ 78 ( OCFS2_SB(sb)->s_feature_incompat & (mask) ) 79 #define OCFS2_SET_COMPAT_FEATURE(sb,mask) \ 80 OCFS2_SB(sb)->s_feature_compat |= (mask) 81 #define OCFS2_SET_RO_COMPAT_FEATURE(sb,mask) \ 82 OCFS2_SB(sb)->s_feature_ro_compat |= (mask) 83 #define OCFS2_SET_INCOMPAT_FEATURE(sb,mask) \ 84 OCFS2_SB(sb)->s_feature_incompat |= (mask) 85 #define OCFS2_CLEAR_COMPAT_FEATURE(sb,mask) \ 86 OCFS2_SB(sb)->s_feature_compat &= ~(mask) 87 #define OCFS2_CLEAR_RO_COMPAT_FEATURE(sb,mask) \ 88 OCFS2_SB(sb)->s_feature_ro_compat &= ~(mask) 89 #define OCFS2_CLEAR_INCOMPAT_FEATURE(sb,mask) \ 90 OCFS2_SB(sb)->s_feature_incompat &= ~(mask) 91 92 #define OCFS2_FEATURE_COMPAT_SUPP (OCFS2_FEATURE_COMPAT_BACKUP_SB \ 93 | OCFS2_FEATURE_COMPAT_JBD2_SB) 94 #define OCFS2_FEATURE_INCOMPAT_SUPP (OCFS2_FEATURE_INCOMPAT_LOCAL_MOUNT \ 95 | OCFS2_FEATURE_INCOMPAT_SPARSE_ALLOC \ 96 | OCFS2_FEATURE_INCOMPAT_INLINE_DATA \ 97 | OCFS2_FEATURE_INCOMPAT_EXTENDED_SLOT_MAP \ 98 | OCFS2_FEATURE_INCOMPAT_USERSPACE_STACK \ 99 | OCFS2_FEATURE_INCOMPAT_XATTR \ 100 | OCFS2_FEATURE_INCOMPAT_META_ECC \ 101 | OCFS2_FEATURE_INCOMPAT_INDEXED_DIRS \ 102 | OCFS2_FEATURE_INCOMPAT_REFCOUNT_TREE \ 103 | OCFS2_FEATURE_INCOMPAT_DISCONTIG_BG \ 104 | OCFS2_FEATURE_INCOMPAT_CLUSTERINFO \ 105 | OCFS2_FEATURE_INCOMPAT_APPEND_DIO) 106 #define OCFS2_FEATURE_RO_COMPAT_SUPP (OCFS2_FEATURE_RO_COMPAT_UNWRITTEN \ 107 | OCFS2_FEATURE_RO_COMPAT_USRQUOTA \ 108 | OCFS2_FEATURE_RO_COMPAT_GRPQUOTA) 109 110 /* 111 * Heartbeat-only devices are missing journals and other files. The 112 * filesystem driver can't load them, but the library can. Never put 113 * this in OCFS2_FEATURE_INCOMPAT_SUPP, *ever*. 114 */ 115 #define OCFS2_FEATURE_INCOMPAT_HEARTBEAT_DEV 0x0002 116 117 /* 118 * tunefs sets this incompat flag before starting the resize and clears it 119 * at the end. This flag protects users from inadvertently mounting the fs 120 * after an aborted run without fsck-ing. 121 */ 122 #define OCFS2_FEATURE_INCOMPAT_RESIZE_INPROG 0x0004 123 124 /* Used to denote a non-clustered volume */ 125 #define OCFS2_FEATURE_INCOMPAT_LOCAL_MOUNT 0x0008 126 127 /* Support for sparse allocation in b-trees */ 128 #define OCFS2_FEATURE_INCOMPAT_SPARSE_ALLOC 0x0010 129 130 /* 131 * Tunefs sets this incompat flag before starting an operation which 132 * would require cleanup on abort. This is done to protect users from 133 * inadvertently mounting the fs after an aborted run without 134 * fsck-ing. 135 * 136 * s_tunefs_flags on the super block describes precisely which 137 * operations were in progress. 138 */ 139 #define OCFS2_FEATURE_INCOMPAT_TUNEFS_INPROG 0x0020 140 141 /* Support for data packed into inode blocks */ 142 #define OCFS2_FEATURE_INCOMPAT_INLINE_DATA 0x0040 143 144 /* 145 * Support for alternate, userspace cluster stacks. If set, the superblock 146 * field s_cluster_info contains a tag for the alternate stack in use as 147 * well as the name of the cluster being joined. 148 * mount.ocfs2 must pass in a matching stack name. 149 * 150 * If not set, the classic stack will be used. This is compatbile with 151 * all older versions. 152 */ 153 #define OCFS2_FEATURE_INCOMPAT_USERSPACE_STACK 0x0080 154 155 /* Support for the extended slot map */ 156 #define OCFS2_FEATURE_INCOMPAT_EXTENDED_SLOT_MAP 0x100 157 158 /* Support for extended attributes */ 159 #define OCFS2_FEATURE_INCOMPAT_XATTR 0x0200 160 161 /* Support for indexed directores */ 162 #define OCFS2_FEATURE_INCOMPAT_INDEXED_DIRS 0x0400 163 164 /* Metadata checksum and error correction */ 165 #define OCFS2_FEATURE_INCOMPAT_META_ECC 0x0800 166 167 /* Refcount tree support */ 168 #define OCFS2_FEATURE_INCOMPAT_REFCOUNT_TREE 0x1000 169 170 /* Discontiguous block groups */ 171 #define OCFS2_FEATURE_INCOMPAT_DISCONTIG_BG 0x2000 172 173 /* 174 * Incompat bit to indicate useable clusterinfo with stackflags for all 175 * cluster stacks (userspace adnd o2cb). If this bit is set, 176 * INCOMPAT_USERSPACE_STACK becomes superfluous and thus should not be set. 177 */ 178 #define OCFS2_FEATURE_INCOMPAT_CLUSTERINFO 0x4000 179 180 /* 181 * Append Direct IO support 182 */ 183 #define OCFS2_FEATURE_INCOMPAT_APPEND_DIO 0x8000 184 185 /* 186 * backup superblock flag is used to indicate that this volume 187 * has backup superblocks. 188 */ 189 #define OCFS2_FEATURE_COMPAT_BACKUP_SB 0x0001 190 191 /* 192 * The filesystem will correctly handle journal feature bits. 193 */ 194 #define OCFS2_FEATURE_COMPAT_JBD2_SB 0x0002 195 196 /* 197 * Unwritten extents support. 198 */ 199 #define OCFS2_FEATURE_RO_COMPAT_UNWRITTEN 0x0001 200 201 /* 202 * Maintain quota information for this filesystem 203 */ 204 #define OCFS2_FEATURE_RO_COMPAT_USRQUOTA 0x0002 205 #define OCFS2_FEATURE_RO_COMPAT_GRPQUOTA 0x0004 206 207 208 /* The byte offset of the first backup block will be 1G. 209 * The following will be 4G, 16G, 64G, 256G and 1T. 210 */ 211 #define OCFS2_BACKUP_SB_START 1 << 30 212 213 /* the max backup superblock nums */ 214 #define OCFS2_MAX_BACKUP_SUPERBLOCKS 6 215 216 /* 217 * Flags on ocfs2_super_block.s_tunefs_flags 218 */ 219 #define OCFS2_TUNEFS_INPROG_REMOVE_SLOT 0x0001 /* Removing slots */ 220 221 /* 222 * Flags on ocfs2_dinode.i_flags 223 */ 224 #define OCFS2_VALID_FL (0x00000001) /* Inode is valid */ 225 #define OCFS2_UNUSED2_FL (0x00000002) 226 #define OCFS2_ORPHANED_FL (0x00000004) /* On the orphan list */ 227 #define OCFS2_UNUSED3_FL (0x00000008) 228 /* System inode flags */ 229 #define OCFS2_SYSTEM_FL (0x00000010) /* System inode */ 230 #define OCFS2_SUPER_BLOCK_FL (0x00000020) /* Super block */ 231 #define OCFS2_LOCAL_ALLOC_FL (0x00000040) /* Slot local alloc bitmap */ 232 #define OCFS2_BITMAP_FL (0x00000080) /* Allocation bitmap */ 233 #define OCFS2_JOURNAL_FL (0x00000100) /* Slot local journal */ 234 #define OCFS2_HEARTBEAT_FL (0x00000200) /* Heartbeat area */ 235 #define OCFS2_CHAIN_FL (0x00000400) /* Chain allocator */ 236 #define OCFS2_DEALLOC_FL (0x00000800) /* Truncate log */ 237 #define OCFS2_QUOTA_FL (0x00001000) /* Quota file */ 238 #define OCFS2_DIO_ORPHANED_FL (0X00002000) /* On the orphan list especially 239 * for dio */ 240 241 /* 242 * Flags on ocfs2_dinode.i_dyn_features 243 * 244 * These can change much more often than i_flags. When adding flags, 245 * keep in mind that i_dyn_features is only 16 bits wide. 246 */ 247 #define OCFS2_INLINE_DATA_FL (0x0001) /* Data stored in inode block */ 248 #define OCFS2_HAS_XATTR_FL (0x0002) 249 #define OCFS2_INLINE_XATTR_FL (0x0004) 250 #define OCFS2_INDEXED_DIR_FL (0x0008) 251 #define OCFS2_HAS_REFCOUNT_FL (0x0010) 252 253 /* Inode attributes, keep in sync with EXT2 */ 254 #define OCFS2_SECRM_FL FS_SECRM_FL /* Secure deletion */ 255 #define OCFS2_UNRM_FL FS_UNRM_FL /* Undelete */ 256 #define OCFS2_COMPR_FL FS_COMPR_FL /* Compress file */ 257 #define OCFS2_SYNC_FL FS_SYNC_FL /* Synchronous updates */ 258 #define OCFS2_IMMUTABLE_FL FS_IMMUTABLE_FL /* Immutable file */ 259 #define OCFS2_APPEND_FL FS_APPEND_FL /* writes to file may only append */ 260 #define OCFS2_NODUMP_FL FS_NODUMP_FL /* do not dump file */ 261 #define OCFS2_NOATIME_FL FS_NOATIME_FL /* do not update atime */ 262 /* Reserved for compression usage... */ 263 #define OCFS2_DIRTY_FL FS_DIRTY_FL 264 #define OCFS2_COMPRBLK_FL FS_COMPRBLK_FL /* One or more compressed clusters */ 265 #define OCFS2_NOCOMP_FL FS_NOCOMP_FL /* Don't compress */ 266 #define OCFS2_ECOMPR_FL FS_ECOMPR_FL /* Compression error */ 267 /* End compression flags --- maybe not all used */ 268 #define OCFS2_BTREE_FL FS_BTREE_FL /* btree format dir */ 269 #define OCFS2_INDEX_FL FS_INDEX_FL /* hash-indexed directory */ 270 #define OCFS2_IMAGIC_FL FS_IMAGIC_FL /* AFS directory */ 271 #define OCFS2_JOURNAL_DATA_FL FS_JOURNAL_DATA_FL /* Reserved for ext3 */ 272 #define OCFS2_NOTAIL_FL FS_NOTAIL_FL /* file tail should not be merged */ 273 #define OCFS2_DIRSYNC_FL FS_DIRSYNC_FL /* dirsync behaviour (directories only) */ 274 #define OCFS2_TOPDIR_FL FS_TOPDIR_FL /* Top of directory hierarchies*/ 275 #define OCFS2_RESERVED_FL FS_RESERVED_FL /* reserved for ext2 lib */ 276 277 #define OCFS2_FL_VISIBLE FS_FL_USER_VISIBLE /* User visible flags */ 278 #define OCFS2_FL_MODIFIABLE FS_FL_USER_MODIFIABLE /* User modifiable flags */ 279 280 /* 281 * Extent record flags (e_node.leaf.flags) 282 */ 283 #define OCFS2_EXT_UNWRITTEN (0x01) /* Extent is allocated but 284 * unwritten */ 285 #define OCFS2_EXT_REFCOUNTED (0x02) /* Extent is reference 286 * counted in an associated 287 * refcount tree */ 288 289 /* 290 * Journal Flags (ocfs2_dinode.id1.journal1.i_flags) 291 */ 292 #define OCFS2_JOURNAL_DIRTY_FL (0x00000001) /* Journal needs recovery */ 293 294 /* 295 * superblock s_state flags 296 */ 297 #define OCFS2_ERROR_FS (0x00000001) /* FS saw errors */ 298 299 /* Limit of space in ocfs2_dir_entry */ 300 #define OCFS2_MAX_FILENAME_LEN 255 301 302 /* Maximum slots on an ocfs2 file system */ 303 #define OCFS2_MAX_SLOTS 255 304 305 /* Slot map indicator for an empty slot */ 306 #define OCFS2_INVALID_SLOT -1 307 308 #define OCFS2_VOL_UUID_LEN 16 309 #define OCFS2_MAX_VOL_LABEL_LEN 64 310 311 /* The cluster stack fields */ 312 #define OCFS2_STACK_LABEL_LEN 4 313 #define OCFS2_CLUSTER_NAME_LEN 16 314 315 /* Classic (historically speaking) cluster stack */ 316 #define OCFS2_CLASSIC_CLUSTER_STACK "o2cb" 317 318 /* Journal limits (in bytes) */ 319 #define OCFS2_MIN_JOURNAL_SIZE (4 * 1024 * 1024) 320 321 /* 322 * Inline extended attribute size (in bytes) 323 * The value chosen should be aligned to 16 byte boundaries. 324 */ 325 #define OCFS2_MIN_XATTR_INLINE_SIZE 256 326 327 /* 328 * Cluster info flags (ocfs2_cluster_info.ci_stackflags) 329 */ 330 #define OCFS2_CLUSTER_O2CB_GLOBAL_HEARTBEAT (0x01) 331 332 struct ocfs2_system_inode_info { 333 char *si_name; 334 int si_iflags; 335 int si_mode; 336 }; 337 338 /* System file index */ 339 enum { 340 BAD_BLOCK_SYSTEM_INODE = 0, 341 GLOBAL_INODE_ALLOC_SYSTEM_INODE, 342 SLOT_MAP_SYSTEM_INODE, 343 #define OCFS2_FIRST_ONLINE_SYSTEM_INODE SLOT_MAP_SYSTEM_INODE 344 HEARTBEAT_SYSTEM_INODE, 345 GLOBAL_BITMAP_SYSTEM_INODE, 346 USER_QUOTA_SYSTEM_INODE, 347 GROUP_QUOTA_SYSTEM_INODE, 348 #define OCFS2_LAST_GLOBAL_SYSTEM_INODE GROUP_QUOTA_SYSTEM_INODE 349 #define OCFS2_FIRST_LOCAL_SYSTEM_INODE ORPHAN_DIR_SYSTEM_INODE 350 ORPHAN_DIR_SYSTEM_INODE, 351 EXTENT_ALLOC_SYSTEM_INODE, 352 INODE_ALLOC_SYSTEM_INODE, 353 JOURNAL_SYSTEM_INODE, 354 LOCAL_ALLOC_SYSTEM_INODE, 355 TRUNCATE_LOG_SYSTEM_INODE, 356 LOCAL_USER_QUOTA_SYSTEM_INODE, 357 LOCAL_GROUP_QUOTA_SYSTEM_INODE, 358 #define OCFS2_LAST_LOCAL_SYSTEM_INODE LOCAL_GROUP_QUOTA_SYSTEM_INODE 359 NUM_SYSTEM_INODES 360 }; 361 #define NUM_GLOBAL_SYSTEM_INODES OCFS2_FIRST_LOCAL_SYSTEM_INODE 362 #define NUM_LOCAL_SYSTEM_INODES \ 363 (NUM_SYSTEM_INODES - OCFS2_FIRST_LOCAL_SYSTEM_INODE) 364 365 static struct ocfs2_system_inode_info ocfs2_system_inodes[NUM_SYSTEM_INODES] = { 366 /* Global system inodes (single copy) */ 367 /* The first two are only used from userspace mfks/tunefs */ 368 [BAD_BLOCK_SYSTEM_INODE] = { "bad_blocks", 0, S_IFREG | 0644 }, 369 [GLOBAL_INODE_ALLOC_SYSTEM_INODE] = { "global_inode_alloc", OCFS2_BITMAP_FL | OCFS2_CHAIN_FL, S_IFREG | 0644 }, 370 371 /* These are used by the running filesystem */ 372 [SLOT_MAP_SYSTEM_INODE] = { "slot_map", 0, S_IFREG | 0644 }, 373 [HEARTBEAT_SYSTEM_INODE] = { "heartbeat", OCFS2_HEARTBEAT_FL, S_IFREG | 0644 }, 374 [GLOBAL_BITMAP_SYSTEM_INODE] = { "global_bitmap", 0, S_IFREG | 0644 }, 375 [USER_QUOTA_SYSTEM_INODE] = { "aquota.user", OCFS2_QUOTA_FL, S_IFREG | 0644 }, 376 [GROUP_QUOTA_SYSTEM_INODE] = { "aquota.group", OCFS2_QUOTA_FL, S_IFREG | 0644 }, 377 378 /* Slot-specific system inodes (one copy per slot) */ 379 [ORPHAN_DIR_SYSTEM_INODE] = { "orphan_dir:%04d", 0, S_IFDIR | 0755 }, 380 [EXTENT_ALLOC_SYSTEM_INODE] = { "extent_alloc:%04d", OCFS2_BITMAP_FL | OCFS2_CHAIN_FL, S_IFREG | 0644 }, 381 [INODE_ALLOC_SYSTEM_INODE] = { "inode_alloc:%04d", OCFS2_BITMAP_FL | OCFS2_CHAIN_FL, S_IFREG | 0644 }, 382 [JOURNAL_SYSTEM_INODE] = { "journal:%04d", OCFS2_JOURNAL_FL, S_IFREG | 0644 }, 383 [LOCAL_ALLOC_SYSTEM_INODE] = { "local_alloc:%04d", OCFS2_BITMAP_FL | OCFS2_LOCAL_ALLOC_FL, S_IFREG | 0644 }, 384 [TRUNCATE_LOG_SYSTEM_INODE] = { "truncate_log:%04d", OCFS2_DEALLOC_FL, S_IFREG | 0644 }, 385 [LOCAL_USER_QUOTA_SYSTEM_INODE] = { "aquota.user:%04d", OCFS2_QUOTA_FL, S_IFREG | 0644 }, 386 [LOCAL_GROUP_QUOTA_SYSTEM_INODE] = { "aquota.group:%04d", OCFS2_QUOTA_FL, S_IFREG | 0644 }, 387 }; 388 389 /* Parameter passed from mount.ocfs2 to module */ 390 #define OCFS2_HB_NONE "heartbeat=none" 391 #define OCFS2_HB_LOCAL "heartbeat=local" 392 #define OCFS2_HB_GLOBAL "heartbeat=global" 393 394 /* 395 * OCFS2 directory file types. Only the low 3 bits are used. The 396 * other bits are reserved for now. 397 */ 398 #define OCFS2_FT_UNKNOWN 0 399 #define OCFS2_FT_REG_FILE 1 400 #define OCFS2_FT_DIR 2 401 #define OCFS2_FT_CHRDEV 3 402 #define OCFS2_FT_BLKDEV 4 403 #define OCFS2_FT_FIFO 5 404 #define OCFS2_FT_SOCK 6 405 #define OCFS2_FT_SYMLINK 7 406 407 #define OCFS2_FT_MAX 8 408 409 /* 410 * OCFS2_DIR_PAD defines the directory entries boundaries 411 * 412 * NOTE: It must be a multiple of 4 413 */ 414 #define OCFS2_DIR_PAD 4 415 #define OCFS2_DIR_ROUND (OCFS2_DIR_PAD - 1) 416 #define OCFS2_DIR_MEMBER_LEN offsetof(struct ocfs2_dir_entry, name) 417 #define OCFS2_DIR_REC_LEN(name_len) (((name_len) + OCFS2_DIR_MEMBER_LEN + \ 418 OCFS2_DIR_ROUND) & \ 419 ~OCFS2_DIR_ROUND) 420 #define OCFS2_DIR_MIN_REC_LEN OCFS2_DIR_REC_LEN(1) 421 422 #define OCFS2_LINK_MAX 32000 423 #define OCFS2_DX_LINK_MAX ((1U << 31) - 1U) 424 #define OCFS2_LINKS_HI_SHIFT 16 425 #define OCFS2_DX_ENTRIES_MAX (0xffffffffU) 426 427 #define S_SHIFT 12 428 static unsigned char ocfs2_type_by_mode[S_IFMT >> S_SHIFT] = { 429 [S_IFREG >> S_SHIFT] = OCFS2_FT_REG_FILE, 430 [S_IFDIR >> S_SHIFT] = OCFS2_FT_DIR, 431 [S_IFCHR >> S_SHIFT] = OCFS2_FT_CHRDEV, 432 [S_IFBLK >> S_SHIFT] = OCFS2_FT_BLKDEV, 433 [S_IFIFO >> S_SHIFT] = OCFS2_FT_FIFO, 434 [S_IFSOCK >> S_SHIFT] = OCFS2_FT_SOCK, 435 [S_IFLNK >> S_SHIFT] = OCFS2_FT_SYMLINK, 436 }; 437 438 439 /* 440 * Convenience casts 441 */ 442 #define OCFS2_RAW_SB(dinode) (&((dinode)->id2.i_super)) 443 444 /* 445 * Block checking structure. This is used in metadata to validate the 446 * contents. If OCFS2_FEATURE_INCOMPAT_META_ECC is not set, it is all 447 * zeros. 