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 #define OCFS2_FEATURE_RO_COMPAT_SUPP OCFS2_FEATURE_RO_COMPAT_UNWRITTEN
98 * Heartbeat-only devices are missing journals and other files. The
99 * filesystem driver can't load them, but the library can. Never put
100 * this in OCFS2_FEATURE_INCOMPAT_SUPP, *ever*.
102 #define OCFS2_FEATURE_INCOMPAT_HEARTBEAT_DEV 0x0002
105 * tunefs sets this incompat flag before starting the resize and clears it
106 * at the end. This flag protects users from inadvertently mounting the fs
107 * after an aborted run without fsck-ing.
109 #define OCFS2_FEATURE_INCOMPAT_RESIZE_INPROG 0x0004
111 /* Used to denote a non-clustered volume */
112 #define OCFS2_FEATURE_INCOMPAT_LOCAL_MOUNT 0x0008
114 /* Support for sparse allocation in b-trees */
115 #define OCFS2_FEATURE_INCOMPAT_SPARSE_ALLOC 0x0010
118 * Tunefs sets this incompat flag before starting an operation which
119 * would require cleanup on abort. This is done to protect users from
120 * inadvertently mounting the fs after an aborted run without
123 * s_tunefs_flags on the super block describes precisely which
124 * operations were in progress.
126 #define OCFS2_FEATURE_INCOMPAT_TUNEFS_INPROG 0x0020
128 /* Support for data packed into inode blocks */
129 #define OCFS2_FEATURE_INCOMPAT_INLINE_DATA 0x0040
131 /* Support for the extended slot map */
132 #define OCFS2_FEATURE_INCOMPAT_EXTENDED_SLOT_MAP 0x100
136 * Support for alternate, userspace cluster stacks. If set, the superblock
137 * field s_cluster_info contains a tag for the alternate stack in use as
138 * well as the name of the cluster being joined.
139 * mount.ocfs2 must pass in a matching stack name.
141 * If not set, the classic stack will be used. This is compatbile with
142 * all older versions.
144 #define OCFS2_FEATURE_INCOMPAT_USERSPACE_STACK 0x0080
147 * backup superblock flag is used to indicate that this volume
148 * has backup superblocks.
150 #define OCFS2_FEATURE_COMPAT_BACKUP_SB 0x0001
153 * Unwritten extents support.
155 #define OCFS2_FEATURE_RO_COMPAT_UNWRITTEN 0x0001
157 /* The byte offset of the first backup block will be 1G.
158 * The following will be 4G, 16G, 64G, 256G and 1T.
160 #define OCFS2_BACKUP_SB_START 1 << 30
162 /* the max backup superblock nums */
163 #define OCFS2_MAX_BACKUP_SUPERBLOCKS 6
166 * Flags on ocfs2_super_block.s_tunefs_flags
168 #define OCFS2_TUNEFS_INPROG_REMOVE_SLOT 0x0001 /* Removing slots */
171 * Flags on ocfs2_dinode.i_flags
173 #define OCFS2_VALID_FL (0x00000001) /* Inode is valid */
174 #define OCFS2_UNUSED2_FL (0x00000002)
175 #define OCFS2_ORPHANED_FL (0x00000004) /* On the orphan list */
176 #define OCFS2_UNUSED3_FL (0x00000008)
177 /* System inode flags */
178 #define OCFS2_SYSTEM_FL (0x00000010) /* System inode */
179 #define OCFS2_SUPER_BLOCK_FL (0x00000020) /* Super block */
180 #define OCFS2_LOCAL_ALLOC_FL (0x00000040) /* Slot local alloc bitmap */
181 #define OCFS2_BITMAP_FL (0x00000080) /* Allocation bitmap */
182 #define OCFS2_JOURNAL_FL (0x00000100) /* Slot local journal */
183 #define OCFS2_HEARTBEAT_FL (0x00000200) /* Heartbeat area */
184 #define OCFS2_CHAIN_FL (0x00000400) /* Chain allocator */
185 #define OCFS2_DEALLOC_FL (0x00000800) /* Truncate log */
188 * Flags on ocfs2_dinode.i_dyn_features
190 * These can change much more often than i_flags. When adding flags,
191 * keep in mind that i_dyn_features is only 16 bits wide.
