1 /* -*- mode: c; c-basic-offset: 8; -*-
2 * vim: noexpandtab sw=8 ts=8 sts=0:
6 * On-disk structures for OCFS2.
8 * Copyright (C) 2002, 2004 Oracle. All rights reserved.
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.
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.
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.
29 #define OCFS2_MAJOR_REV_LEVEL 0
30 #define OCFS2_MINOR_REV_LEVEL 90
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.
38 * All other structures are found from the superblock information.
40 * OCFS2_SUPER_BLOCK_BLKNO is in blocks, not sectors. eg, for a
41 * blocksize of 2K, it is 4096 bytes into disk.
43 #define OCFS2_SUPER_BLOCK_BLKNO 2
46 * Cluster size limits. The maximum is kept arbitrarily at 1 MB, and could
49 #define OCFS2_MIN_CLUSTERSIZE 4096
50 #define OCFS2_MAX_CLUSTERSIZE 1048576
53 * Blocks cannot be bigger than clusters, so the maximum blocksize is the
54 * minimum cluster size.
56 #define OCFS2_MIN_BLOCKSIZE 512
57 #define OCFS2_MAX_BLOCKSIZE OCFS2_MIN_CLUSTERSIZE
59 /* Filesystem magic number */
60 #define OCFS2_SUPER_MAGIC 0x7461636f
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"
69 /* Compatibility flags */
70 #define OCFS2_HAS_COMPAT_FEATURE(sb,mask) \
71 ( OCFS2_SB(sb)->s_feature_compat & (mask) )
72 #define OCFS2_HAS_RO_COMPAT_FEATURE(sb,mask) \
73 ( OCFS2_SB(sb)->s_feature_ro_compat & (mask) )
74 #define OCFS2_HAS_INCOMPAT_FEATURE(sb,mask) \
75 ( OCFS2_SB(sb)->s_feature_incompat & (mask) )
76 #define OCFS2_SET_COMPAT_FEATURE(sb,mask) \
77 OCFS2_SB(sb)->s_feature_compat |= (mask)
78 #define OCFS2_SET_RO_COMPAT_FEATURE(sb,mask) \
79 OCFS2_SB(sb)->s_feature_ro_compat |= (mask)
80 #define OCFS2_SET_INCOMPAT_FEATURE(sb,mask) \
81 OCFS2_SB(sb)->s_feature_incompat |= (mask)
82 #define OCFS2_CLEAR_COMPAT_FEATURE(sb,mask) \
83 OCFS2_SB(sb)->s_feature_compat &= ~(mask)
84 #define OCFS2_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
85 OCFS2_SB(sb)->s_feature_ro_compat &= ~(mask)
86 #define OCFS2_CLEAR_INCOMPAT_FEATURE(sb,mask) \
87 OCFS2_SB(sb)->s_feature_incompat &= ~(mask)
89 #define OCFS2_FEATURE_COMPAT_SUPP OCFS2_FEATURE_COMPAT_BACKUP_SB
90 #define OCFS2_FEATURE_INCOMPAT_SUPP (OCFS2_FEATURE_INCOMPAT_LOCAL_MOUNT \
91 | OCFS2_FEATURE_INCOMPAT_SPARSE_ALLOC \
92 | OCFS2_FEATURE_INCOMPAT_INLINE_DATA \
93 | OCFS2_FEATURE_INCOMPAT_EXTENDED_SLOT_MAP \
94 | OCFS2_FEATURE_INCOMPAT_USERSPACE_STACK \
95 | OCFS2_FEATURE_INCOMPAT_XATTR)
96 #define OCFS2_FEATURE_RO_COMPAT_SUPP OCFS2_FEATURE_RO_COMPAT_UNWRITTEN
99 * Heartbeat-only devices are missing journals and other files. The
100 * filesystem driver can't load them, but the library can. Never put
101 * this in OCFS2_FEATURE_INCOMPAT_SUPP, *ever*.
103 #define OCFS2_FEATURE_INCOMPAT_HEARTBEAT_DEV 0x0002
106 * tunefs sets this incompat flag before starting the resize and clears it
107 * at the end. This flag protects users from inadvertently mounting the fs
108 * after an aborted run without fsck-ing.
110 #define OCFS2_FEATURE_INCOMPAT_RESIZE_INPROG 0x0004
112 /* Used to denote a non-clustered volume */
113 #define OCFS2_FEATURE_INCOMPAT_LOCAL_MOUNT 0x0008
115 /* Support for sparse allocation in b-trees */
116 #define OCFS2_FEATURE_INCOMPAT_SPARSE_ALLOC 0x0010
119 * Tunefs sets this incompat flag before starting an operation which
120 * would require cleanup on abort. This is done to protect users from
121 * inadvertently mounting the fs after an aborted run without
124 * s_tunefs_flags on the super block describes precisely which
125 * operations were in progress.
127 #define OCFS2_FEATURE_INCOMPAT_TUNEFS_INPROG 0x0020
129 /* Support for data packed into inode blocks */
130 #define OCFS2_FEATURE_INCOMPAT_INLINE_DATA 0x0040
133 * Support for alternate, userspace cluster stacks. If set, the superblock
134 * field s_cluster_info contains a tag for the alternate stack in use as
135 * well as the name of the cluster being joined.