448 */ 449 struct ocfs2_block_check { 450 /*00*/ __le32 bc_crc32e; /* 802.3 Ethernet II CRC32 */ 451 __le16 bc_ecc; /* Single-error-correction parity vector. 452 This is a simple Hamming code dependent 453 on the blocksize. OCFS2's maximum 454 blocksize, 4K, requires 16 parity bits, 455 so we fit in __le16. */ 456 __le16 bc_reserved1; 457 /*08*/ 458 }; 459 460 /* 461 * On disk extent record for OCFS2 462 * It describes a range of clusters on disk. 463 * 464 * Length fields are divided into interior and leaf node versions. 465 * This leaves room for a flags field (OCFS2_EXT_*) in the leaf nodes. 466 */ 467 struct ocfs2_extent_rec { 468 /*00*/ __le32 e_cpos; /* Offset into the file, in clusters */ 469 union { 470 __le32 e_int_clusters; /* Clusters covered by all children */ 471 struct { 472 __le16 e_leaf_clusters; /* Clusters covered by this 473 extent */ 474 __u8 e_reserved1; 475 __u8 e_flags; /* Extent flags */ 476 }; 477 }; 478 __le64 e_blkno; /* Physical disk offset, in blocks */ 479 /*10*/ 480 }; 481 482 struct ocfs2_chain_rec { 483 __le32 c_free; /* Number of free bits in this chain. */ 484 __le32 c_total; /* Number of total bits in this chain */ 485 __le64 c_blkno; /* Physical disk offset (blocks) of 1st group */ 486 }; 487 488 struct ocfs2_truncate_rec { 489 __le32 t_start; /* 1st cluster in this log */ 490 __le32 t_clusters; /* Number of total clusters covered */ 491 }; 492 493 /* 494 * On disk extent list for OCFS2 (node in the tree). Note that this 495 * is contained inside ocfs2_dinode or ocfs2_extent_block, so the 496 * offsets are relative to ocfs2_dinode.id2.i_list or 497 * ocfs2_extent_block.h_list, respectively. 498 */ 499 struct ocfs2_extent_list { 500 /*00*/ __le16 l_tree_depth; /* Extent tree depth from this 501 point. 0 means data extents 502 hang directly off this 503 header (a leaf) 504 NOTE: The high 8 bits cannot be 505 used - tree_depth is never that big. 506 */ 507 __le16 l_count; /* Number of extent records */ 508 __le16 l_next_free_rec; /* Next unused extent slot */ 509 __le16 l_reserved1; 510 __le64 l_reserved2; /* Pad to 511 sizeof(ocfs2_extent_rec) */ 512 /*10*/ struct ocfs2_extent_rec l_recs[0]; /* Extent records */ 513 }; 514 515 /* 516 * On disk allocation chain list for OCFS2. Note that this is 517 * contained inside ocfs2_dinode, so the offsets are relative to 518 * ocfs2_dinode.id2.i_chain. 519 */ 520 struct ocfs2_chain_list { 521 /*00*/ __le16 cl_cpg; /* Clusters per Block Group */ 522 __le16 cl_bpc; /* Bits per cluster */ 523 __le16 cl_count; /* Total chains in this list */ 524 __le16 cl_next_free_rec; /* Next unused chain slot */ 525 __le64 cl_reserved1; 526 /*10*/ struct ocfs2_chain_rec cl_recs[0]; /* Chain records */ 527 }; 528 529 /* 530 * On disk deallocation log for OCFS2. Note that this is 531 * contained inside ocfs2_dinode, so the offsets are relative to 532 * ocfs2_dinode.id2.i_dealloc. 533 */ 534 struct ocfs2_truncate_log { 535 /*00*/ __le16 tl_count; /* Total records in this log */ 536 __le16 tl_used; /* Number of records in use */ 537 __le32 tl_reserved1; 538 /*08*/ struct ocfs2_truncate_rec tl_recs[0]; /* Truncate records */ 539 }; 540 541 /* 542 * On disk extent block (indirect block) for OCFS2 543 */ 544 struct ocfs2_extent_block 545 { 546 /*00*/ __u8 h_signature[8]; /* Signature for verification */ 547 struct ocfs2_block_check h_check; /* Error checking */ 548 /*10*/ __le16 h_suballoc_slot; /* Slot suballocator this 549 extent_header belongs to */ 550 __le16 h_suballoc_bit; /* Bit offset in suballocator 551 block group */ 552 __le32 h_fs_generation; /* Must match super block */ 553 __le64 h_blkno; /* Offset on disk, in blocks */ 554 /*20*/ __le64 h_suballoc_loc; /* Suballocator block group this 555 eb belongs to. Only valid 556 if allocated from a 557 discontiguous block group */ 558 __le64 h_next_leaf_blk; /* Offset on disk, in blocks, 559 of next leaf header pointing 560 to data */ 561 /*30*/ struct ocfs2_extent_list h_list; /* Extent record list */ 562 /* Actual on-disk size is one block */ 563 }; 564 565 /* 566 * On disk slot map for OCFS2. This defines the contents of the "slot_map" 567 * system file. A slot is valid if it contains a node number >= 0. The 568 * value -1 (0xFFFF) is OCFS2_INVALID_SLOT. This marks a slot empty. 569 */ 570 struct ocfs2_slot_map { 571 /*00*/ __le16 sm_slots[0]; 572 /* 573 * Actual on-disk size is one block. OCFS2_MAX_SLOTS is 255, 574 * 255 * sizeof(__le16) == 512B, within the 512B block minimum blocksize. 575 */ 576 }; 577 578 struct ocfs2_extended_slot { 579 /*00*/ __u8 es_valid; 580 __u8 es_reserved1[3]; 581 __le32 es_node_num; 582 /*08*/ 583 }; 584 585 /* 586 * The extended slot map, used when OCFS2_FEATURE_INCOMPAT_EXTENDED_SLOT_MAP 587 * is set. It separates out the valid marker from the node number, and 588 * has room to grow. Unlike the old slot map, this format is defined by 589 * i_size. 590 */ 591 struct ocfs2_slot_map_extended { 592 /*00*/ struct ocfs2_extended_slot se_slots[0]; 593 /* 594 * Actual size is i_size of the slot_map system file. It should 595 * match s_max_slots * sizeof(struct ocfs2_extended_slot) 596 */ 597 }; 598 599 /* 600 * ci_stackflags is only valid if the incompat bit 601 * OCFS2_FEATURE_INCOMPAT_CLUSTERINFO is set. 602 */ 603 struct ocfs2_cluster_info { 604 /*00*/ __u8 ci_stack[OCFS2_STACK_LABEL_LEN]; 605 union { 606 __le32 ci_reserved; 607 struct { 608 __u8 ci_stackflags; 609 __u8 ci_reserved1; 610 __u8 ci_reserved2; 611 __u8 ci_reserved3; 612 }; 613 }; 614 /*08*/ __u8 ci_cluster[OCFS2_CLUSTER_NAME_LEN]; 615 /*18*/ 616 }; 617 618 /* 619 * On disk superblock for OCFS2 620 * Note that it is contained inside an ocfs2_dinode, so all offsets 621 * are relative to the start of ocfs2_dinode.id2. 