193 #define OCFS2_INLINE_DATA_FL (0x0001) /* Data stored in inode block */
194 #define OCFS2_HAS_XATTR_FL (0x0002)
195 #define OCFS2_INLINE_XATTR_FL (0x0004)
196 #define OCFS2_INDEXED_DIR_FL (0x0008)
198 /* Inode attributes, keep in sync with EXT2 */
199 #define OCFS2_SECRM_FL (0x00000001) /* Secure deletion */
200 #define OCFS2_UNRM_FL (0x00000002) /* Undelete */
201 #define OCFS2_COMPR_FL (0x00000004) /* Compress file */
202 #define OCFS2_SYNC_FL (0x00000008) /* Synchronous updates */
203 #define OCFS2_IMMUTABLE_FL (0x00000010) /* Immutable file */
204 #define OCFS2_APPEND_FL (0x00000020) /* writes to file may only append */
205 #define OCFS2_NODUMP_FL (0x00000040) /* do not dump file */
206 #define OCFS2_NOATIME_FL (0x00000080) /* do not update atime */
207 #define OCFS2_DIRSYNC_FL (0x00010000) /* dirsync behaviour (directories only) */
209 #define OCFS2_FL_VISIBLE (0x000100FF) /* User visible flags */
210 #define OCFS2_FL_MODIFIABLE (0x000100FF) /* User modifiable flags */
213 * Extent record flags (e_node.leaf.flags)
215 #define OCFS2_EXT_UNWRITTEN (0x01) /* Extent is allocated but
221 #define OCFS2_IOC_GETFLAGS _IOR('f', 1, long)
222 #define OCFS2_IOC_SETFLAGS _IOW('f', 2, long)
223 #define OCFS2_IOC32_GETFLAGS _IOR('f', 1, int)
224 #define OCFS2_IOC32_SETFLAGS _IOW('f', 2, int)
227 * Space reservation / allocation / free ioctls and argument structure
228 * are designed to be compatible with XFS.
230 * ALLOCSP* and FREESP* are not and will never be supported, but are
231 * included here for completeness.
233 struct ocfs2_space_resv {
237 __s64 l_len; /* len == 0 means until end of file */
240 __s32 l_pad[4]; /* reserve area */
243 #define OCFS2_IOC_ALLOCSP _IOW ('X', 10, struct ocfs2_space_resv)
244 #define OCFS2_IOC_FREESP _IOW ('X', 11, struct ocfs2_space_resv)
245 #define OCFS2_IOC_RESVSP _IOW ('X', 40, struct ocfs2_space_resv)
246 #define OCFS2_IOC_UNRESVSP _IOW ('X', 41, struct ocfs2_space_resv)
247 #define OCFS2_IOC_ALLOCSP64 _IOW ('X', 36, struct ocfs2_space_resv)
248 #define OCFS2_IOC_FREESP64 _IOW ('X', 37, struct ocfs2_space_resv)
249 #define OCFS2_IOC_RESVSP64 _IOW ('X', 42, struct ocfs2_space_resv)
250 #define OCFS2_IOC_UNRESVSP64 _IOW ('X', 43, struct ocfs2_space_resv)
252 /* Used to pass group descriptor data when online resize is done */
253 struct ocfs2_new_group_input {
254 __u64 group; /* Group descriptor's blkno. */
255 __u32 clusters; /* Total number of clusters in this group */
256 __u32 frees; /* Total free clusters in this group */
257 __u16 chain; /* Chain for this group */
262 #define OCFS2_IOC_GROUP_EXTEND _IOW('o', 1, int)
263 #define OCFS2_IOC_GROUP_ADD _IOW('o', 2,struct ocfs2_new_group_input)
264 #define OCFS2_IOC_GROUP_ADD64 _IOW('o', 3,struct ocfs2_new_group_input)
267 * Journal Flags (ocfs2_dinode.id1.journal1.i_flags)
269 #define OCFS2_JOURNAL_DIRTY_FL (0x00000001) /* Journal needs recovery */
272 * superblock s_state flags
274 #define OCFS2_ERROR_FS (0x00000001) /* FS saw errors */
276 /* Limit of space in ocfs2_dir_entry */
277 #define OCFS2_MAX_FILENAME_LEN 255
279 /* Maximum slots on an ocfs2 file system */
280 #define OCFS2_MAX_SLOTS 255
282 /* Slot map indicator for an empty slot */
283 #define OCFS2_INVALID_SLOT -1
285 #define OCFS2_VOL_UUID_LEN 16
286 #define OCFS2_MAX_VOL_LABEL_LEN 64
288 /* The alternate, userspace stack fields */
289 #define OCFS2_STACK_LABEL_LEN 4
290 #define OCFS2_CLUSTER_NAME_LEN 16
292 /* Journal limits (in bytes) */
293 #define OCFS2_MIN_JOURNAL_SIZE (4 * 1024 * 1024)
296 * Default local alloc size (in megabytes)
298 * The value chosen should be such that most allocations, including new
299 * block groups, use local alloc.
