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
303 struct ocfs2_system_inode_info
{
309 /* System file index */
311 BAD_BLOCK_SYSTEM_INODE
= 0,
312 GLOBAL_INODE_ALLOC_SYSTEM_INODE
,
313 SLOT_MAP_SYSTEM_INODE
,
314 #define OCFS2_FIRST_ONLINE_SYSTEM_INODE SLOT_MAP_SYSTEM_INODE
315 HEARTBEAT_SYSTEM_INODE
,
316 GLOBAL_BITMAP_SYSTEM_INODE
,
317 #define OCFS2_LAST_GLOBAL_SYSTEM_INODE GLOBAL_BITMAP_SYSTEM_INODE
318 ORPHAN_DIR_SYSTEM_INODE
,
319 EXTENT_ALLOC_SYSTEM_INODE
,
320 INODE_ALLOC_SYSTEM_INODE
,
321 JOURNAL_SYSTEM_INODE
,
322 LOCAL_ALLOC_SYSTEM_INODE
,
323 TRUNCATE_LOG_SYSTEM_INODE
,
327 static struct ocfs2_system_inode_info ocfs2_system_inodes
[NUM_SYSTEM_INODES
] = {
328 /* Global system inodes (single copy) */
329 /* The first two are only used from userspace mfks/tunefs */
330 [BAD_BLOCK_SYSTEM_INODE
] = { "bad_blocks", 0, S_IFREG
| 0644 },
331 [GLOBAL_INODE_ALLOC_SYSTEM_INODE
] = { "global_inode_alloc", OCFS2_BITMAP_FL
| OCFS2_CHAIN_FL
, S_IFREG
| 0644 },
333 /* These are used by the running filesystem */
334 [SLOT_MAP_SYSTEM_INODE
] = { "slot_map", 0, S_IFREG
| 0644 },
335 [HEARTBEAT_SYSTEM_INODE
] = { "heartbeat", OCFS2_HEARTBEAT_FL
, S_IFREG
| 0644 },
336 [GLOBAL_BITMAP_SYSTEM_INODE
] = { "global_bitmap", 0, S_IFREG
| 0644 },
338 /* Slot-specific system inodes (one copy per slot) */
339 [ORPHAN_DIR_SYSTEM_INODE
] = { "orphan_dir:%04d", 0, S_IFDIR
| 0755 },
340 [EXTENT_ALLOC_SYSTEM_INODE
] = { "extent_alloc:%04d", OCFS2_BITMAP_FL
| OCFS2_CHAIN_FL
, S_IFREG
| 0644 },
341 [INODE_ALLOC_SYSTEM_INODE
] = { "inode_alloc:%04d", OCFS2_BITMAP_FL
| OCFS2_CHAIN_FL
, S_IFREG
| 0644 },
342 [JOURNAL_SYSTEM_INODE
] = { "journal:%04d", OCFS2_JOURNAL_FL
, S_IFREG
| 0644 },
343 [LOCAL_ALLOC_SYSTEM_INODE
] = { "local_alloc:%04d", OCFS2_BITMAP_FL
| OCFS2_LOCAL_ALLOC_FL
, S_IFREG
| 0644 },
344 [TRUNCATE_LOG_SYSTEM_INODE
] = { "truncate_log:%04d", OCFS2_DEALLOC_FL
, S_IFREG
| 0644 }
347 /* Parameter passed from mount.ocfs2 to module */
348 #define OCFS2_HB_NONE "heartbeat=none"
349 #define OCFS2_HB_LOCAL "heartbeat=local"
352 * OCFS2 directory file types. Only the low 3 bits are used. The
353 * other bits are reserved for now.