622 */ 623 struct ocfs2_super_block { 624 /*00*/ __le16 s_major_rev_level; 625 __le16 s_minor_rev_level; 626 __le16 s_mnt_count; 627 __le16 s_max_mnt_count; 628 __le16 s_state; /* File system state */ 629 __le16 s_errors; /* Behaviour when detecting errors */ 630 __le32 s_checkinterval; /* Max time between checks */ 631 /*10*/ __le64 s_lastcheck; /* Time of last check */ 632 __le32 s_creator_os; /* OS */ 633 __le32 s_feature_compat; /* Compatible feature set */ 634 /*20*/ __le32 s_feature_incompat; /* Incompatible feature set */ 635 __le32 s_feature_ro_compat; /* Readonly-compatible feature set */ 636 __le64 s_root_blkno; /* Offset, in blocks, of root directory 637 dinode */ 638 /*30*/ __le64 s_system_dir_blkno; /* Offset, in blocks, of system 639 directory dinode */ 640 __le32 s_blocksize_bits; /* Blocksize for this fs */ 641 __le32 s_clustersize_bits; /* Clustersize for this fs */ 642 /*40*/ __le16 s_max_slots; /* Max number of simultaneous mounts 643 before tunefs required */ 644 __le16 s_tunefs_flag; 645 __le32 s_uuid_hash; /* hash value of uuid */ 646 __le64 s_first_cluster_group; /* Block offset of 1st cluster 647 * group header */ 648 /*50*/ __u8 s_label[OCFS2_MAX_VOL_LABEL_LEN]; /* Label for mounting, etc. */ 649 /*90*/ __u8 s_uuid[OCFS2_VOL_UUID_LEN]; /* 128-bit uuid */ 650 /*A0*/ struct ocfs2_cluster_info s_cluster_info; /* Only valid if either 651 userspace or clusterinfo 652 INCOMPAT flag set. */ 653 /*B8*/ __le16 s_xattr_inline_size; /* extended attribute inline size 654 for this fs*/ 655 __le16 s_reserved0; 656 __le32 s_dx_seed[3]; /* seed[0-2] for dx dir hash. 657 * s_uuid_hash serves as seed[3]. */ 658 /*C0*/ __le64 s_reserved2[15]; /* Fill out superblock */ 659 /*140*/ 660 661 /* 662 * NOTE: As stated above, all offsets are relative to 663 * ocfs2_dinode.id2, which is at 0xC0 in the inode. 664 * 0xC0 + 0x140 = 0x200 or 512 bytes. A superblock must fit within 665 * our smallest blocksize, which is 512 bytes. To ensure this, 666 * we reserve the space in s_reserved2. Anything past s_reserved2 667 * will not be available on the smallest blocksize. 668 */ 669 }; 670 671 /* 672 * Local allocation bitmap for OCFS2 slots 673 * Note that it exists inside an ocfs2_dinode, so all offsets are 674 * relative to the start of ocfs2_dinode.id2. 675 */ 676 struct ocfs2_local_alloc 677 { 678 /*00*/ __le32 la_bm_off; /* Starting bit offset in main bitmap */ 679 __le16 la_size; /* Size of included bitmap, in bytes */ 680 __le16 la_reserved1; 681 __le64 la_reserved2; 682 /*10*/ __u8 la_bitmap[0]; 683 }; 684 685 /* 686 * Data-in-inode header. This is only used if i_dyn_features has 687 * OCFS2_INLINE_DATA_FL set. 688 */ 689 struct ocfs2_inline_data 690 { 691 /*00*/ __le16 id_count; /* Number of bytes that can be used 692 * for data, starting at id_data */ 693 __le16 id_reserved0; 694 __le32 id_reserved1; 695 __u8 id_data[0]; /* Start of user data */ 696 }; 697 698 /* 699 * On disk inode for OCFS2 700 */ 701 struct ocfs2_dinode { 702 /*00*/ __u8 i_signature[8]; /* Signature for validation */ 703 __le32 i_generation; /* Generation number */ 704 __le16 i_suballoc_slot; /* Slot suballocator this inode 705 belongs to */ 706 __le16 i_suballoc_bit; /* Bit offset in suballocator 707 block group */ 708 /*10*/ __le16 i_links_count_hi; /* High 16 bits of links count */ 709 __le16 i_xattr_inline_size; 710 __le32 i_clusters; /* Cluster count */ 711 __le32 i_uid; /* Owner UID */ 712 __le32 i_gid; /* Owning GID */ 713 /*20*/ __le64 i_size; /* Size in bytes */ 714 __le16 i_mode; /* File mode */ 715 __le16 i_links_count; /* Links count */ 716 __le32 i_flags; /* File flags */ 717 /*30*/ __le64 i_atime; /* Access time */ 718 __le64 i_ctime; /* Creation time */ 719 /*40*/ __le64 i_mtime; /* Modification time */ 720 __le64 i_dtime; /* Deletion time */ 721 /*50*/ __le64 i_blkno; /* Offset on disk, in blocks */ 722 __le64 i_last_eb_blk; /* Pointer to last extent 723 block */ 724 /*60*/ __le32 i_fs_generation; /* Generation per fs-instance */ 725 __le32 i_atime_nsec; 726 __le32 i_ctime_nsec; 727 __le32 i_mtime_nsec; 728 /*70*/ __le32 i_attr; 729 __le16 i_orphaned_slot; /* Only valid when OCFS2_ORPHANED_FL 730 was set in i_flags */ 731 __le16 i_dyn_features; 732 __le64 i_xattr_loc; 733 /*80*/ struct ocfs2_block_check i_check; /* Error checking */ 734 /*88*/ __le64 i_dx_root; /* Pointer to dir index root block */ 735 /*90*/ __le64 i_refcount_loc; 736 __le64 i_suballoc_loc; /* Suballocator block group this 737 inode belongs to. Only valid 738 if allocated from a 739 discontiguous block group */ 740 /*A0*/ __le16 i_dio_orphaned_slot; /* only used for append dio write */ 741 __le16 i_reserved1[3]; 742 __le64 i_reserved2[2]; 743 /*B8*/ union { 744 __le64 i_pad1; /* Generic way to refer to this 745 64bit union */ 746 struct { 747 __le64 i_rdev; /* Device number */ 748 } dev1; 749 struct { /* Info for bitmap system 750 inodes */ 751 __le32 i_used; /* Bits (ie, clusters) used */ 752 __le32 i_total; /* Total bits (clusters) 753 available */ 754 } bitmap1; 755 struct { /* Info for journal system 756 inodes */ 757 __le32 ij_flags; /* Mounted, version, etc. */ 758 __le32 ij_recovery_generation; /* Incremented when the 759 journal is recovered 760 after an unclean 761 shutdown */ 762 } journal1; 763 } id1; /* Inode type dependent 1 */ 764 /*C0*/ union { 765 struct ocfs2_super_block i_super; 766 struct ocfs2_local_alloc i_lab; 767 struct ocfs2_chain_list i_chain; 768 struct ocfs2_extent_list i_list; 769 struct ocfs2_truncate_log i_dealloc; 770 struct ocfs2_inline_data i_data; 771 __u8 i_symlink[0]; 772 } id2; 773 /* Actual on-disk size is one block */ 774 }; 775 776 /* 777 * On-disk directory entry structure for OCFS2 778 * 779 * Packed as this structure could be accessed unaligned on 64-bit platforms 780 */ 781 struct ocfs2_dir_entry { 782 /*00*/ __le64 inode; /* Inode number */ 783 __le16 rec_len; /* Directory entry length */ 784 __u8 name_len; /* Name length */ 785 __u8 file_type; 786 /*0C*/ char name[OCFS2_MAX_FILENAME_LEN]; /* File name */ 787 /* Actual on-disk length specified by rec_len */ 788 } __attribute__ ((packed)); 789 790 /* 791 * Per-block record for the unindexed directory btree. This is carefully 792 * crafted so that the rec_len and name_len records of an ocfs2_dir_entry are 793 * mirrored. That way, the directory manipulation code needs a minimal amount 794 * of update. 795 * 796 * NOTE: Keep this structure aligned to a multiple of 4 bytes. 