136 * mount.ocfs2 must pass in a matching stack name.
138 * If not set, the classic stack will be used. This is compatbile with
139 * all older versions.
141 #define OCFS2_FEATURE_INCOMPAT_USERSPACE_STACK 0x0080
143 /* Support for the extended slot map */
144 #define OCFS2_FEATURE_INCOMPAT_EXTENDED_SLOT_MAP 0x100
146 /* Support for extended attributes */
147 #define OCFS2_FEATURE_INCOMPAT_XATTR 0x0200
150 * backup superblock flag is used to indicate that this volume
151 * has backup superblocks.
153 #define OCFS2_FEATURE_COMPAT_BACKUP_SB 0x0001
156 * Unwritten extents support.
158 #define OCFS2_FEATURE_RO_COMPAT_UNWRITTEN 0x0001
160 /* The byte offset of the first backup block will be 1G.
161 * The following will be 4G, 16G, 64G, 256G and 1T.
163 #define OCFS2_BACKUP_SB_START 1 << 30
165 /* the max backup superblock nums */
166 #define OCFS2_MAX_BACKUP_SUPERBLOCKS 6
169 * Flags on ocfs2_super_block.s_tunefs_flags
171 #define OCFS2_TUNEFS_INPROG_REMOVE_SLOT 0x0001 /* Removing slots */
174 * Flags on ocfs2_dinode.i_flags
176 #define OCFS2_VALID_FL (0x00000001) /* Inode is valid */
177 #define OCFS2_UNUSED2_FL (0x00000002)
178 #define OCFS2_ORPHANED_FL (0x00000004) /* On the orphan list */
179 #define OCFS2_UNUSED3_FL (0x00000008)
180 /* System inode flags */
181 #define OCFS2_SYSTEM_FL (0x00000010) /* System inode */
182 #define OCFS2_SUPER_BLOCK_FL (0x00000020) /* Super block */
183 #define OCFS2_LOCAL_ALLOC_FL (0x00000040) /* Slot local alloc bitmap */
184 #define OCFS2_BITMAP_FL (0x00000080) /* Allocation bitmap */
185 #define OCFS2_JOURNAL_FL (0x00000100) /* Slot local journal */
186 #define OCFS2_HEARTBEAT_FL (0x00000200) /* Heartbeat area */
187 #define OCFS2_CHAIN_FL (0x00000400) /* Chain allocator */
188 #define OCFS2_DEALLOC_FL (0x00000800) /* Truncate log */
191 * Flags on ocfs2_dinode.i_dyn_features
193 * These can change much more often than i_flags. When adding flags,
194 * keep in mind that i_dyn_features is only 16 bits wide.
196 #define OCFS2_INLINE_DATA_FL (0x0001) /* Data stored in inode block */
197 #define OCFS2_HAS_XATTR_FL (0x0002)
198 #define OCFS2_INLINE_XATTR_FL (0x0004)
199 #define OCFS2_INDEXED_DIR_FL (0x0008)
201 /* Inode attributes, keep in sync with EXT2 */
202 #define OCFS2_SECRM_FL (0x00000001) /* Secure deletion */
203 #define OCFS2_UNRM_FL (0x00000002) /* Undelete */
204 #define OCFS2_COMPR_FL (0x00000004) /* Compress file */
205 #define OCFS2_SYNC_FL (0x00000008) /* Synchronous updates */
206 #define OCFS2_IMMUTABLE_FL (0x00000010) /* Immutable file */
207 #define OCFS2_APPEND_FL (0x00000020) /* writes to file may only append */
208 #define OCFS2_NODUMP_FL (0x00000040) /* do not dump file */
209 #define OCFS2_NOATIME_FL (0x00000080) /* do not update atime */
210 #define OCFS2_DIRSYNC_FL (0x00010000) /* dirsync behaviour (directories only) */
212 #define OCFS2_FL_VISIBLE (0x000100FF) /* User visible flags */
213 #define OCFS2_FL_MODIFIABLE (0x000100FF) /* User modifiable flags */
216 * Extent record flags (e_node.leaf.flags)
218 #define OCFS2_EXT_UNWRITTEN (0x01) /* Extent is allocated but
224 #define OCFS2_IOC_GETFLAGS _IOR('f', 1, long)
225 #define OCFS2_IOC_SETFLAGS _IOW('f', 2, long)
226 #define OCFS2_IOC32_GETFLAGS _IOR('f', 1, int)
227 #define OCFS2_IOC32_SETFLAGS _IOW('f', 2, int)
230 * Space reservation / allocation / free ioctls and argument structure
231 * are designed to be compatible with XFS.
233 * ALLOCSP* and FREESP* are not and will never be supported, but are
234 * included here for completeness.