301 #define OCFS2_DEFAULT_LOCAL_ALLOC_SIZE 8
304 * Inline extended attribute size (in bytes)
305 * The value chosen should be aligned to 16 byte boundaries.
307 #define OCFS2_MIN_XATTR_INLINE_SIZE 256
309 struct ocfs2_system_inode_info {
315 /* System file index */
317 BAD_BLOCK_SYSTEM_INODE = 0,
318 GLOBAL_INODE_ALLOC_SYSTEM_INODE,
319 SLOT_MAP_SYSTEM_INODE,
320 #define OCFS2_FIRST_ONLINE_SYSTEM_INODE SLOT_MAP_SYSTEM_INODE
321 HEARTBEAT_SYSTEM_INODE,
322 GLOBAL_BITMAP_SYSTEM_INODE,
323 #define OCFS2_LAST_GLOBAL_SYSTEM_INODE GLOBAL_BITMAP_SYSTEM_INODE
324 ORPHAN_DIR_SYSTEM_INODE,
325 EXTENT_ALLOC_SYSTEM_INODE,
326 INODE_ALLOC_SYSTEM_INODE,
327 JOURNAL_SYSTEM_INODE,
328 LOCAL_ALLOC_SYSTEM_INODE,
329 TRUNCATE_LOG_SYSTEM_INODE,
333 static struct ocfs2_system_inode_info ocfs2_system_inodes[NUM_SYSTEM_INODES] = {
334 /* Global system inodes (single copy) */
335 /* The first two are only used from userspace mfks/tunefs */
336 [BAD_BLOCK_SYSTEM_INODE] = { "bad_blocks", 0, S_IFREG | 0644 },
337 [GLOBAL_INODE_ALLOC_SYSTEM_INODE] = { "global_inode_alloc", OCFS2_BITMAP_FL | OCFS2_CHAIN_FL, S_IFREG | 0644 },
339 /* These are used by the running filesystem */
340 [SLOT_MAP_SYSTEM_INODE] = { "slot_map", 0, S_IFREG | 0644 },
341 [HEARTBEAT_SYSTEM_INODE] = { "heartbeat", OCFS2_HEARTBEAT_FL, S_IFREG | 0644 },
342 [GLOBAL_BITMAP_SYSTEM_INODE] = { "global_bitmap", 0, S_IFREG | 0644 },
344 /* Slot-specific system inodes (one copy per slot) */
345 [ORPHAN_DIR_SYSTEM_INODE] = { "orphan_dir:%04d", 0, S_IFDIR | 0755 },
346 [EXTENT_ALLOC_SYSTEM_INODE] = { "extent_alloc:%04d", OCFS2_BITMAP_FL | OCFS2_CHAIN_FL, S_IFREG | 0644 },
347 [INODE_ALLOC_SYSTEM_INODE] = { "inode_alloc:%04d", OCFS2_BITMAP_FL | OCFS2_CHAIN_FL, S_IFREG | 0644 },
348 [JOURNAL_SYSTEM_INODE] = { "journal:%04d", OCFS2_JOURNAL_FL, S_IFREG | 0644 },
349 [LOCAL_ALLOC_SYSTEM_INODE] = { "local_alloc:%04d", OCFS2_BITMAP_FL | OCFS2_LOCAL_ALLOC_FL, S_IFREG | 0644 },
350 [TRUNCATE_LOG_SYSTEM_INODE] = { "truncate_log:%04d", OCFS2_DEALLOC_FL, S_IFREG | 0644 }
353 /* Parameter passed from mount.ocfs2 to module */
354 #define OCFS2_HB_NONE "heartbeat=none"
355 #define OCFS2_HB_LOCAL "heartbeat=local"
358 * OCFS2 directory file types. Only the low 3 bits are used. The
359 * other bits are reserved for now.