797 */ 798 struct ocfs2_dir_block_trailer { 799 /*00*/ __le64 db_compat_inode; /* Always zero. Was inode */ 800 801 __le16 db_compat_rec_len; /* Backwards compatible with 802 * ocfs2_dir_entry. */ 803 __u8 db_compat_name_len; /* Always zero. Was name_len */ 804 __u8 db_reserved0; 805 __le16 db_reserved1; 806 __le16 db_free_rec_len; /* Size of largest empty hole 807 * in this block. (unused) */ 808 /*10*/ __u8 db_signature[8]; /* Signature for verification */ 809 __le64 db_reserved2; 810 __le64 db_free_next; /* Next block in list (unused) */ 811 /*20*/ __le64 db_blkno; /* Offset on disk, in blocks */ 812 __le64 db_parent_dinode; /* dinode which owns me, in 813 blocks */ 814 /*30*/ struct ocfs2_block_check db_check; /* Error checking */ 815 /*40*/ 816 }; 817 818 /* 819 * A directory entry in the indexed tree. We don't store the full name here, 820 * but instead provide a pointer to the full dirent in the unindexed tree. 821 * 822 * We also store name_len here so as to reduce the number of leaf blocks we 823 * need to search in case of collisions. 824 */ 825 struct ocfs2_dx_entry { 826 __le32 dx_major_hash; /* Used to find logical 827 * cluster in index */ 828 __le32 dx_minor_hash; /* Lower bits used to find 829 * block in cluster */ 830 __le64 dx_dirent_blk; /* Physical block in unindexed 831 * tree holding this dirent. */ 832 }; 833 834 struct ocfs2_dx_entry_list { 835 __le32 de_reserved; 836 __le16 de_count; /* Maximum number of entries 837 * possible in de_entries */ 838 __le16 de_num_used; /* Current number of 839 * de_entries entries */ 840 struct ocfs2_dx_entry de_entries[0]; /* Indexed dir entries 841 * in a packed array of 842 * length de_num_used */ 843 }; 844 845 #define OCFS2_DX_FLAG_INLINE 0x01 846 847 /* 848 * A directory indexing block. Each indexed directory has one of these, 849 * pointed to by ocfs2_dinode. 850 * 851 * This block stores an indexed btree root, and a set of free space 852 * start-of-list pointers. 853 */ 854 struct ocfs2_dx_root_block { 855 __u8 dr_signature[8]; /* Signature for verification */ 856 struct ocfs2_block_check dr_check; /* Error checking */ 857 __le16 dr_suballoc_slot; /* Slot suballocator this 858 * block belongs to. */ 859 __le16 dr_suballoc_bit; /* Bit offset in suballocator 860 * block group */ 861 __le32 dr_fs_generation; /* Must match super block */ 862 __le64 dr_blkno; /* Offset on disk, in blocks */ 863 __le64 dr_last_eb_blk; /* Pointer to last 864 * extent block */ 865 __le32 dr_clusters; /* Clusters allocated 866 * to the indexed tree. */ 867 __u8 dr_flags; /* OCFS2_DX_FLAG_* flags */ 868 __u8 dr_reserved0; 869 __le16 dr_reserved1; 870 __le64 dr_dir_blkno; /* Pointer to parent inode */ 871 __le32 dr_num_entries; /* Total number of 872 * names stored in 873 * this directory.*/ 874 __le32 dr_reserved2; 875 __le64 dr_free_blk; /* Pointer to head of free 876 * unindexed block list. */ 877 __le64 dr_suballoc_loc; /* Suballocator block group 878 this root belongs to. 879 Only valid if allocated 880 from a discontiguous 881 block group */ 882 __le64 dr_reserved3[14]; 883 union { 884 struct ocfs2_extent_list dr_list; /* Keep this aligned to 128 885 * bits for maximum space 886 * efficiency. */ 887 struct ocfs2_dx_entry_list dr_entries; /* In-root-block list of 888 * entries. We grow out 889 * to extents if this 890 * gets too big. */ 891 }; 892 }; 893 894 /* 895 * The header of a leaf block in the indexed tree. 896 */ 897 struct ocfs2_dx_leaf { 898 __u8 dl_signature[8];/* Signature for verification */ 899 struct ocfs2_block_check dl_check; /* Error checking */ 900 __le64 dl_blkno; /* Offset on disk, in blocks */ 901 __le32 dl_fs_generation;/* Must match super block */ 902 __le32 dl_reserved0; 903 __le64 dl_reserved1; 904 struct ocfs2_dx_entry_list dl_list; 905 }; 906 907 /* 908 * Largest bitmap for a block (suballocator) group in bytes. This limit 909 * does not affect cluster groups (global allocator). Cluster group 910 * bitmaps run to the end of the block. 911 */ 912 #define OCFS2_MAX_BG_BITMAP_SIZE 256 913 914 /* 915 * On disk allocator group structure for OCFS2 916 */ 917 struct ocfs2_group_desc 918 { 919 /*00*/ __u8 bg_signature[8]; /* Signature for validation */ 920 __le16 bg_size; /* Size of included bitmap in 921 bytes. */ 922 __le16 bg_bits; /* Bits represented by this 923 group. */ 924 __le16 bg_free_bits_count; /* Free bits count */ 925 __le16 bg_chain; /* What chain I am in. */ 926 /*10*/ __le32 bg_generation; 927 __le32 bg_reserved1; 928 __le64 bg_next_group; /* Next group in my list, in 929 blocks */ 930 /*20*/ __le64 bg_parent_dinode; /* dinode which owns me, in 931 blocks */ 932 __le64 bg_blkno; /* Offset on disk, in blocks */ 933 /*30*/ struct ocfs2_block_check bg_check; /* Error checking */ 934 __le64 bg_reserved2; 935 /*40*/ union { 936 __u8 bg_bitmap[0]; 937 struct { 938 /* 939 * Block groups may be discontiguous when 940 * OCFS2_FEATURE_INCOMPAT_DISCONTIG_BG is set. 941 * The extents of a discontiguous block group are 942 * stored in bg_list. It is a flat list. 943 * l_tree_depth must always be zero. A 944 * discontiguous group is signified by a non-zero 945 * bg_list->l_next_free_rec. Only block groups 946 * can be discontiguous; Cluster groups cannot. 947 * We've never made a block group with more than 948 * 2048 blocks (256 bytes of bg_bitmap). This 949 * codifies that limit so that we can fit bg_list. 950 * bg_size of a discontiguous block group will 951 * be 256 to match bg_bitmap_filler. 952 */ 953 __u8 bg_bitmap_filler[OCFS2_MAX_BG_BITMAP_SIZE]; 954 /*140*/ struct ocfs2_extent_list bg_list; 955 }; 956 }; 957 /* Actual on-disk size is one block */ 958 }; 959 960 struct ocfs2_refcount_rec { 961 /*00*/ __le64 r_cpos; /* Physical offset, in clusters */ 962 __le32 r_clusters; /* Clusters covered by this extent */ 963 __le32 r_refcount; /* Reference count of this extent */ 964 /*10*/ 965 }; 966 #define OCFS2_32BIT_POS_MASK (0xffffffffULL) 967 968 #define OCFS2_REFCOUNT_LEAF_FL (0x00000001) 969 #define OCFS2_REFCOUNT_TREE_FL (0x00000002) 970 971 struct ocfs2_refcount_list { 972 /*00*/ __le16 rl_count; /* Maximum number of entries possible 973 in rl_records */ 974 __le16 rl_used; /* Current number of used records */ 975 __le32 rl_reserved2; 976 __le64 rl_reserved1; /* Pad to sizeof(ocfs2_refcount_record) */ 977 /*10*/ struct ocfs2_refcount_rec rl_recs[0]; /* Refcount records */ 978 }; 979 980 981 struct ocfs2_refcount_block { 982 /*00*/ __u8 rf_signature[8]; /* Signature for verification */ 983 __le16 rf_suballoc_slot; /* Slot suballocator this block 984 belongs to */ 985 __le16 rf_suballoc_bit; /* Bit offset in suballocator 986 block group */ 987 __le32 rf_fs_generation; /* Must match superblock */ 988 /*10*/ __le64 rf_blkno; /* Offset on disk, in blocks */ 989 __le64 rf_parent; /* Parent block, only valid if 990 OCFS2_REFCOUNT_LEAF_FL is set in 991 rf_flags */ 992 /*20*/ struct ocfs2_block_check rf_check; /* Error checking */ 993 __le64 rf_last_eb_blk; /* Pointer to last extent block */ 994 /*30*/ __le32 rf_count; /* Number of inodes sharing this 995 refcount tree */ 996 __le32 rf_flags; /* See the flags above */ 997 __le32 rf_clusters; /* clusters covered by refcount tree. */ 998 __le32 rf_cpos; /* cluster offset in refcount tree.