236 struct ocfs2_space_resv {
240 __s64 l_len; /* len == 0 means until end of file */
243 __s32 l_pad[4]; /* reserve area */
246 #define OCFS2_IOC_ALLOCSP _IOW ('X', 10, struct ocfs2_space_resv)
247 #define OCFS2_IOC_FREESP _IOW ('X', 11, struct ocfs2_space_resv)
248 #define OCFS2_IOC_RESVSP _IOW ('X', 40, struct ocfs2_space_resv)
249 #define OCFS2_IOC_UNRESVSP _IOW ('X', 41, struct ocfs2_space_resv)
250 #define OCFS2_IOC_ALLOCSP64 _IOW ('X', 36, struct ocfs2_space_resv)
251 #define OCFS2_IOC_FREESP64 _IOW ('X', 37, struct ocfs2_space_resv)
252 #define OCFS2_IOC_RESVSP64 _IOW ('X', 42, struct ocfs2_space_resv)
253 #define OCFS2_IOC_UNRESVSP64 _IOW ('X', 43, struct ocfs2_space_resv)
255 /* Used to pass group descriptor data when online resize is done */
256 struct ocfs2_new_group_input {
257 __u64 group; /* Group descriptor's blkno. */
258 __u32 clusters; /* Total number of clusters in this group */
259 __u32 frees; /* Total free clusters in this group */
260 __u16 chain; /* Chain for this group */
265 #define OCFS2_IOC_GROUP_EXTEND _IOW('o', 1, int)
266 #define OCFS2_IOC_GROUP_ADD _IOW('o', 2,struct ocfs2_new_group_input)
267 #define OCFS2_IOC_GROUP_ADD64 _IOW('o', 3,struct ocfs2_new_group_input)
270 * Journal Flags (ocfs2_dinode.id1.journal1.i_flags)
272 #define OCFS2_JOURNAL_DIRTY_FL (0x00000001) /* Journal needs recovery */
275 * superblock s_state flags
277 #define OCFS2_ERROR_FS (0x00000001) /* FS saw errors */
279 /* Limit of space in ocfs2_dir_entry */
280 #define OCFS2_MAX_FILENAME_LEN 255
282 /* Maximum slots on an ocfs2 file system */
283 #define OCFS2_MAX_SLOTS 255
285 /* Slot map indicator for an empty slot */
286 #define OCFS2_INVALID_SLOT -1
288 #define OCFS2_VOL_UUID_LEN 16
289 #define OCFS2_MAX_VOL_LABEL_LEN 64
291 /* The alternate, userspace stack fields */
292 #define OCFS2_STACK_LABEL_LEN 4
293 #define OCFS2_CLUSTER_NAME_LEN 16
295 /* Journal limits (in bytes) */
296 #define OCFS2_MIN_JOURNAL_SIZE (4 * 1024 * 1024)
299 * Default local alloc size (in megabytes)
301 * The value chosen should be such that most allocations, including new
302 * block groups, use local alloc.
304 #define OCFS2_DEFAULT_LOCAL_ALLOC_SIZE 8
307 * Inline extended attribute size (in bytes)
308 * The value chosen should be aligned to 16 byte boundaries.
310 #define OCFS2_MIN_XATTR_INLINE_SIZE 256
312 struct ocfs2_system_inode_info {
318 /* System file index */
320 BAD_BLOCK_SYSTEM_INODE = 0,
321 GLOBAL_INODE_ALLOC_SYSTEM_INODE,
322 SLOT_MAP_SYSTEM_INODE,
323 #define OCFS2_FIRST_ONLINE_SYSTEM_INODE SLOT_MAP_SYSTEM_INODE
324 HEARTBEAT_SYSTEM_INODE,
325 GLOBAL_BITMAP_SYSTEM_INODE,
326 #define OCFS2_LAST_GLOBAL_SYSTEM_INODE GLOBAL_BITMAP_SYSTEM_INODE
327 ORPHAN_DIR_SYSTEM_INODE,
328 EXTENT_ALLOC_SYSTEM_INODE,
329 INODE_ALLOC_SYSTEM_INODE,
330 JOURNAL_SYSTEM_INODE,
331 LOCAL_ALLOC_SYSTEM_INODE,
332 TRUNCATE_LOG_SYSTEM_INODE,
336 static struct ocfs2_system_inode_info ocfs2_system_inodes[NUM_SYSTEM_INODES] = {
337 /* Global system inodes (single copy) */
338 /* The first two are only used from userspace mfks/tunefs */
339 [BAD_BLOCK_SYSTEM_INODE] = { "bad_blocks", 0, S_IFREG | 0644 },
340 [GLOBAL_INODE_ALLOC_SYSTEM_INODE] = { "global_inode_alloc", OCFS2_BITMAP_FL | OCFS2_CHAIN_FL, S_IFREG | 0644 },
342 /* These are used by the running filesystem */
343 [SLOT_MAP_SYSTEM_INODE] = { "slot_map", 0, S_IFREG | 0644 },
344 [HEARTBEAT_SYSTEM_INODE] = { "heartbeat", OCFS2_HEARTBEAT_FL, S_IFREG | 0644 },
345 [GLOBAL_BITMAP_SYSTEM_INODE] = { "global_bitmap", 0, S_IFREG | 0644 },
347 /* Slot-specific system inodes (one copy per slot) */
348 [ORPHAN_DIR_SYSTEM_INODE] = { "orphan_dir:%04d", 0, S_IFDIR | 0755 },
349 [EXTENT_ALLOC_SYSTEM_INODE] = { "extent_alloc:%04d", OCFS2_BITMAP_FL | OCFS2_CHAIN_FL, S_IFREG | 0644 },
350 [INODE_ALLOC_SYSTEM_INODE] = { "inode_alloc:%04d", OCFS2_BITMAP_FL | OCFS2_CHAIN_FL, S_IFREG | 0644 },
351 [JOURNAL_SYSTEM_INODE] = { "journal:%04d", OCFS2_JOURNAL_FL, S_IFREG | 0644 },
352 [LOCAL_ALLOC_SYSTEM_INODE] = { "local_alloc:%04d", OCFS2_BITMAP_FL | OCFS2_LOCAL_ALLOC_FL, S_IFREG | 0644 },
353 [TRUNCATE_LOG_SYSTEM_INODE] = { "truncate_log:%04d", OCFS2_DEALLOC_FL, S_IFREG | 0644 }
356 /* Parameter passed from mount.ocfs2 to module */
357 #define OCFS2_HB_NONE "heartbeat=none"
358 #define OCFS2_HB_LOCAL "heartbeat=local"
361 * OCFS2 directory file types. Only the low 3 bits are used. The
362 * other bits are reserved for now.