355 #define OCFS2_FT_UNKNOWN 0
356 #define OCFS2_FT_REG_FILE 1
357 #define OCFS2_FT_DIR 2
358 #define OCFS2_FT_CHRDEV 3
359 #define OCFS2_FT_BLKDEV 4
360 #define OCFS2_FT_FIFO 5
361 #define OCFS2_FT_SOCK 6
362 #define OCFS2_FT_SYMLINK 7
364 #define OCFS2_FT_MAX 8
367 * OCFS2_DIR_PAD defines the directory entries boundaries
369 * NOTE: It must be a multiple of 4
371 #define OCFS2_DIR_PAD 4
372 #define OCFS2_DIR_ROUND (OCFS2_DIR_PAD - 1)
373 #define OCFS2_DIR_MEMBER_LEN offsetof(struct ocfs2_dir_entry, name)
374 #define OCFS2_DIR_REC_LEN(name_len) (((name_len) + OCFS2_DIR_MEMBER_LEN + \
378 #define OCFS2_LINK_MAX 32000
381 static unsigned char ocfs2_type_by_mode
[S_IFMT
>> S_SHIFT
] = {
382 [S_IFREG
>> S_SHIFT
] = OCFS2_FT_REG_FILE
,
383 [S_IFDIR
>> S_SHIFT
] = OCFS2_FT_DIR
,
384 [S_IFCHR
>> S_SHIFT
] = OCFS2_FT_CHRDEV
,
385 [S_IFBLK
>> S_SHIFT
] = OCFS2_FT_BLKDEV
,
386 [S_IFIFO
>> S_SHIFT
] = OCFS2_FT_FIFO
,
387 [S_IFSOCK
>> S_SHIFT
] = OCFS2_FT_SOCK
,
388 [S_IFLNK
>> S_SHIFT
] = OCFS2_FT_SYMLINK
,
395 #define OCFS2_RAW_SB(dinode) (&((dinode)->id2.i_super))
398 * On disk extent record for OCFS2
399 * It describes a range of clusters on disk.
401 * Length fields are divided into interior and leaf node versions.
402 * This leaves room for a flags field (OCFS2_EXT_*) in the leaf nodes.
404 struct ocfs2_extent_rec
{
405 /*00*/ __le32 e_cpos
; /* Offset into the file, in clusters */
407 __le32 e_int_clusters
; /* Clusters covered by all children */
409 __le16 e_leaf_clusters
; /* Clusters covered by this
412 __u8 e_flags
; /* Extent flags */
415 __le64 e_blkno
; /* Physical disk offset, in blocks */
419 struct ocfs2_chain_rec
{
420 __le32 c_free
; /* Number of free bits in this chain. */
421 __le32 c_total
; /* Number of total bits in this chain */
422 __le64 c_blkno
; /* Physical disk offset (blocks) of 1st group */
425 struct ocfs2_truncate_rec
{
426 __le32 t_start
; /* 1st cluster in this log */
427 __le32 t_clusters
; /* Number of total clusters covered */
431 * On disk extent list for OCFS2 (node in the tree). Note that this
432 * is contained inside ocfs2_dinode or ocfs2_extent_block, so the
433 * offsets are relative to ocfs2_dinode.id2.i_list or
434 * ocfs2_extent_block.h_list, respectively.
436 struct ocfs2_extent_list
{
437 /*00*/ __le16 l_tree_depth
; /* Extent tree depth from this
438 point. 0 means data extents
439 hang directly off this
441 NOTE: The high 8 bits cannot be
442 used - tree_depth is never that big.
444 __le16 l_count
; /* Number of extent records */
445 __le16 l_next_free_rec
; /* Next unused extent slot */
447 __le64 l_reserved2
; /* Pad to
448 sizeof(ocfs2_extent_rec) */
449 /*10*/ struct ocfs2_extent_rec l_recs
[0]; /* Extent records */
453 * On disk allocation chain list for OCFS2. Note that this is
454 * contained inside ocfs2_dinode, so the offsets are relative to
455 * ocfs2_dinode.id2.i_chain.