*/ 999 /*40*/ __le32 rf_generation; /* generation number. all be the same 1000 * for the same refcount tree. */ 1001 __le32 rf_reserved0; 1002 __le64 rf_suballoc_loc; /* Suballocator block group this 1003 refcount block belongs to. Only 1004 valid if allocated from a 1005 discontiguous block group */ 1006 /*50*/ __le64 rf_reserved1[6]; 1007 /*80*/ union { 1008 struct ocfs2_refcount_list rf_records; /* List of refcount 1009 records */ 1010 struct ocfs2_extent_list rf_list; /* Extent record list, 1011 only valid if 1012 OCFS2_REFCOUNT_TREE_FL 1013 is set in rf_flags */ 1014 }; 1015 /* Actual on-disk size is one block */ 1016 }; 1017 1018 /* 1019 * On disk extended attribute structure for OCFS2. 1020 */ 1021 1022 /* 1023 * ocfs2_xattr_entry indicates one extend attribute. 1024 * 1025 * Note that it can be stored in inode, one block or one xattr bucket. 1026 */ 1027 struct ocfs2_xattr_entry { 1028 __le32 xe_name_hash; /* hash value of xattr prefix+suffix. */ 1029 __le16 xe_name_offset; /* byte offset from the 1st entry in the 1030 local xattr storage(inode, xattr block or 1031 xattr bucket). */ 1032 __u8 xe_name_len; /* xattr name len, doesn't include prefix. */ 1033 __u8 xe_type; /* the low 7 bits indicate the name prefix 1034 * type and the highest bit indicates whether 1035 * the EA is stored in the local storage. */ 1036 __le64 xe_value_size; /* real xattr value length. */ 1037 }; 1038 1039 /* 1040 * On disk structure for xattr header. 1041 * 1042 * One ocfs2_xattr_header describes how many ocfs2_xattr_entry records in 1043 * the local xattr storage. 1044 */ 1045 struct ocfs2_xattr_header { 1046 __le16 xh_count; /* contains the count of how 1047 many records are in the 1048 local xattr storage. */ 1049 __le16 xh_free_start; /* current offset for storing 1050 xattr. */ 1051 __le16 xh_name_value_len; /* total length of name/value 1052 length in this bucket. */ 1053 __le16 xh_num_buckets; /* Number of xattr buckets 1054 in this extent record, 1055 only valid in the first 1056 bucket. */ 1057 struct ocfs2_block_check xh_check; /* Error checking 1058 (Note, this is only 1059 used for xattr 1060 buckets. A block uses 1061 xb_check and sets 1062 this field to zero.) */ 1063 struct ocfs2_xattr_entry xh_entries[0]; /* xattr entry list. */ 1064 }; 1065 1066 /* 1067 * On disk structure for xattr value root. 1068 * 1069 * When an xattr's value is large enough, it is stored in an external 1070 * b-tree like file data. The xattr value root points to this structure. 1071 */ 1072 struct ocfs2_xattr_value_root { 1073 /*00*/ __le32 xr_clusters; /* clusters covered by xattr value. */ 1074 __le32 xr_reserved0; 1075 __le64 xr_last_eb_blk; /* Pointer to last extent block */ 1076 /*10*/ struct ocfs2_extent_list xr_list; /* Extent record list */ 1077 }; 1078 1079 /* 1080 * On disk structure for xattr tree root. 1081 * 1082 * It is used when there are too many extended attributes for one file. These 1083 * attributes will be organized and stored in an indexed-btree. 1084 */ 1085 struct ocfs2_xattr_tree_root { 1086 /*00*/ __le32 xt_clusters; /* clusters covered by xattr. */ 1087 __le32 xt_reserved0; 1088 __le64 xt_last_eb_blk; /* Pointer to last extent block */ 1089 /*10*/ struct ocfs2_extent_list xt_list; /* Extent record list */ 1090 }; 1091 1092 #define OCFS2_XATTR_INDEXED 0x1 1093 #define OCFS2_HASH_SHIFT 5 1094 #define OCFS2_XATTR_ROUND 3 1095 #define OCFS2_XATTR_SIZE(size) (((size) + OCFS2_XATTR_ROUND) & \ 1096 ~(OCFS2_XATTR_ROUND)) 1097 1098 #define OCFS2_XATTR_BUCKET_SIZE 4096 1099 #define OCFS2_XATTR_MAX_BLOCKS_PER_BUCKET (OCFS2_XATTR_BUCKET_SIZE \ 1100 / OCFS2_MIN_BLOCKSIZE) 1101 1102 /* 1103 * On disk structure for xattr block. 1104 */ 1105 struct ocfs2_xattr_block { 1106 /*00*/ __u8 xb_signature[8]; /* Signature for verification */ 1107 __le16 xb_suballoc_slot; /* Slot suballocator this 1108 block belongs to. */ 1109 __le16 xb_suballoc_bit; /* Bit offset in suballocator 1110 block group */ 1111 __le32 xb_fs_generation; /* Must match super block */ 1112 /*10*/ __le64 xb_blkno; /* Offset on disk, in blocks */ 1113 struct ocfs2_block_check xb_check; /* Error checking */ 1114 /*20*/ __le16 xb_flags; /* Indicates whether this block contains 1115 real xattr or a xattr tree. */ 1116 __le16 xb_reserved0; 1117 __le32 xb_reserved1; 1118 __le64 xb_suballoc_loc; /* Suballocator block group this 1119 xattr block belongs to. Only 1120 valid if allocated from a 1121 discontiguous block group */ 1122 /*30*/ union { 1123 struct ocfs2_xattr_header xb_header; /* xattr header if this 1124 block contains xattr */ 1125 struct ocfs2_xattr_tree_root xb_root;/* xattr tree root if this 1126 block cotains xattr 1127 tree. */ 1128 } xb_attrs; 1129 }; 1130 1131 #define OCFS2_XATTR_ENTRY_LOCAL 0x80 1132 #define OCFS2_XATTR_TYPE_MASK 0x7F 1133 static inline void ocfs2_xattr_set_local(struct ocfs2_xattr_entry *xe, 1134 int local) 1135 { 1136 if (local) 1137 xe->xe_type |= OCFS2_XATTR_ENTRY_LOCAL; 1138 else 1139 xe->xe_type &= ~OCFS2_XATTR_ENTRY_LOCAL; 1140 } 1141 1142 static inline int ocfs2_xattr_is_local(struct ocfs2_xattr_entry *xe) 1143 { 1144 return xe->xe_type & OCFS2_XATTR_ENTRY_LOCAL; 1145 } 1146 1147 static inline void ocfs2_xattr_set_type(struct ocfs2_xattr_entry *xe, int type) 1148 { 1149 xe->xe_type |= type & OCFS2_XATTR_TYPE_MASK; 1150 } 1151 1152 static inline int ocfs2_xattr_get_type(struct ocfs2_xattr_entry *xe) 1153 { 1154 return xe->xe_type & OCFS2_XATTR_TYPE_MASK; 1155 } 1156 1157 /* 1158 * On disk structures for global quota file 1159 */ 1160 1161 /* Magic numbers and known versions for global quota files */ 1162 #define OCFS2_GLOBAL_QMAGICS {\ 1163 0x0cf52470, /* USRQUOTA */ \ 1164 0x0cf52471 /* GRPQUOTA */ \ 1165 } 1166 1167 #define OCFS2_GLOBAL_QVERSIONS {\ 1168 0, \ 1169 0, \ 1170 } 1171 1172 1173 /* Each block of each quota file has a certain fixed number of bytes reserved 1174 * for OCFS2 internal use at its end. OCFS2 can use it for things like 1175 * checksums, etc. */ 1176 #define OCFS2_QBLK_RESERVED_SPACE 8 1177 1178 /* Generic header of all quota files */ 1179 struct ocfs2_disk_dqheader { 1180 __le32 dqh_magic; /* Magic number identifying file */ 1181 __le32 dqh_version; /* Quota format version */ 1182 }; 1183 1184 #define OCFS2_GLOBAL_INFO_OFF (sizeof(struct ocfs2_disk_dqheader)) 1185 1186 /* Information header of global quota file (immediately follows the generic 1187 * header) */ 1188 struct ocfs2_global_disk_dqinfo { 1189 /*00*/ __le32 dqi_bgrace; /* Grace time for space softlimit excess */ 1190 __le32 dqi_igrace; /* Grace time for inode softlimit excess */ 1191 __le32 dqi_syncms; /* Time after which we sync local changes to 1192 * global quota file */ 1193 __le32 dqi_blocks; /* Number of blocks in quota file */ 1194 /*10*/ __le32 dqi_free_blk; /* First free block in quota file */ 1195 __le32 dqi_free_entry; /* First block with free dquot entry in quota 1196 * file */ 1197 }; 1198 1199 /* Structure with global user / group information. We reserve some space 1200 * for future use. */ 1201 struct ocfs2_global_disk_dqblk { 1202 /*00*/ __le32 dqb_id; /* ID the structure belongs to */ 1203 __le32 dqb_use_count; /* Number of nodes having reference to this structure */ 1204 __le64 dqb_ihardlimit; /* absolute limit on allocated inodes */ 1205 /*10*/ __le64 dqb_isoftlimit; /* preferred inode limit */ 1206 __le64 dqb_curinodes; /* current # allocated inodes */ 1207 /*20*/ __le64 dqb_bhardlimit; /* absolute limit on disk space */ 1208 __le64 dqb_bsoftlimit; /* preferred limit on disk space */ 1209 /*30*/ __le64 dqb_curspace; /* current space occupied */ 1210 __le64 dqb_btime; /* time limit for excessive disk use */ 1211 /*40*/ __le64 dqb_itime; /* time limit for excessive inode use */ 1212 __le64 dqb_pad1; 1213 /*50*/ __le64 dqb_pad2; 1214 }; 1215 1216 /* 1217 * On-disk structures for local quota file 1218 */ 1219 1220 /* Magic numbers and known versions for local quota files */ 1221 #define OCFS2_LOCAL_QMAGICS {\ 1222 0x0cf524c0, /* USRQUOTA */ \ 1223 0x0cf524c1 /* GRPQUOTA */ \ 1224 } 1225 1226 #define OCFS2_LOCAL_QVERSIONS {\ 1227 0, \ 1228 0, \ 1229 } 1230 1231 /* Quota flags in dqinfo header */ 1232 #define OLQF_CLEAN 0x0001 /* Quota file is empty (this should be after\ 1233 * quota has been cleanly turned off) */ 1234 1235 #define OCFS2_LOCAL_INFO_OFF (sizeof(struct ocfs2_disk_dqheader)) 1236 1237 /* Information header of local quota file (immediately follows the generic 1238 * header) */ 1239 struct ocfs2_local_disk_dqinfo { 1240 __le32 dqi_flags; /* Flags for quota file */ 1241 __le32 dqi_chunks; /* Number of chunks of quota structures 1242 * with a bitmap */ 1243 __le32 dqi_blocks; /* Number of blocks allocated for quota file */ 1244 }; 1245 1246 /* Header of one chunk of a quota file */ 1247 struct ocfs2_local_disk_chunk { 1248 __le32 dqc_free; /* Number of free entries in the bitmap */ 1249 __u8 dqc_bitmap[0]; /* Bitmap of entries in the corresponding 1250 * chunk of quota file */ 1251 }; 1252 1253 /* One entry in local quota file */ 1254 struct ocfs2_local_disk_dqblk { 1255 /*00*/ __le64 dqb_id; /* id this quota applies to */ 1256 __le64 dqb_spacemod; /* Change in the amount of used space */ 1257 /*10*/ __le64 dqb_inodemod; /* Change in the amount of used inodes */ 1258 }; 1259 1260 1261 /* 1262 * The quota trailer lives at the end of each quota block. 1263 */ 1264 1265 struct ocfs2_disk_dqtrailer { 1266 /*00*/ struct ocfs2_block_check dq_check; /* Error checking */ 1267 /*08*/ /* Cannot be larger than OCFS2_QBLK_RESERVED_SPACE */ 1268 }; 1269 1270 static inline struct ocfs2_disk_dqtrailer *ocfs2_block_dqtrailer(int blocksize, 1271 void *buf) 1272 { 1273 char *ptr = buf; 1274 ptr += blocksize - OCFS2_QBLK_RESERVED_SPACE; 1275 1276 return (struct ocfs2_disk_dqtrailer *)ptr; 1277 } 1278 1279 #ifdef __KERNEL__ 1280 static inline int ocfs2_fast_symlink_chars(struct super_block *sb) 1281 { 1282 return sb->s_blocksize - 1283 offsetof(struct ocfs2_dinode, id2.i_symlink); 1284 } 1285 1286 static inline int ocfs2_max_inline_data_with_xattr(struct super_block *sb, 1287 struct ocfs2_dinode *di) 1288 { 1289 unsigned int xattrsize = le16_to_cpu(di->i_xattr_inline_size); 1290 1291 if (le16_to_cpu(di->i_dyn_features) & OCFS2_INLINE_XATTR_FL) 1292 return sb->s_blocksize - 1293 offsetof(struct ocfs2_dinode, id2.i_data.id_data) - 1294 xattrsize; 1295 else 1296 return sb->s_blocksize - 1297 offsetof(struct ocfs2_dinode, id2.i_data.id_data); 1298 } 1299 1300 static inline int ocfs2_extent_recs_per_inode(struct super_block *sb) 1301 { 1302 int size; 1303 1304 size = sb->s_blocksize - 1305 offsetof(struct ocfs2_dinode, id2.i_list.l_recs); 1306 1307 return size / sizeof(struct ocfs2_extent_rec); 1308 } 1309 1310 static inline int ocfs2_extent_recs_per_inode_with_xattr( 1311 struct super_block *sb, 1312 struct ocfs2_dinode *di) 1313 { 1314 int size; 1315 unsigned int xattrsize = le16_to_cpu(di->i_xattr_inline_size); 1316 1317 if (le16_to_cpu(di->i_dyn_features) & OCFS2_INLINE_XATTR_FL) 1318 size = sb->s_blocksize - 1319 offsetof(struct ocfs2_dinode, id2.i_list.l_recs) - 1320 xattrsize; 1321 else 1322 size = sb->s_blocksize - 1323 offsetof(struct ocfs2_dinode, id2.i_list.l_recs); 1324 1325 return size / sizeof(struct ocfs2_extent_rec); 1326 } 1327 1328 static inline int ocfs2_extent_recs_per_dx_root(struct super_block *sb) 1329 { 1330 int size; 1331 1332 size = sb->s_blocksize - 1333 offsetof(struct ocfs2_dx_root_block, dr_list.l_recs); 1334 1335 return size / sizeof(struct ocfs2_extent_rec); 1336 } 1337 1338 static inline int ocfs2_chain_recs_per_inode(struct super_block *sb) 1339 { 1340 int size; 1341 1342 size = sb->s_blocksize - 1343 offsetof(struct ocfs2_dinode, id2.i_chain.cl_recs); 1344 1345 return size / sizeof(struct ocfs2_chain_rec); 1346 } 1347 1348 static inline u16 ocfs2_extent_recs_per_eb(struct super_block *sb) 1349 { 1350 int size; 1351 1352 size = sb->s_blocksize - 1353 offsetof(struct ocfs2_extent_block, h_list.l_recs); 1354 1355 return size / sizeof(struct ocfs2_extent_rec); 1356 } 1357 1358 static inline u16 ocfs2_extent_recs_per_gd(struct super_block *sb) 1359 { 1360 int size; 1361 1362 size = sb->s_blocksize - 1363 offsetof(struct ocfs2_group_desc, bg_list.l_recs); 1364 1365 return size / sizeof(struct ocfs2_extent_rec); 1366 } 1367 1368 static inline int ocfs2_dx_entries_per_leaf(struct super_block *sb) 1369 { 1370 int size; 1371 1372 size = sb->s_blocksize - 1373 offsetof(struct ocfs2_dx_leaf, dl_list.de_entries); 1374 1375 return size / sizeof(struct ocfs2_dx_entry); 1376 } 1377 1378 static inline int ocfs2_dx_entries_per_root(struct super_block *sb) 1379 { 1380 int size; 1381 1382 size = sb->s_blocksize - 1383 offsetof(struct ocfs2_dx_root_block, dr_entries.de_entries); 1384 1385 return size / sizeof(struct ocfs2_dx_entry); 1386 } 1387 1388 static inline u16 ocfs2_local_alloc_size(struct super_block *sb) 1389 { 1390 u16 size; 1391 1392 size = sb->s_blocksize - 1393 offsetof(struct ocfs2_dinode, id2.i_lab.la_bitmap); 1394 1395 return size; 1396 } 1397 1398 static inline int ocfs2_group_bitmap_size(struct super_block *sb, 1399 int suballocator, 1400 u32 feature_incompat) 1401 { 1402 int size = sb->s_blocksize - 1403 offsetof(struct ocfs2_group_desc, bg_bitmap); 1404 1405 /* 1406 * The cluster allocator uses the entire block. Suballocators have 1407 * never used more than OCFS2_MAX_BG_BITMAP_SIZE. Unfortunately, older 1408 * code expects bg_size set to the maximum. Thus we must keep 1409 * bg_size as-is unless discontig_bg is enabled. 