364 #define OCFS2_FT_UNKNOWN 0
365 #define OCFS2_FT_REG_FILE 1
366 #define OCFS2_FT_DIR 2
367 #define OCFS2_FT_CHRDEV 3
368 #define OCFS2_FT_BLKDEV 4
369 #define OCFS2_FT_FIFO 5
370 #define OCFS2_FT_SOCK 6
371 #define OCFS2_FT_SYMLINK 7
373 #define OCFS2_FT_MAX 8
376 * OCFS2_DIR_PAD defines the directory entries boundaries
378 * NOTE: It must be a multiple of 4
380 #define OCFS2_DIR_PAD 4
381 #define OCFS2_DIR_ROUND (OCFS2_DIR_PAD - 1)
382 #define OCFS2_DIR_MEMBER_LEN offsetof(struct ocfs2_dir_entry, name)
383 #define OCFS2_DIR_REC_LEN(name_len) (((name_len) + OCFS2_DIR_MEMBER_LEN + \
387 #define OCFS2_LINK_MAX 32000
390 static unsigned char ocfs2_type_by_mode[S_IFMT >> S_SHIFT] = {
391 [S_IFREG >> S_SHIFT] = OCFS2_FT_REG_FILE,
392 [S_IFDIR >> S_SHIFT] = OCFS2_FT_DIR,
393 [S_IFCHR >> S_SHIFT] = OCFS2_FT_CHRDEV,
394 [S_IFBLK >> S_SHIFT] = OCFS2_FT_BLKDEV,
395 [S_IFIFO >> S_SHIFT] = OCFS2_FT_FIFO,
396 [S_IFSOCK >> S_SHIFT] = OCFS2_FT_SOCK,
397 [S_IFLNK >> S_SHIFT] = OCFS2_FT_SYMLINK,
404 #define OCFS2_RAW_SB(dinode) (&((dinode)->id2.i_super))
407 * On disk extent record for OCFS2
408 * It describes a range of clusters on disk.
410 * Length fields are divided into interior and leaf node versions.
411 * This leaves room for a flags field (OCFS2_EXT_*) in the leaf nodes.
413 struct ocfs2_extent_rec {
414 /*00*/ __le32 e_cpos; /* Offset into the file, in clusters */
416 __le32 e_int_clusters; /* Clusters covered by all children */
418 __le16 e_leaf_clusters; /* Clusters covered by this
421 __u8 e_flags; /* Extent flags */
424 __le64 e_blkno; /* Physical disk offset, in blocks */
428 struct ocfs2_chain_rec {
429 __le32 c_free; /* Number of free bits in this chain. */
430 __le32 c_total; /* Number of total bits in this chain */
431 __le64 c_blkno; /* Physical disk offset (blocks) of 1st group */
434 struct ocfs2_truncate_rec {
435 __le32 t_start; /* 1st cluster in this log */
436 __le32 t_clusters; /* Number of total clusters covered */
440 * On disk extent list for OCFS2 (node in the tree). Note that this
441 * is contained inside ocfs2_dinode or ocfs2_extent_block, so the
442 * offsets are relative to ocfs2_dinode.id2.i_list or
443 * ocfs2_extent_block.h_list, respectively.
445 struct ocfs2_extent_list {
446 /*00*/ __le16 l_tree_depth; /* Extent tree depth from this
447 point. 0 means data extents
448 hang directly off this
450 NOTE: The high 8 bits cannot be
451 used - tree_depth is never that big.
453 __le16 l_count; /* Number of extent records */
454 __le16 l_next_free_rec; /* Next unused extent slot */
456 __le64 l_reserved2; /* Pad to
457 sizeof(ocfs2_extent_rec) */
458 /*10*/ struct ocfs2_extent_rec l_recs[0]; /* Extent records */
462 * On disk allocation chain list for OCFS2. Note that this is
463 * contained inside ocfs2_dinode, so the offsets are relative to
464 * ocfs2_dinode.id2.i_chain.
466 struct ocfs2_chain_list {
467 /*00*/ __le16 cl_cpg; /* Clusters per Block Group */
468 __le16 cl_bpc; /* Bits per cluster */
469 __le16 cl_count; /* Total chains in this list */
470 __le16 cl_next_free_rec; /* Next unused chain slot */
472 /*10*/ struct ocfs2_chain_rec cl_recs[0]; /* Chain records */
476 * On disk deallocation log for OCFS2. Note that this is
477 * contained inside ocfs2_dinode, so the offsets are relative to
478 * ocfs2_dinode.id2.i_dealloc.