361 #define OCFS2_FT_UNKNOWN 0
362 #define OCFS2_FT_REG_FILE 1
363 #define OCFS2_FT_DIR 2
364 #define OCFS2_FT_CHRDEV 3
365 #define OCFS2_FT_BLKDEV 4
366 #define OCFS2_FT_FIFO 5
367 #define OCFS2_FT_SOCK 6
368 #define OCFS2_FT_SYMLINK 7
370 #define OCFS2_FT_MAX 8
373 * OCFS2_DIR_PAD defines the directory entries boundaries
375 * NOTE: It must be a multiple of 4
377 #define OCFS2_DIR_PAD 4
378 #define OCFS2_DIR_ROUND (OCFS2_DIR_PAD - 1)
379 #define OCFS2_DIR_MEMBER_LEN offsetof(struct ocfs2_dir_entry, name)
380 #define OCFS2_DIR_REC_LEN(name_len) (((name_len) + OCFS2_DIR_MEMBER_LEN + \
384 #define OCFS2_LINK_MAX 32000
387 static unsigned char ocfs2_type_by_mode[S_IFMT >> S_SHIFT] = {
388 [S_IFREG >> S_SHIFT] = OCFS2_FT_REG_FILE,
389 [S_IFDIR >> S_SHIFT] = OCFS2_FT_DIR,
390 [S_IFCHR >> S_SHIFT] = OCFS2_FT_CHRDEV,
391 [S_IFBLK >> S_SHIFT] = OCFS2_FT_BLKDEV,
392 [S_IFIFO >> S_SHIFT] = OCFS2_FT_FIFO,
393 [S_IFSOCK >> S_SHIFT] = OCFS2_FT_SOCK,
394 [S_IFLNK >> S_SHIFT] = OCFS2_FT_SYMLINK,
401 #define OCFS2_RAW_SB(dinode) (&((dinode)->id2.i_super))
404 * On disk extent record for OCFS2
405 * It describes a range of clusters on disk.
407 * Length fields are divided into interior and leaf node versions.
408 * This leaves room for a flags field (OCFS2_EXT_*) in the leaf nodes.
410 struct ocfs2_extent_rec {
411 /*00*/ __le32 e_cpos; /* Offset into the file, in clusters */
413 __le32 e_int_clusters; /* Clusters covered by all children */
415 __le16 e_leaf_clusters; /* Clusters covered by this
418 __u8 e_flags; /* Extent flags */
421 __le64 e_blkno; /* Physical disk offset, in blocks */
425 struct ocfs2_chain_rec {
426 __le32 c_free; /* Number of free bits in this chain. */
427 __le32 c_total; /* Number of total bits in this chain */
428 __le64 c_blkno; /* Physical disk offset (blocks) of 1st group */
431 struct ocfs2_truncate_rec {
432 __le32 t_start; /* 1st cluster in this log */
433 __le32 t_clusters; /* Number of total clusters covered */
437 * On disk extent list for OCFS2 (node in the tree). Note that this
438 * is contained inside ocfs2_dinode or ocfs2_extent_block, so the
439 * offsets are relative to ocfs2_dinode.id2.i_list or
440 * ocfs2_extent_block.h_list, respectively.
442 struct ocfs2_extent_list {
443 /*00*/ __le16 l_tree_depth; /* Extent tree depth from this
444 point. 0 means data extents
445 hang directly off this
447 NOTE: The high 8 bits cannot be
448 used - tree_depth is never that big.
450 __le16 l_count; /* Number of extent records */
451 __le16 l_next_free_rec; /* Next unused extent slot */
453 __le64 l_reserved2; /* Pad to
454 sizeof(ocfs2_extent_rec) */
455 /*10*/ struct ocfs2_extent_rec l_recs[0]; /* Extent records */
459 * On disk allocation chain list for OCFS2. Note that this is
460 * contained inside ocfs2_dinode, so the offsets are relative to
461 * ocfs2_dinode.id2.i_chain.
463 struct ocfs2_chain_list {
464 /*00*/ __le16 cl_cpg; /* Clusters per Block Group */
465 __le16 cl_bpc; /* Bits per cluster */
466 __le16 cl_count; /* Total chains in this list */
467 __le16 cl_next_free_rec; /* Next unused chain slot */
469 /*10*/ struct ocfs2_chain_rec cl_recs[0]; /* Chain records */
473 * On disk deallocation log for OCFS2. Note that this is
474 * contained inside ocfs2_dinode, so the offsets are relative to
475 * ocfs2_dinode.id2.i_dealloc.
477 struct ocfs2_truncate_log {
478 /*00*/ __le16 tl_count; /* Total records in this log */
479 __le16 tl_used; /* Number of records in use */
481 /*08*/ struct ocfs2_truncate_rec tl_recs[0]; /* Truncate records */
485 * On disk extent block (indirect block) for OCFS2
487 struct ocfs2_extent_block
489 /*00*/ __u8 h_signature[8]; /* Signature for verification */
491 /*10*/ __le16 h_suballoc_slot; /* Slot suballocator this
492 extent_header belongs to */
493 __le16 h_suballoc_bit; /* Bit offset in suballocator
495 __le32 h_fs_generation; /* Must match super block */
496 __le64 h_blkno; /* Offset on disk, in blocks */
497 /*20*/ __le64 h_reserved3;
498 __le64 h_next_leaf_blk; /* Offset on disk, in blocks,
499 of next leaf header pointing
501 /*30*/ struct ocfs2_extent_list h_list; /* Extent record list */
502 /* Actual on-disk size is one block */
506 * On disk slot map for OCFS2. This defines the contents of the "slot_map"
507 * system file. A slot is valid if it contains a node number >= 0. The
508 * value -1 (0xFFFF) is OCFS2_INVALID_SLOT. This marks a slot empty.