457 struct ocfs2_chain_list
{
458 /*00*/ __le16 cl_cpg
; /* Clusters per Block Group */
459 __le16 cl_bpc
; /* Bits per cluster */
460 __le16 cl_count
; /* Total chains in this list */
461 __le16 cl_next_free_rec
; /* Next unused chain slot */
463 /*10*/ struct ocfs2_chain_rec cl_recs
[0]; /* Chain records */
467 * On disk deallocation log for OCFS2. Note that this is
468 * contained inside ocfs2_dinode, so the offsets are relative to
469 * ocfs2_dinode.id2.i_dealloc.
471 struct ocfs2_truncate_log
{
472 /*00*/ __le16 tl_count
; /* Total records in this log */
473 __le16 tl_used
; /* Number of records in use */
475 /*08*/ struct ocfs2_truncate_rec tl_recs
[0]; /* Truncate records */
479 * On disk extent block (indirect block) for OCFS2
481 struct ocfs2_extent_block
483 /*00*/ __u8 h_signature
[8]; /* Signature for verification */
485 /*10*/ __le16 h_suballoc_slot
; /* Slot suballocator this
486 extent_header belongs to */
487 __le16 h_suballoc_bit
; /* Bit offset in suballocator
489 __le32 h_fs_generation
; /* Must match super block */
490 __le64 h_blkno
; /* Offset on disk, in blocks */
491 /*20*/ __le64 h_reserved3
;
492 __le64 h_next_leaf_blk
; /* Offset on disk, in blocks,
493 of next leaf header pointing
495 /*30*/ struct ocfs2_extent_list h_list
; /* Extent record list */
496 /* Actual on-disk size is one block */
500 * On disk slot map for OCFS2. This defines the contents of the "slot_map"
501 * system file. A slot is valid if it contains a node number >= 0. The
502 * value -1 (0xFFFF) is OCFS2_INVALID_SLOT. This marks a slot empty.
504 struct ocfs2_slot_map
{
505 /*00*/ __le16 sm_slots
[0];
507 * Actual on-disk size is one block. OCFS2_MAX_SLOTS is 255,
508 * 255 * sizeof(__le16) == 512B, within the 512B block minimum blocksize.
512 struct ocfs2_extended_slot
{
513 /*00*/ __u8 es_valid
;
514 __u8 es_reserved1
[3];
520 * The extended slot map, used when OCFS2_FEATURE_INCOMPAT_EXTENDED_SLOT_MAP
521 * is set. It separates out the valid marker from the node number, and
522 * has room to grow. Unlike the old slot map, this format is defined by
525 struct ocfs2_slot_map_extended
{
526 /*00*/ struct ocfs2_extended_slot se_slots
[0];
528 * Actual size is i_size of the slot_map system file. It should
529 * match s_max_slots * sizeof(struct ocfs2_extended_slot)
533 struct ocfs2_cluster_info
{
534 /*00*/ __u8 ci_stack
[OCFS2_STACK_LABEL_LEN
];
536 /*08*/ __u8 ci_cluster
[OCFS2_CLUSTER_NAME_LEN
];
541 * On disk superblock for OCFS2
542 * Note that it is contained inside an ocfs2_dinode, so all offsets
543 * are relative to the start of ocfs2_dinode.id2.