480 struct ocfs2_truncate_log {
481 /*00*/ __le16 tl_count; /* Total records in this log */
482 __le16 tl_used; /* Number of records in use */
484 /*08*/ struct ocfs2_truncate_rec tl_recs[0]; /* Truncate records */
488 * On disk extent block (indirect block) for OCFS2
490 struct ocfs2_extent_block
492 /*00*/ __u8 h_signature[8]; /* Signature for verification */
494 /*10*/ __le16 h_suballoc_slot; /* Slot suballocator this
495 extent_header belongs to */
496 __le16 h_suballoc_bit; /* Bit offset in suballocator
498 __le32 h_fs_generation; /* Must match super block */
499 __le64 h_blkno; /* Offset on disk, in blocks */
500 /*20*/ __le64 h_reserved3;
501 __le64 h_next_leaf_blk; /* Offset on disk, in blocks,
502 of next leaf header pointing
504 /*30*/ struct ocfs2_extent_list h_list; /* Extent record list */
505 /* Actual on-disk size is one block */
509 * On disk slot map for OCFS2. This defines the contents of the "slot_map"
510 * system file. A slot is valid if it contains a node number >= 0. The
511 * value -1 (0xFFFF) is OCFS2_INVALID_SLOT. This marks a slot empty.
513 struct ocfs2_slot_map {
514 /*00*/ __le16 sm_slots[0];
516 * Actual on-disk size is one block. OCFS2_MAX_SLOTS is 255,
517 * 255 * sizeof(__le16) == 512B, within the 512B block minimum blocksize.
521 struct ocfs2_extended_slot {
522 /*00*/ __u8 es_valid;
523 __u8 es_reserved1[3];
529 * The extended slot map, used when OCFS2_FEATURE_INCOMPAT_EXTENDED_SLOT_MAP
530 * is set. It separates out the valid marker from the node number, and
531 * has room to grow. Unlike the old slot map, this format is defined by
534 struct ocfs2_slot_map_extended {
535 /*00*/ struct ocfs2_extended_slot se_slots[0];
537 * Actual size is i_size of the slot_map system file. It should
538 * match s_max_slots * sizeof(struct ocfs2_extended_slot)
542 struct ocfs2_cluster_info {
543 /*00*/ __u8 ci_stack[OCFS2_STACK_LABEL_LEN];
545 /*08*/ __u8 ci_cluster[OCFS2_CLUSTER_NAME_LEN];
550 * On disk superblock for OCFS2
551 * Note that it is contained inside an ocfs2_dinode, so all offsets
552 * are relative to the start of ocfs2_dinode.id2.
554 struct ocfs2_super_block {
555 /*00*/ __le16 s_major_rev_level;
556 __le16 s_minor_rev_level;
558 __le16 s_max_mnt_count;
559 __le16 s_state; /* File system state */
560 __le16 s_errors; /* Behaviour when detecting errors */
561 __le32 s_checkinterval; /* Max time between checks */
562 /*10*/ __le64 s_lastcheck; /* Time of last check */
563 __le32 s_creator_os; /* OS */
564 __le32 s_feature_compat; /* Compatible feature set */
565 /*20*/ __le32 s_feature_incompat; /* Incompatible feature set */
566 __le32 s_feature_ro_compat; /* Readonly-compatible feature set */
567 __le64 s_root_blkno; /* Offset, in blocks, of root directory
569 /*30*/ __le64 s_system_dir_blkno; /* Offset, in blocks, of system
571 __le32 s_blocksize_bits; /* Blocksize for this fs */
572 __le32 s_clustersize_bits; /* Clustersize for this fs */
573 /*40*/ __le16 s_max_slots; /* Max number of simultaneous mounts
574 before tunefs required */
575 __le16 s_tunefs_flag;
576 __le32 s_uuid_hash; /* hash value of uuid */
577 __le64 s_first_cluster_group; /* Block offset of 1st cluster
579 /*50*/ __u8 s_label[OCFS2_MAX_VOL_LABEL_LEN]; /* Label for mounting, etc. */
580 /*90*/ __u8 s_uuid[OCFS2_VOL_UUID_LEN]; /* 128-bit uuid */
581 /*A0*/ struct ocfs2_cluster_info s_cluster_info; /* Selected userspace
583 with INCOMPAT flag. */
584 /*B8*/ __le16 s_xattr_inline_size; /* extended attribute inline size
588 /*C0*/ __le64 s_reserved2[16]; /* Fill out superblock */
592 * NOTE: As stated above, all offsets are relative to
593 * ocfs2_dinode.id2, which is at 0xC0 in the inode.
594 * 0xC0 + 0x140 = 0x200 or 512 bytes. A superblock must fit within
595 * our smallest blocksize, which is 512 bytes. To ensure this,
596 * we reserve the space in s_reserved2. Anything past s_reserved2
597 * will not be available on the smallest blocksize.
602 * Local allocation bitmap for OCFS2 slots
603 * Note that it exists inside an ocfs2_dinode, so all offsets are
604 * relative to the start of ocfs2_dinode.id2.