510 struct ocfs2_slot_map {
511 /*00*/ __le16 sm_slots[0];
513 * Actual on-disk size is one block. OCFS2_MAX_SLOTS is 255,
514 * 255 * sizeof(__le16) == 512B, within the 512B block minimum blocksize.
518 struct ocfs2_extended_slot {
519 /*00*/ __u8 es_valid;
520 __u8 es_reserved1[3];
526 * The extended slot map, used when OCFS2_FEATURE_INCOMPAT_EXTENDED_SLOT_MAP
527 * is set. It separates out the valid marker from the node number, and
528 * has room to grow. Unlike the old slot map, this format is defined by
531 struct ocfs2_slot_map_extended {
532 /*00*/ struct ocfs2_extended_slot se_slots[0];
534 * Actual size is i_size of the slot_map system file. It should
535 * match s_max_slots * sizeof(struct ocfs2_extended_slot)
539 struct ocfs2_cluster_info {
540 /*00*/ __u8 ci_stack[OCFS2_STACK_LABEL_LEN];
542 /*08*/ __u8 ci_cluster[OCFS2_CLUSTER_NAME_LEN];
547 * On disk superblock for OCFS2
548 * Note that it is contained inside an ocfs2_dinode, so all offsets
549 * are relative to the start of ocfs2_dinode.id2.
551 struct ocfs2_super_block {
552 /*00*/ __le16 s_major_rev_level;
553 __le16 s_minor_rev_level;
555 __le16 s_max_mnt_count;
556 __le16 s_state; /* File system state */
557 __le16 s_errors; /* Behaviour when detecting errors */
558 __le32 s_checkinterval; /* Max time between checks */
559 /*10*/ __le64 s_lastcheck; /* Time of last check */
560 __le32 s_creator_os; /* OS */
561 __le32 s_feature_compat; /* Compatible feature set */
562 /*20*/ __le32 s_feature_incompat; /* Incompatible feature set */
563 __le32 s_feature_ro_compat; /* Readonly-compatible feature set */
564 __le64 s_root_blkno; /* Offset, in blocks, of root directory
566 /*30*/ __le64 s_system_dir_blkno; /* Offset, in blocks, of system
568 __le32 s_blocksize_bits; /* Blocksize for this fs */
569 __le32 s_clustersize_bits; /* Clustersize for this fs */
570 /*40*/ __le16 s_max_slots; /* Max number of simultaneous mounts
571 before tunefs required */
572 __le16 s_tunefs_flag;
573 __le32 s_uuid_hash; /* hash value of uuid */
574 __le64 s_first_cluster_group; /* Block offset of 1st cluster
576 /*50*/ __u8 s_label[OCFS2_MAX_VOL_LABEL_LEN]; /* Label for mounting, etc. */
577 /*90*/ __u8 s_uuid[OCFS2_VOL_UUID_LEN]; /* 128-bit uuid */
578 /*A0*/ struct ocfs2_cluster_info s_cluster_info; /* Selected userspace
580 with INCOMPAT flag. */
581 /*B8*/ __le64 s_reserved2[17]; /* Fill out superblock */
585 * NOTE: As stated above, all offsets are relative to
586 * ocfs2_dinode.id2, which is at 0xC0 in the inode.
587 * 0xC0 + 0x140 = 0x200 or 512 bytes. A superblock must fit within
588 * our smallest blocksize, which is 512 bytes. To ensure this,
589 * we reserve the space in s_reserved2. Anything past s_reserved2
590 * will not be available on the smallest blocksize.
595 * Local allocation bitmap for OCFS2 slots
596 * Note that it exists inside an ocfs2_dinode, so all offsets are
597 * relative to the start of ocfs2_dinode.id2.
599 struct ocfs2_local_alloc
601 /*00*/ __le32 la_bm_off; /* Starting bit offset in main bitmap */
602 __le16 la_size; /* Size of included bitmap, in bytes */
605 /*10*/ __u8 la_bitmap[0];
609 * Data-in-inode header. This is only used if i_dyn_features has
610 * OCFS2_INLINE_DATA_FL set.