1410 */ 1411 if (suballocator && 1412 (feature_incompat & OCFS2_FEATURE_INCOMPAT_DISCONTIG_BG)) 1413 size = OCFS2_MAX_BG_BITMAP_SIZE; 1414 1415 return size; 1416 } 1417 1418 static inline int ocfs2_truncate_recs_per_inode(struct super_block *sb) 1419 { 1420 int size; 1421 1422 size = sb->s_blocksize - 1423 offsetof(struct ocfs2_dinode, id2.i_dealloc.tl_recs); 1424 1425 return size / sizeof(struct ocfs2_truncate_rec); 1426 } 1427 1428 static inline u64 ocfs2_backup_super_blkno(struct super_block *sb, int index) 1429 { 1430 u64 offset = OCFS2_BACKUP_SB_START; 1431 1432 if (index >= 0 && index < OCFS2_MAX_BACKUP_SUPERBLOCKS) { 1433 offset <<= (2 * index); 1434 offset >>= sb->s_blocksize_bits; 1435 return offset; 1436 } 1437 1438 return 0; 1439 1440 } 1441 1442 static inline u16 ocfs2_xattr_recs_per_xb(struct super_block *sb) 1443 { 1444 int size; 1445 1446 size = sb->s_blocksize - 1447 offsetof(struct ocfs2_xattr_block, 1448 xb_attrs.xb_root.xt_list.l_recs); 1449 1450 return size / sizeof(struct ocfs2_extent_rec); 1451 } 1452 1453 static inline u16 ocfs2_extent_recs_per_rb(struct super_block *sb) 1454 { 1455 int size; 1456 1457 size = sb->s_blocksize - 1458 offsetof(struct ocfs2_refcount_block, rf_list.l_recs); 1459 1460 return size / sizeof(struct ocfs2_extent_rec); 1461 } 1462 1463 static inline u16 ocfs2_refcount_recs_per_rb(struct super_block *sb) 1464 { 1465 int size; 1466 1467 size = sb->s_blocksize - 1468 offsetof(struct ocfs2_refcount_block, rf_records.rl_recs); 1469 1470 return size / sizeof(struct ocfs2_refcount_rec); 1471 } 1472 1473 static inline u32 1474 ocfs2_get_ref_rec_low_cpos(const struct ocfs2_refcount_rec *rec) 1475 { 1476 return le64_to_cpu(rec->r_cpos) & OCFS2_32BIT_POS_MASK; 1477 } 1478 #else 1479 static inline int ocfs2_fast_symlink_chars(int blocksize) 1480 { 1481 return blocksize - offsetof(struct ocfs2_dinode, id2.i_symlink); 1482 } 1483 1484 static inline int ocfs2_max_inline_data_with_xattr(int blocksize, 1485 struct ocfs2_dinode *di) 1486 { 1487 if (di && (di->i_dyn_features & OCFS2_INLINE_XATTR_FL)) 1488 return blocksize - 1489 offsetof(struct ocfs2_dinode, id2.i_data.id_data) - 1490 di->i_xattr_inline_size; 1491 else 1492 return blocksize - 1493 offsetof(struct ocfs2_dinode, id2.i_data.id_data); 1494 } 1495 1496 static inline int ocfs2_extent_recs_per_inode(int blocksize) 1497 { 1498 int size; 1499 1500 size = blocksize - 1501 offsetof(struct ocfs2_dinode, id2.i_list.l_recs); 1502 1503 return size / sizeof(struct ocfs2_extent_rec); 1504 } 1505 1506 static inline int ocfs2_chain_recs_per_inode(int blocksize) 1507 { 1508 int size; 1509 1510 size = blocksize - 1511 offsetof(struct ocfs2_dinode, id2.i_chain.cl_recs); 1512 1513 return size / sizeof(struct ocfs2_chain_rec); 1514 } 1515 1516 static inline int ocfs2_extent_recs_per_eb(int blocksize) 1517 { 1518 int size; 1519 1520 size = blocksize - 1521 offsetof(struct ocfs2_extent_block, h_list.l_recs); 1522 1523 return size / sizeof(struct ocfs2_extent_rec); 1524 } 1525 1526 static inline int ocfs2_extent_recs_per_gd(int blocksize) 1527 { 1528 int size; 1529 1530 size = blocksize - 1531 offsetof(struct ocfs2_group_desc, bg_list.l_recs); 1532 1533 return size / sizeof(struct ocfs2_extent_rec); 1534 } 1535 1536 static inline int ocfs2_local_alloc_size(int blocksize) 1537 { 1538 int size; 1539 1540 size = blocksize - 1541 offsetof(struct ocfs2_dinode, id2.i_lab.la_bitmap); 1542 1543 return size; 1544 } 1545 1546 static inline int ocfs2_group_bitmap_size(int blocksize, 1547 int suballocator, 1548 uint32_t feature_incompat) 1549 { 1550 int size = sb->s_blocksize - 1551 offsetof(struct ocfs2_group_desc, bg_bitmap); 1552 1553 /* 1554 * The cluster allocator uses the entire block. Suballocators have 1555 * never used more than OCFS2_MAX_BG_BITMAP_SIZE. Unfortunately, older 1556 * code expects bg_size set to the maximum. Thus we must keep 1557 * bg_size as-is unless discontig_bg is enabled. 1558 */ 1559 if (suballocator && 1560 (feature_incompat & OCFS2_FEATURE_INCOMPAT_DISCONTIG_BG)) 1561 size = OCFS2_MAX_BG_BITMAP_SIZE; 1562 1563 return size; 1564 } 1565 1566 static inline int ocfs2_truncate_recs_per_inode(int blocksize) 1567 { 1568 int size; 1569 1570 size = blocksize - 1571 offsetof(struct ocfs2_dinode, id2.i_dealloc.tl_recs); 1572 1573 return size / sizeof(struct ocfs2_truncate_rec); 1574 } 1575 1576 static inline uint64_t ocfs2_backup_super_blkno(int blocksize, int index) 1577 { 1578 uint64_t offset = OCFS2_BACKUP_SB_START; 1579 1580 if (index >= 0 && index < OCFS2_MAX_BACKUP_SUPERBLOCKS) { 1581 offset <<= (2 * index); 1582 offset /= blocksize; 1583 return offset; 1584 } 1585 1586 return 0; 1587 } 1588 1589 static inline int ocfs2_xattr_recs_per_xb(int blocksize) 1590 { 1591 int size; 1592 1593 size = blocksize - 1594 offsetof(struct ocfs2_xattr_block, 1595 xb_attrs.xb_root.xt_list.l_recs); 1596 1597 return size / sizeof(struct ocfs2_extent_rec); 1598 } 1599 #endif /* __KERNEL__ */ 1600 1601 1602 static inline int ocfs2_system_inode_is_global(int type) 1603 { 1604 return ((type >= 0) && 1605 (type <= OCFS2_LAST_GLOBAL_SYSTEM_INODE)); 1606 } 1607 1608 static inline int ocfs2_sprintf_system_inode_name(char *buf, int len, 1609 int type, int slot) 1610 { 1611 int chars; 1612 1613 /* 1614 * Global system inodes can only have one copy. Everything 1615 * after OCFS2_LAST_GLOBAL_SYSTEM_INODE in the system inode 1616 * list has a copy per slot. 1617 */ 1618 if (type <= OCFS2_LAST_GLOBAL_SYSTEM_INODE) 1619 chars = snprintf(buf, len, "%s", 1620 ocfs2_system_inodes[type].si_name); 1621 else 1622 chars = snprintf(buf, len, 1623 ocfs2_system_inodes[type].si_name, 1624 slot); 1625 1626 return chars; 1627 } 1628 1629 static inline void ocfs2_set_de_type(struct ocfs2_dir_entry *de, 1630 umode_t mode) 1631 { 1632 de->file_type = ocfs2_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 1633 } 1634 1635 static inline int ocfs2_gd_is_discontig(struct ocfs2_group_desc *gd) 1636 { 1637 if ((offsetof(struct ocfs2_group_desc, bg_bitmap) + 1638 le16_to_cpu(gd->bg_size)) != 1639 offsetof(struct ocfs2_group_desc, bg_list)) 1640 return 0; 1641 /* 1642 * Only valid to check l_next_free_rec if 1643 * bg_bitmap + bg_size == bg_list. 1644 */ 1645 if (!gd->bg_list.l_next_free_rec) 1646 return 0; 1647 return 1; 1648 } 1649 #endif /* _OCFS2_FS_H */ 1650 1651