606 struct ocfs2_local_alloc
608 /*00*/ __le32 la_bm_off; /* Starting bit offset in main bitmap */
609 __le16 la_size; /* Size of included bitmap, in bytes */
612 /*10*/ __u8 la_bitmap[0];
616 * Data-in-inode header. This is only used if i_dyn_features has
617 * OCFS2_INLINE_DATA_FL set.
619 struct ocfs2_inline_data
621 /*00*/ __le16 id_count; /* Number of bytes that can be used
622 * for data, starting at id_data */
625 __u8 id_data[0]; /* Start of user data */
629 * On disk inode for OCFS2
631 struct ocfs2_dinode {
632 /*00*/ __u8 i_signature[8]; /* Signature for validation */
633 __le32 i_generation; /* Generation number */
634 __le16 i_suballoc_slot; /* Slot suballocator this inode
636 __le16 i_suballoc_bit; /* Bit offset in suballocator
638 /*10*/ __le16 i_reserved0;
639 __le16 i_xattr_inline_size;
640 __le32 i_clusters; /* Cluster count */
641 __le32 i_uid; /* Owner UID */
642 __le32 i_gid; /* Owning GID */
643 /*20*/ __le64 i_size; /* Size in bytes */
644 __le16 i_mode; /* File mode */
645 __le16 i_links_count; /* Links count */
646 __le32 i_flags; /* File flags */
647 /*30*/ __le64 i_atime; /* Access time */
648 __le64 i_ctime; /* Creation time */
649 /*40*/ __le64 i_mtime; /* Modification time */
650 __le64 i_dtime; /* Deletion time */
651 /*50*/ __le64 i_blkno; /* Offset on disk, in blocks */
652 __le64 i_last_eb_blk; /* Pointer to last extent
654 /*60*/ __le32 i_fs_generation; /* Generation per fs-instance */
658 /*70*/ __le32 i_attr;
659 __le16 i_orphaned_slot; /* Only valid when OCFS2_ORPHANED_FL
660 was set in i_flags */
661 __le16 i_dyn_features;
663 /*80*/ __le64 i_reserved2[7];
665 __le64 i_pad1; /* Generic way to refer to this
668 __le64 i_rdev; /* Device number */
670 struct { /* Info for bitmap system
672 __le32 i_used; /* Bits (ie, clusters) used */
673 __le32 i_total; /* Total bits (clusters)
676 struct { /* Info for journal system
678 __le32 ij_flags; /* Mounted, version, etc. */
679 __le32 ij_recovery_generation; /* Incremented when the
684 } id1; /* Inode type dependant 1 */
686 struct ocfs2_super_block i_super;
687 struct ocfs2_local_alloc i_lab;
688 struct ocfs2_chain_list i_chain;
689 struct ocfs2_extent_list i_list;
690 struct ocfs2_truncate_log i_dealloc;
691 struct ocfs2_inline_data i_data;
694 /* Actual on-disk size is one block */
698 * On-disk directory entry structure for OCFS2
700 * Packed as this structure could be accessed unaligned on 64-bit platforms
702 struct ocfs2_dir_entry {
703 /*00*/ __le64 inode; /* Inode number */
704 __le16 rec_len; /* Directory entry length */
705 __u8 name_len; /* Name length */
707 /*0C*/ char name[OCFS2_MAX_FILENAME_LEN]; /* File name */
708 /* Actual on-disk length specified by rec_len */
709 } __attribute__ ((packed));
712 * On disk allocator group structure for OCFS2
714 struct ocfs2_group_desc
716 /*00*/ __u8 bg_signature[8]; /* Signature for validation */
717 __le16 bg_size; /* Size of included bitmap in
719 __le16 bg_bits; /* Bits represented by this
721 __le16 bg_free_bits_count; /* Free bits count */
722 __le16 bg_chain; /* What chain I am in. */
723 /*10*/ __le32 bg_generation;
725 __le64 bg_next_group; /* Next group in my list, in
727 /*20*/ __le64 bg_parent_dinode; /* dinode which owns me, in
729 __le64 bg_blkno; /* Offset on disk, in blocks */
730 /*30*/ __le64 bg_reserved2[2];
731 /*40*/ __u8 bg_bitmap[0];
735 * On disk extended attribute structure for OCFS2.
739 * ocfs2_xattr_entry indicates one extend attribute.
741 * Note that it can be stored in inode, one block or one xattr bucket.
743 struct ocfs2_xattr_entry {
744 __le32 xe_name_hash; /* hash value of xattr prefix+suffix. */
745 __le16 xe_name_offset; /* byte offset from the 1st entry in the
746 local xattr storage(inode, xattr block or
748 __u8 xe_name_len; /* xattr name len, does't include prefix. */
749 __u8 xe_type; /* the low 7 bits indicate the name prefix
750 * type and the highest bit indicates whether
751 * the EA is stored in the local storage. */
752 __le64 xe_value_size; /* real xattr value length. */
756 * On disk structure for xattr header.
758 * One ocfs2_xattr_header describes how many ocfs2_xattr_entry records in
759 * the local xattr storage.