612 struct ocfs2_inline_data
614 /*00*/ __le16 id_count; /* Number of bytes that can be used
615 * for data, starting at id_data */
618 __u8 id_data[0]; /* Start of user data */
622 * On disk inode for OCFS2
624 struct ocfs2_dinode {
625 /*00*/ __u8 i_signature[8]; /* Signature for validation */
626 __le32 i_generation; /* Generation number */
627 __le16 i_suballoc_slot; /* Slot suballocator this inode
629 __le16 i_suballoc_bit; /* Bit offset in suballocator
631 /*10*/ __le16 i_reserved0;
632 __le16 i_xattr_inline_size;
633 __le32 i_clusters; /* Cluster count */
634 __le32 i_uid; /* Owner UID */
635 __le32 i_gid; /* Owning GID */
636 /*20*/ __le64 i_size; /* Size in bytes */
637 __le16 i_mode; /* File mode */
638 __le16 i_links_count; /* Links count */
639 __le32 i_flags; /* File flags */
640 /*30*/ __le64 i_atime; /* Access time */
641 __le64 i_ctime; /* Creation time */
642 /*40*/ __le64 i_mtime; /* Modification time */
643 __le64 i_dtime; /* Deletion time */
644 /*50*/ __le64 i_blkno; /* Offset on disk, in blocks */
645 __le64 i_last_eb_blk; /* Pointer to last extent
647 /*60*/ __le32 i_fs_generation; /* Generation per fs-instance */
651 /*70*/ __le32 i_attr;
652 __le16 i_orphaned_slot; /* Only valid when OCFS2_ORPHANED_FL
653 was set in i_flags */
654 __le16 i_dyn_features;
656 /*80*/ __le64 i_reserved2[7];
658 __le64 i_pad1; /* Generic way to refer to this
661 __le64 i_rdev; /* Device number */
663 struct { /* Info for bitmap system
665 __le32 i_used; /* Bits (ie, clusters) used */
666 __le32 i_total; /* Total bits (clusters)
669 struct { /* Info for journal system
671 __le32 ij_flags; /* Mounted, version, etc. */
672 __le32 ij_recovery_generation; /* Incremented when the
677 } id1; /* Inode type dependant 1 */
679 struct ocfs2_super_block i_super;
680 struct ocfs2_local_alloc i_lab;
681 struct ocfs2_chain_list i_chain;
682 struct ocfs2_extent_list i_list;
683 struct ocfs2_truncate_log i_dealloc;
684 struct ocfs2_inline_data i_data;
687 /* Actual on-disk size is one block */
691 * On-disk directory entry structure for OCFS2
693 * Packed as this structure could be accessed unaligned on 64-bit platforms
695 struct ocfs2_dir_entry {
696 /*00*/ __le64 inode; /* Inode number */
697 __le16 rec_len; /* Directory entry length */
698 __u8 name_len; /* Name length */
700 /*0C*/ char name[OCFS2_MAX_FILENAME_LEN]; /* File name */
701 /* Actual on-disk length specified by rec_len */
702 } __attribute__ ((packed));
705 * On disk allocator group structure for OCFS2
707 struct ocfs2_group_desc
709 /*00*/ __u8 bg_signature[8]; /* Signature for validation */
710 __le16 bg_size; /* Size of included bitmap in
712 __le16 bg_bits; /* Bits represented by this
714 __le16 bg_free_bits_count; /* Free bits count */
715 __le16 bg_chain; /* What chain I am in. */
716 /*10*/ __le32 bg_generation;
718 __le64 bg_next_group; /* Next group in my list, in
720 /*20*/ __le64 bg_parent_dinode; /* dinode which owns me, in
722 __le64 bg_blkno; /* Offset on disk, in blocks */
723 /*30*/ __le64 bg_reserved2[2];
724 /*40*/ __u8 bg_bitmap[0];
728 * On disk extended attribute structure for OCFS2.
732 * ocfs2_xattr_entry indicates one extend attribute.
734 * Note that it can be stored in inode, one block or one xattr bucket.
736 struct ocfs2_xattr_entry {
737 __le32 xe_name_hash; /* hash value of xattr prefix+suffix. */
738 __le16 xe_name_offset; /* byte offset from the 1st etnry in the local
739 local xattr storage(inode, xattr block or
741 __u8 xe_name_len; /* xattr name len, does't include prefix. */
742 __u8 xe_type; /* the low 7 bits indicates the name prefix's
743 * type and the highest 1 bits indicate whether
744 * the EA is stored in the local storage. */
745 __le64 xe_value_size; /* real xattr value length. */
749 * On disk structure for xattr header.
751 * One ocfs2_xattr_header describes how many ocfs2_xattr_entry records in
752 * the local xattr storage.
754 struct ocfs2_xattr_header {
755 __le16 xh_count; /* contains the count of how
756 many records are in the
757 local xattr storage. */
761 struct ocfs2_xattr_entry xh_entries[0]; /* xattr entry list. */
765 * On disk structure for xattr value root.