545 struct ocfs2_super_block
{
546 /*00*/ __le16 s_major_rev_level
;
547 __le16 s_minor_rev_level
;
549 __le16 s_max_mnt_count
;
550 __le16 s_state
; /* File system state */
551 __le16 s_errors
; /* Behaviour when detecting errors */
552 __le32 s_checkinterval
; /* Max time between checks */
553 /*10*/ __le64 s_lastcheck
; /* Time of last check */
554 __le32 s_creator_os
; /* OS */
555 __le32 s_feature_compat
; /* Compatible feature set */
556 /*20*/ __le32 s_feature_incompat
; /* Incompatible feature set */
557 __le32 s_feature_ro_compat
; /* Readonly-compatible feature set */
558 __le64 s_root_blkno
; /* Offset, in blocks, of root directory
560 /*30*/ __le64 s_system_dir_blkno
; /* Offset, in blocks, of system
562 __le32 s_blocksize_bits
; /* Blocksize for this fs */
563 __le32 s_clustersize_bits
; /* Clustersize for this fs */
564 /*40*/ __le16 s_max_slots
; /* Max number of simultaneous mounts
565 before tunefs required */
566 __le16 s_tunefs_flag
;
568 __le64 s_first_cluster_group
; /* Block offset of 1st cluster
570 /*50*/ __u8 s_label
[OCFS2_MAX_VOL_LABEL_LEN
]; /* Label for mounting, etc. */
571 /*90*/ __u8 s_uuid
[OCFS2_VOL_UUID_LEN
]; /* 128-bit uuid */
572 /*A0*/ struct ocfs2_cluster_info s_cluster_info
; /* Selected userspace
574 with INCOMPAT flag. */
575 /*B8*/ __le64 s_reserved2
[17]; /* Fill out superblock */
579 * NOTE: As stated above, all offsets are relative to
580 * ocfs2_dinode.id2, which is at 0xC0 in the inode.
581 * 0xC0 + 0x140 = 0x200 or 512 bytes. A superblock must fit within
582 * our smallest blocksize, which is 512 bytes. To ensure this,
583 * we reserve the space in s_reserved2. Anything past s_reserved2
584 * will not be available on the smallest blocksize.
589 * Local allocation bitmap for OCFS2 slots
590 * Note that it exists inside an ocfs2_dinode, so all offsets are
591 * relative to the start of ocfs2_dinode.id2.
593 struct ocfs2_local_alloc
595 /*00*/ __le32 la_bm_off
; /* Starting bit offset in main bitmap */
596 __le16 la_size
; /* Size of included bitmap, in bytes */
599 /*10*/ __u8 la_bitmap
[0];
603 * Data-in-inode header. This is only used if i_dyn_features has
604 * OCFS2_INLINE_DATA_FL set.
606 struct ocfs2_inline_data
608 /*00*/ __le16 id_count
; /* Number of bytes that can be used
609 * for data, starting at id_data */
612 __u8 id_data
[0]; /* Start of user data */
616 * On disk inode for OCFS2
618 struct ocfs2_dinode
{
619 /*00*/ __u8 i_signature
[8]; /* Signature for validation */
620 __le32 i_generation
; /* Generation number */
621 __le16 i_suballoc_slot
; /* Slot suballocator this inode
623 __le16 i_suballoc_bit
; /* Bit offset in suballocator
625 /*10*/ __le32 i_reserved0
;
626 __le32 i_clusters
; /* Cluster count */
627 __le32 i_uid
; /* Owner UID */
628 __le32 i_gid
; /* Owning GID */
629 /*20*/ __le64 i_size
; /* Size in bytes */
630 __le16 i_mode
; /* File mode */
631 __le16 i_links_count
; /* Links count */
632 __le32 i_flags
; /* File flags */
633 /*30*/ __le64 i_atime
; /* Access time */
634 __le64 i_ctime
; /* Creation time */
635 /*40*/ __le64 i_mtime
; /* Modification time */
636 __le64 i_dtime
; /* Deletion time */
637 /*50*/ __le64 i_blkno
; /* Offset on disk, in blocks */
638 __le64 i_last_eb_blk
; /* Pointer to last extent
640 /*60*/ __le32 i_fs_generation
; /* Generation per fs-instance */
645 __le16 i_orphaned_slot
; /* Only valid when OCFS2_ORPHANED_FL
646 was set in i_flags */
647 __le16 i_dyn_features
;
648 /*70*/ __le64 i_reserved2