761 struct ocfs2_xattr_header {
762 __le16 xh_count; /* contains the count of how
763 many records are in the
764 local xattr storage. */
765 __le16 xh_free_start; /* current offset for storing
767 __le16 xh_name_value_len; /* total length of name/value
768 length in this bucket. */
769 __le16 xh_num_buckets; /* Number of xattr buckets
770 in this extent record,
771 only valid in the first
774 struct ocfs2_xattr_entry xh_entries[0]; /* xattr entry list. */
778 * On disk structure for xattr value root.
780 * When an xattr's value is large enough, it is stored in an external
781 * b-tree like file data. The xattr value root points to this structure.
783 struct ocfs2_xattr_value_root {
784 /*00*/ __le32 xr_clusters; /* clusters covered by xattr value. */
786 __le64 xr_last_eb_blk; /* Pointer to last extent block */
787 /*10*/ struct ocfs2_extent_list xr_list; /* Extent record list */
791 * On disk structure for xattr tree root.
793 * It is used when there are too many extended attributes for one file. These
794 * attributes will be organized and stored in an indexed-btree.
796 struct ocfs2_xattr_tree_root {
797 /*00*/ __le32 xt_clusters; /* clusters covered by xattr. */
799 __le64 xt_last_eb_blk; /* Pointer to last extent block */
800 /*10*/ struct ocfs2_extent_list xt_list; /* Extent record list */
803 #define OCFS2_XATTR_INDEXED 0x1
804 #define OCFS2_HASH_SHIFT 5
805 #define OCFS2_XATTR_ROUND 3
806 #define OCFS2_XATTR_SIZE(size) (((size) + OCFS2_XATTR_ROUND) & \
807 ~(OCFS2_XATTR_ROUND))
809 #define OCFS2_XATTR_BUCKET_SIZE 4096
810 #define OCFS2_XATTR_MAX_BLOCKS_PER_BUCKET (OCFS2_XATTR_BUCKET_SIZE \
811 / OCFS2_MIN_BLOCKSIZE)
814 * On disk structure for xattr block.
816 struct ocfs2_xattr_block {
817 /*00*/ __u8 xb_signature[8]; /* Signature for verification */
818 __le16 xb_suballoc_slot; /* Slot suballocator this
820 __le16 xb_suballoc_bit; /* Bit offset in suballocator
822 __le32 xb_fs_generation; /* Must match super block */
823 /*10*/ __le64 xb_blkno; /* Offset on disk, in blocks */
825 /*20*/ __le16 xb_flags; /* Indicates whether this block contains
826 real xattr or a xattr tree. */
831 struct ocfs2_xattr_header xb_header; /* xattr header if this
832 block contains xattr */
833 struct ocfs2_xattr_tree_root xb_root;/* xattr tree root if this
839 #define OCFS2_XATTR_ENTRY_LOCAL 0x80
840 #define OCFS2_XATTR_TYPE_MASK 0x7F
841 static inline void ocfs2_xattr_set_local(struct ocfs2_xattr_entry *xe,
845 xe->xe_type |= OCFS2_XATTR_ENTRY_LOCAL;
847 xe->xe_type &= ~OCFS2_XATTR_ENTRY_LOCAL;
850 static inline int ocfs2_xattr_is_local(struct ocfs2_xattr_entry *xe)
852 return xe->xe_type & OCFS2_XATTR_ENTRY_LOCAL;
855 static inline void ocfs2_xattr_set_type(struct ocfs2_xattr_entry *xe, int type)
857 xe->xe_type |= type & OCFS2_XATTR_TYPE_MASK;
860 static inline int ocfs2_xattr_get_type(struct ocfs2_xattr_entry *xe)
862 return xe->xe_type & OCFS2_XATTR_TYPE_MASK;
866 static inline int ocfs2_fast_symlink_chars(struct super_block *sb)
868 return sb->s_blocksize -
869 offsetof(struct ocfs2_dinode, id2.i_symlink);
872 static inline int ocfs2_max_inline_data(struct super_block *sb)
874 return sb->s_blocksize -
875 offsetof(struct ocfs2_dinode, id2.i_data.id_data);
878 static inline int ocfs2_max_inline_data_with_xattr(struct super_block *sb,
879 struct ocfs2_dinode *di)
881 unsigned int xattrsize = le16_to_cpu(di->i_xattr_inline_size);
883 if (le16_to_cpu(di->i_dyn_features) & OCFS2_INLINE_XATTR_FL)
884 return sb->s_blocksize -
885 offsetof(struct ocfs2_dinode, id2.i_data.id_data) -
888 return sb->s_blocksize -
889 offsetof(struct ocfs2_dinode, id2.i_data.id_data);
892 static inline int ocfs2_extent_recs_per_inode(struct super_block *sb)
896 size = sb->s_blocksize -
897 offsetof(struct ocfs2_dinode, id2.i_list.l_recs);
899 return size / sizeof(struct ocfs2_extent_rec);
902 static inline int ocfs2_extent_recs_per_inode_with_xattr(
903 struct super_block *sb,
904 struct ocfs2_dinode *di)
907 unsigned int xattrsize = le16_to_cpu(di->i_xattr_inline_size);
909 if (le16_to_cpu(di->i_dyn_features) & OCFS2_INLINE_XATTR_FL)