767 * It is used when one extended attribute's size is larger, and we will save it
768 * in an outside cluster. It will stored in a b-tree like file content.
770 struct ocfs2_xattr_value_root {
771 /*00*/ __le32 xr_clusters; /* clusters covered by xattr value. */
773 __le64 xr_last_eb_blk; /* Pointer to last extent block */
774 /*10*/ struct ocfs2_extent_list xr_list; /* Extent record list */
778 * On disk structure for xattr tree root.
780 * It is used when there are too many extended attributes for one file. These
781 * attributes will be organized and stored in an indexed-btree.
783 struct ocfs2_xattr_tree_root {
784 /*00*/ __le32 xt_clusters; /* clusters covered by xattr. */
786 __le64 xt_last_eb_blk; /* Pointer to last extent block */
787 /*10*/ struct ocfs2_extent_list xt_list; /* Extent record list */
790 #define OCFS2_XATTR_INDEXED 0x1
791 #define OCFS2_HASH_SHIFT 5
792 #define OCFS2_XATTR_ROUND 3
793 #define OCFS2_XATTR_SIZE(size) (((size) + OCFS2_XATTR_ROUND) & \
794 ~(OCFS2_XATTR_ROUND))
797 * On disk structure for xattr block.
799 struct ocfs2_xattr_block {
800 /*00*/ __u8 xb_signature[8]; /* Signature for verification */
801 __le16 xb_suballoc_slot; /* Slot suballocator this
803 __le16 xb_suballoc_bit; /* Bit offset in suballocator
805 __le32 xb_fs_generation; /* Must match super block */
806 /*10*/ __le64 xb_blkno; /* Offset on disk, in blocks */
808 /*20*/ __le16 xb_flags; /* Indicates whether this block contains
809 real xattr or a xattr tree. */
814 struct ocfs2_xattr_header xb_header; /* xattr header if this
815 block contains xattr */
816 struct ocfs2_xattr_tree_root xb_root;/* xattr tree root if this
822 #define OCFS2_XATTR_ENTRY_LOCAL 0x80
823 #define OCFS2_XATTR_TYPE_MASK 0x7F
824 static inline void ocfs2_xattr_set_local(struct ocfs2_xattr_entry *xe,
828 xe->xe_type |= OCFS2_XATTR_ENTRY_LOCAL;
830 xe->xe_type &= ~OCFS2_XATTR_ENTRY_LOCAL;
833 static inline int ocfs2_xattr_is_local(struct ocfs2_xattr_entry *xe)
835 return xe->xe_type & OCFS2_XATTR_ENTRY_LOCAL;
838 static inline void ocfs2_xattr_set_type(struct ocfs2_xattr_entry *xe, int type)
840 xe->xe_type |= type & OCFS2_XATTR_TYPE_MASK;
843 static inline int ocfs2_xattr_get_type(struct ocfs2_xattr_entry *xe)
845 return xe->xe_type & OCFS2_XATTR_TYPE_MASK;
849 static inline int ocfs2_fast_symlink_chars(struct super_block *sb)
851 return sb->s_blocksize -
852 offsetof(struct ocfs2_dinode, id2.i_symlink);
855 static inline int ocfs2_max_inline_data(struct super_block *sb)
857 return sb->s_blocksize -
858 offsetof(struct ocfs2_dinode, id2.i_data.id_data);
861 static inline int ocfs2_max_inline_data_with_xattr(struct super_block *sb,
862 struct ocfs2_dinode *di)
864 unsigned int xattrsize = le16_to_cpu(di->i_xattr_inline_size);
866 if (le16_to_cpu(di->i_dyn_features) & OCFS2_INLINE_XATTR_FL)
867 return sb->s_blocksize -
868 offsetof(struct ocfs2_dinode, id2.i_data.id_data) -
871 return sb->s_blocksize -
872 offsetof(struct ocfs2_dinode, id2.i_data.id_data);
875 static inline int ocfs2_extent_recs_per_inode(struct super_block *sb)
879 size = sb->s_blocksize -
880 offsetof(struct ocfs2_dinode, id2.i_list.l_recs);
882 return size / sizeof(struct ocfs2_extent_rec);
885 static inline int ocfs2_extent_recs_per_inode_with_xattr(
886 struct super_block *sb,
887 struct ocfs2_dinode *di)
890 unsigned int xattrsize = le16_to_cpu(di->i_xattr_inline_size);
892 if (le16_to_cpu(di->i_dyn_features) & OCFS2_INLINE_XATTR_FL)
893 size = sb->s_blocksize -
894 offsetof(struct ocfs2_dinode, id2.i_list.l_recs) -