[8];
650 __le64 i_pad1
; /* Generic way to refer to this
653 __le64 i_rdev
; /* Device number */
655 struct { /* Info for bitmap system
657 __le32 i_used
; /* Bits (ie, clusters) used */
658 __le32 i_total
; /* Total bits (clusters)
661 struct { /* Info for journal system
663 __le32 ij_flags
; /* Mounted, version, etc. */
664 __le32 ij_recovery_generation
; /* Incremented when the
669 } id1
; /* Inode type dependant 1 */
671 struct ocfs2_super_block i_super
;
672 struct ocfs2_local_alloc i_lab
;
673 struct ocfs2_chain_list i_chain
;
674 struct ocfs2_extent_list i_list
;
675 struct ocfs2_truncate_log i_dealloc
;
676 struct ocfs2_inline_data i_data
;
679 /* Actual on-disk size is one block */
683 * On-disk directory entry structure for OCFS2
685 * Packed as this structure could be accessed unaligned on 64-bit platforms
687 struct ocfs2_dir_entry
{
688 /*00*/ __le64 inode
; /* Inode number */
689 __le16 rec_len
; /* Directory entry length */
690 __u8 name_len
; /* Name length */
692 /*0C*/ char name
[OCFS2_MAX_FILENAME_LEN
]; /* File name */
693 /* Actual on-disk length specified by rec_len */
694 } __attribute__ ((packed
));
697 * On disk allocator group structure for OCFS2
699 struct ocfs2_group_desc
701 /*00*/ __u8 bg_signature
[8]; /* Signature for validation */
702 __le16 bg_size
; /* Size of included bitmap in
704 __le16 bg_bits
; /* Bits represented by this
706 __le16 bg_free_bits_count
; /* Free bits count */
707 __le16 bg_chain
; /* What chain I am in. */
708 /*10*/ __le32 bg_generation
;
710 __le64 bg_next_group
; /* Next group in my list, in
712 /*20*/ __le64 bg_parent_dinode
; /* dinode which owns me, in
714 __le64 bg_blkno
; /* Offset on disk, in blocks */
715 /*30*/ __le64 bg_reserved2
[2];
716 /*40*/ __u8 bg_bitmap
[0];
720 * On disk extended attribute structure for OCFS2.
724 * ocfs2_xattr_entry indicates one extend attribute.
726 * Note that it can be stored in inode, one block or one xattr bucket.
728 struct ocfs2_xattr_entry
{
729 __le32 xe_name_hash
; /* hash value of xattr prefix+suffix. */
730 __le16 xe_name_offset
; /* byte offset from the 1st etnry in the local
731 local xattr storage(inode, xattr block or
733 __u8 xe_name_len
; /* xattr name len, does't include prefix. */
734 __u8 xe_type
; /* the low 7 bits indicates the name prefix's
735 * type and the highest 1 bits indicate whether
736 * the EA is stored in the local storage. */
737 __le64 xe_value_size
; /* real xattr value length. */
741 * On disk structure for xattr header.
743 * One ocfs2_xattr_header describes how many ocfs2_xattr_entry records in
744 * the local xattr storage.
746 struct ocfs2_xattr_header
{
747 __le16 xh_count
; /* contains the count of how
748 many records are in the
749 local xattr storage. */
753 struct ocfs2_xattr_entry xh_entries
[0]; /* xattr entry list. */
757 * On disk structure for xattr value root.
759 * It is used when one extended attribute's size is larger, and we will save it
760 * in an outside cluster. It will stored in a b-tree like file content.
762 struct ocfs2_xattr_value_root
{
763 /*00*/ __le32 xr_clusters
; /* clusters covered by xattr value. */
765 __le64 xr_last_eb_blk
; /* Pointer to last extent block */
766 /*10*/ struct ocfs2_extent_list xr_list
; /* Extent record list */
770 * On disk structure for xattr tree root.
772 * It is used when there are too many extended attributes for one file. These
773 * attributes will be organized and stored in an indexed-btree.