910 size = sb->s_blocksize -
911 offsetof(struct ocfs2_dinode, id2.i_list.l_recs) -
914 size = sb->s_blocksize -
915 offsetof(struct ocfs2_dinode, id2.i_list.l_recs);
917 return size / sizeof(struct ocfs2_extent_rec);
920 static inline int ocfs2_chain_recs_per_inode(struct super_block *sb)
924 size = sb->s_blocksize -
925 offsetof(struct ocfs2_dinode, id2.i_chain.cl_recs);
927 return size / sizeof(struct ocfs2_chain_rec);
930 static inline u16 ocfs2_extent_recs_per_eb(struct super_block *sb)
934 size = sb->s_blocksize -
935 offsetof(struct ocfs2_extent_block, h_list.l_recs);
937 return size / sizeof(struct ocfs2_extent_rec);
940 static inline u16 ocfs2_local_alloc_size(struct super_block *sb)
944 size = sb->s_blocksize -
945 offsetof(struct ocfs2_dinode, id2.i_lab.la_bitmap);
950 static inline int ocfs2_group_bitmap_size(struct super_block *sb)
954 size = sb->s_blocksize -
955 offsetof(struct ocfs2_group_desc, bg_bitmap);
960 static inline int ocfs2_truncate_recs_per_inode(struct super_block *sb)
964 size = sb->s_blocksize -
965 offsetof(struct ocfs2_dinode, id2.i_dealloc.tl_recs);
967 return size / sizeof(struct ocfs2_truncate_rec);
970 static inline u64 ocfs2_backup_super_blkno(struct super_block *sb, int index)
972 u64 offset = OCFS2_BACKUP_SB_START;
974 if (index >= 0 && index < OCFS2_MAX_BACKUP_SUPERBLOCKS) {
975 offset <<= (2 * index);
976 offset >>= sb->s_blocksize_bits;
984 static inline u16 ocfs2_xattr_recs_per_xb(struct super_block *sb)
988 size = sb->s_blocksize -
989 offsetof(struct ocfs2_xattr_block,
990 xb_attrs.xb_root.xt_list.l_recs);
992 return size / sizeof(struct ocfs2_extent_rec);
995 static inline int ocfs2_fast_symlink_chars(int blocksize)
997 return blocksize - offsetof(struct ocfs2_dinode, id2.i_symlink);
1000 static inline int ocfs2_max_inline_data(int blocksize)
1002 return blocksize - offsetof(struct ocfs2_dinode, id2.i_data.id_data);
1005 static inline int ocfs2_extent_recs_per_inode(int blocksize)
1010 offsetof(struct ocfs2_dinode, id2.i_list.l_recs);
1012 return size / sizeof(struct ocfs2_extent_rec);
1015 static inline int ocfs2_chain_recs_per_inode(int blocksize)
1020 offsetof(struct ocfs2_dinode, id2.i_chain.cl_recs);
1022 return size / sizeof(struct ocfs2_chain_rec);
1025 static inline int ocfs2_extent_recs_per_eb(int blocksize)
1030 offsetof(struct ocfs2_extent_block, h_list.l_recs);
1032 return size / sizeof(struct ocfs2_extent_rec);
1035 static inline int ocfs2_local_alloc_size(int blocksize)
1040 offsetof(struct ocfs2_dinode, id2.i_lab.la_bitmap);
1045 static inline int ocfs2_group_bitmap_size(int blocksize)
1050 offsetof(struct ocfs2_group_desc, bg_bitmap);
1055 static inline int ocfs2_truncate_recs_per_inode(int blocksize)
1060 offsetof(struct ocfs2_dinode, id2.i_dealloc.tl_recs);
1062 return size / sizeof(struct ocfs2_truncate_rec);
1065 static inline uint64_t ocfs2_backup_super_blkno(int blocksize, int index)
1067 uint64_t offset = OCFS2_BACKUP_SB_START;
1069 if (index >= 0 && index < OCFS2_MAX_BACKUP_SUPERBLOCKS) {
1070 offset <<= (2 * index);
1071 offset /= blocksize;
1078 static inline int ocfs2_xattr_recs_per_xb(int blocksize)
1083 offsetof(struct ocfs2_xattr_block,
1084 xb_attrs.xb_root.xt_list.l_recs);
1086 return size / sizeof(struct ocfs2_extent_rec);
1088 #endif /* __KERNEL__ */
1091 static inline int ocfs2_system_inode_is_global(int type)
1093 return ((type >= 0) &&
1094 (type <= OCFS2_LAST_GLOBAL_SYSTEM_INODE));
1097 static inline int ocfs2_sprintf_system_inode_name(char *buf, int len,
1103 * Global system inodes can only have one copy. Everything
1104 * after OCFS2_LAST_GLOBAL_SYSTEM_INODE in the system inode
1105 * list has a copy per slot.
1107 if (type <= OCFS2_LAST_GLOBAL_SYSTEM_INODE)
1108 chars = snprintf(buf, len, "%s",
1109 ocfs2_system_inodes[type].si_name);
1111 chars = snprintf(buf, len,
1112 ocfs2_system_inodes[type].si_name,
1118 static inline void ocfs2_set_de_type(struct ocfs2_dir_entry *de,
1121 de->file_type = ocfs2_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
1124 #endif /* _OCFS2_FS_H */