897 size = sb->s_blocksize -
898 offsetof(struct ocfs2_dinode, id2.i_list.l_recs);
900 return size / sizeof(struct ocfs2_extent_rec);
903 static inline int ocfs2_chain_recs_per_inode(struct super_block *sb)
907 size = sb->s_blocksize -
908 offsetof(struct ocfs2_dinode, id2.i_chain.cl_recs);
910 return size / sizeof(struct ocfs2_chain_rec);
913 static inline u16 ocfs2_extent_recs_per_eb(struct super_block *sb)
917 size = sb->s_blocksize -
918 offsetof(struct ocfs2_extent_block, h_list.l_recs);
920 return size / sizeof(struct ocfs2_extent_rec);
923 static inline u16 ocfs2_local_alloc_size(struct super_block *sb)
927 size = sb->s_blocksize -
928 offsetof(struct ocfs2_dinode, id2.i_lab.la_bitmap);
933 static inline int ocfs2_group_bitmap_size(struct super_block *sb)
937 size = sb->s_blocksize -
938 offsetof(struct ocfs2_group_desc, bg_bitmap);
943 static inline int ocfs2_truncate_recs_per_inode(struct super_block *sb)
947 size = sb->s_blocksize -
948 offsetof(struct ocfs2_dinode, id2.i_dealloc.tl_recs);
950 return size / sizeof(struct ocfs2_truncate_rec);
953 static inline u64 ocfs2_backup_super_blkno(struct super_block *sb, int index)
955 u64 offset = OCFS2_BACKUP_SB_START;
957 if (index >= 0 && index < OCFS2_MAX_BACKUP_SUPERBLOCKS) {
958 offset <<= (2 * index);
959 offset >>= sb->s_blocksize_bits;
967 static inline int ocfs2_fast_symlink_chars(int blocksize)
969 return blocksize - offsetof(struct ocfs2_dinode, id2.i_symlink);
972 static inline int ocfs2_max_inline_data(int blocksize)
974 return blocksize - offsetof(struct ocfs2_dinode, id2.i_data.id_data);
977 static inline int ocfs2_extent_recs_per_inode(int blocksize)
982 offsetof(struct ocfs2_dinode, id2.i_list.l_recs);
984 return size / sizeof(struct ocfs2_extent_rec);
987 static inline int ocfs2_chain_recs_per_inode(int blocksize)
992 offsetof(struct ocfs2_dinode, id2.i_chain.cl_recs);
994 return size / sizeof(struct ocfs2_chain_rec);
997 static inline int ocfs2_extent_recs_per_eb(int blocksize)
1002 offsetof(struct ocfs2_extent_block, h_list.l_recs);
1004 return size / sizeof(struct ocfs2_extent_rec);
1007 static inline int ocfs2_local_alloc_size(int blocksize)
1012 offsetof(struct ocfs2_dinode, id2.i_lab.la_bitmap);
1017 static inline int ocfs2_group_bitmap_size(int blocksize)
1022 offsetof(struct ocfs2_group_desc, bg_bitmap);
1027 static inline int ocfs2_truncate_recs_per_inode(int blocksize)
1032 offsetof(struct ocfs2_dinode, id2.i_dealloc.tl_recs);
1034 return size / sizeof(struct ocfs2_truncate_rec);
1037 static inline uint64_t ocfs2_backup_super_blkno(int blocksize, int index)
1039 uint64_t offset = OCFS2_BACKUP_SB_START;
1041 if (index >= 0 && index < OCFS2_MAX_BACKUP_SUPERBLOCKS) {
1042 offset <<= (2 * index);
1043 offset /= blocksize;
1049 #endif /* __KERNEL__ */
1052 static inline int ocfs2_system_inode_is_global(int type)
1054 return ((type >= 0) &&
1055 (type <= OCFS2_LAST_GLOBAL_SYSTEM_INODE));
1058 static inline int ocfs2_sprintf_system_inode_name(char *buf, int len,
1064 * Global system inodes can only have one copy. Everything
1065 * after OCFS2_LAST_GLOBAL_SYSTEM_INODE in the system inode
1066 * list has a copy per slot.
1068 if (type <= OCFS2_LAST_GLOBAL_SYSTEM_INODE)
1069 chars = snprintf(buf, len, "%s",
1070 ocfs2_system_inodes[type].si_name);
1072 chars = snprintf(buf, len,
1073 ocfs2_system_inodes[type].si_name,
1079 static inline void ocfs2_set_de_type(struct ocfs2_dir_entry *de,
1082 de->file_type = ocfs2_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
1085 #endif /* _OCFS2_FS_H */