775 struct ocfs2_xattr_tree_root
{
776 /*00*/ __le32 xt_clusters
; /* clusters covered by xattr. */
778 __le64 xt_last_eb_blk
; /* Pointer to last extent block */
779 /*10*/ struct ocfs2_extent_list xt_list
; /* Extent record list */
782 #define OCFS2_XATTR_INDEXED 0x1
785 * On disk structure for xattr block.
787 struct ocfs2_xattr_block
{
788 /*00*/ __u8 xb_signature
[8]; /* Signature for verification */
789 __le16 xb_suballoc_slot
; /* Slot suballocator this
791 __le16 xb_suballoc_bit
; /* Bit offset in suballocator
793 __le32 xb_fs_generation
; /* Must match super block */
794 /*10*/ __le64 xb_blkno
; /* Offset on disk, in blocks */
796 /*20*/ __le16 xb_flags
; /* Indicates whether this block contains
797 real xattr or a xattr tree. */
802 struct ocfs2_xattr_header xb_header
; /* xattr header if this
803 block contains xattr */
804 struct ocfs2_xattr_tree_root xb_root
;/* xattr tree root if this
810 #define OCFS2_XATTR_ENTRY_LOCAL 0x80
811 #define OCFS2_XATTR_TYPE_MASK 0x7F
812 static inline void ocfs2_xattr_set_local(struct ocfs2_xattr_entry
*xe
,
816 xe
->xe_type
|= OCFS2_XATTR_ENTRY_LOCAL
;
818 xe
->xe_type
&= ~OCFS2_XATTR_ENTRY_LOCAL
;
821 static inline int ocfs2_xattr_is_local(struct ocfs2_xattr_entry
*xe
)
823 return xe
->xe_type
& OCFS2_XATTR_ENTRY_LOCAL
;
826 static inline void ocfs2_xattr_set_type(struct ocfs2_xattr_entry
*xe
, int type
)
828 xe
->xe_type
|= type
& OCFS2_XATTR_TYPE_MASK
;
831 static inline int ocfs2_xattr_get_type(struct ocfs2_xattr_entry
*xe
)
833 return xe
->xe_type
& OCFS2_XATTR_TYPE_MASK
;
837 static inline int ocfs2_fast_symlink_chars(struct super_block
*sb
)
839 return sb
->s_blocksize
-
840 offsetof(struct ocfs2_dinode
, id2
.i_symlink
);
843 static inline int ocfs2_max_inline_data(struct super_block
*sb
)
845 return sb
->s_blocksize
-
846 offsetof(struct ocfs2_dinode
, id2
.i_data
.id_data
);
849 static inline int ocfs2_extent_recs_per_inode(struct super_block
*sb
)
853 size
= sb
->s_blocksize
-
854 offsetof(struct ocfs2_dinode
, id2
.i_list
.l_recs
);
856 return size
/ sizeof(struct ocfs2_extent_rec
);
859 static inline int ocfs2_chain_recs_per_inode(struct super_block
*sb
)
863 size
= sb
->s_blocksize
-
864 offsetof(struct ocfs2_dinode
, id2
.i_chain
.cl_recs
);
866 return size
/ sizeof(struct ocfs2_chain_rec
);
869 static inline u16
ocfs2_extent_recs_per_eb(struct super_block
*sb
)
873 size
= sb
->s_blocksize
-
874 offsetof(struct ocfs2_extent_block
, h_list
.l_recs
);
876 return size
/ sizeof(struct ocfs2_extent_rec
);
879 static inline u16
ocfs2_local_alloc_size(struct super_block
*sb
)
883 size
= sb
->s_blocksize
-
884 offsetof(struct ocfs2_dinode
, id2
.i_lab
.la_bitmap
);
889 static inline int ocfs2_group_bitmap_size(struct super_block
*sb
)
893 size
= sb
->s_blocksize
-
894 offsetof(struct ocfs2_group_desc
, bg_bitmap
);
899 static inline int ocfs2_truncate_recs_per_inode(struct super_block
*sb
)
903 size
= sb
->s_blocksize
-
904 offsetof(struct ocfs2_dinode
, id2
.i_dealloc
.tl_recs
);
906 return size
/ sizeof(struct ocfs2_truncate_rec
);
909 static inline u64
ocfs2_backup_super_blkno(struct super_block
*sb
, int index
)
911 u64 offset
= OCFS2_BACKUP_SB_START
;
913 if (index
>= 0 && index
< OCFS2_MAX_BACKUP_SUPERBLOCKS
) {
914 offset
<<= (2 * index
);
915 offset
>>= sb
->s_blocksize_bits
;
923 static inline int ocfs2_fast_symlink_chars(int blocksize
)
925 return blocksize
- offsetof(struct ocfs2_dinode
, id2
.i_symlink
);
928 static inline int ocfs2_max_inline_data(int blocksize
)
930 return blocksize
- offsetof(struct ocfs2_dinode
, id2
.i_data
.id_data
);
933 static inline int ocfs2_extent_recs_per_inode(int blocksize
)
938 offsetof(struct ocfs2_dinode
, id2
.i_list
.l_recs
);
940 return size
/ sizeof(struct ocfs2_extent_rec
);
943 static inline int ocfs2_chain_recs_per_inode(int blocksize
)
948 offsetof(struct ocfs2_dinode
, id2
.i_chain
.cl_recs
);
950 return size
/ sizeof(struct ocfs2_chain_rec
);
953 static inline int ocfs2_extent_recs_per_eb(int blocksize
)
958 offsetof(struct ocfs2_extent_block
, h_list
.l_recs
);
960 return size
/ sizeof(struct ocfs2_extent_rec
);
963 static inline int ocfs2_local_alloc_size(int blocksize
)
968 offsetof(struct ocfs2_dinode
, id2
.i_lab
.la_bitmap
);
973 static inline int ocfs2_group_bitmap_size(int blocksize
)
978 offsetof(struct ocfs2_group_desc
, bg_bitmap
);
983 static inline int ocfs2_truncate_recs_per_inode(int blocksize
)
988 offsetof(struct ocfs2_dinode
, id2
.i_dealloc
.tl_recs
);
990 return size
/ sizeof(struct ocfs2_truncate_rec
);
993 static inline uint64_t ocfs2_backup_super_blkno(int blocksize
, int index
)
995 uint64_t offset
= OCFS2_BACKUP_SB_START
;
997 if (index
>= 0 && index
< OCFS2_MAX_BACKUP_SUPERBLOCKS
) {
998 offset
<<= (2 * index
);
1005 #endif /* __KERNEL__ */
1008 static inline int ocfs2_system_inode_is_global(int type
)
1010 return ((type
>= 0) &&
1011 (type
<= OCFS2_LAST_GLOBAL_SYSTEM_INODE
));
1014 static inline int ocfs2_sprintf_system_inode_name(char *buf
, int len
,
1020 * Global system inodes can only have one copy. Everything
1021 * after OCFS2_LAST_GLOBAL_SYSTEM_INODE in the system inode
1022 * list has a copy per slot.
1024 if (type
<= OCFS2_LAST_GLOBAL_SYSTEM_INODE
)
1025 chars
= snprintf(buf
, len
, "%s",
1026 ocfs2_system_inodes
[type
].si_name
);
1028 chars
= snprintf(buf
, len
,
1029 ocfs2_system_inodes
[type
].si_name
,
1035 static inline void ocfs2_set_de_type(struct ocfs2_dir_entry
*de
,
1038 de
->file_type
= ocfs2_type_by_mode
[(mode
& S_IFMT
)>>S_SHIFT
];
1041 #endif /* _OCFS2_FS_H */