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 | OCFS2_FEATURE_COMPAT_JBD2_SB)
91 #define OCFS2_FEATURE_INCOMPAT_SUPP (OCFS2_FEATURE_INCOMPAT_LOCAL_MOUNT \
92 | OCFS2_FEATURE_INCOMPAT_SPARSE_ALLOC \
93 | OCFS2_FEATURE_INCOMPAT_INLINE_DATA \
94 | OCFS2_FEATURE_INCOMPAT_EXTENDED_SLOT_MAP \
95 | OCFS2_FEATURE_INCOMPAT_USERSPACE_STACK \
96 | OCFS2_FEATURE_INCOMPAT_XATTR)
97 #define OCFS2_FEATURE_RO_COMPAT_SUPP OCFS2_FEATURE_RO_COMPAT_UNWRITTEN
100 * Heartbeat-only devices are missing journals and other files. The
101 * filesystem driver can't load them, but the library can. Never put
102 * this in OCFS2_FEATURE_INCOMPAT_SUPP, *ever*.
104 #define OCFS2_FEATURE_INCOMPAT_HEARTBEAT_DEV 0x0002
107 * tunefs sets this incompat flag before starting the resize and clears it
108 * at the end. This flag protects users from inadvertently mounting the fs
109 * after an aborted run without fsck-ing.
111 #define OCFS2_FEATURE_INCOMPAT_RESIZE_INPROG 0x0004
113 /* Used to denote a non-clustered volume */
114 #define OCFS2_FEATURE_INCOMPAT_LOCAL_MOUNT 0x0008
116 /* Support for sparse allocation in b-trees */
117 #define OCFS2_FEATURE_INCOMPAT_SPARSE_ALLOC 0x0010
120 * Tunefs sets this incompat flag before starting an operation which
121 * would require cleanup on abort. This is done to protect users from
122 * inadvertently mounting the fs after an aborted run without
125 * s_tunefs_flags on the super block describes precisely which
126 * operations were in progress.
128 #define OCFS2_FEATURE_INCOMPAT_TUNEFS_INPROG 0x0020
130 /* Support for data packed into inode blocks */
131 #define OCFS2_FEATURE_INCOMPAT_INLINE_DATA 0x0040
134 * Support for alternate, userspace cluster stacks. If set, the superblock
135 * field s_cluster_info contains a tag for the alternate stack in use as
136 * well as the name of the cluster being joined.
137 * mount.ocfs2 must pass in a matching stack name.
139 * If not set, the classic stack will be used. This is compatbile with
140 * all older versions.
142 #define OCFS2_FEATURE_INCOMPAT_USERSPACE_STACK 0x0080
144 /* Support for the extended slot map */
145 #define OCFS2_FEATURE_INCOMPAT_EXTENDED_SLOT_MAP 0x100
147 /* Support for extended attributes */
148 #define OCFS2_FEATURE_INCOMPAT_XATTR 0x0200
151 * backup superblock flag is used to indicate that this volume
152 * has backup superblocks.
154 #define OCFS2_FEATURE_COMPAT_BACKUP_SB 0x0001
157 * The filesystem will correctly handle journal feature bits.
159 #define OCFS2_FEATURE_COMPAT_JBD2_SB 0x0002
162 * Unwritten extents support.
164 #define OCFS2_FEATURE_RO_COMPAT_UNWRITTEN 0x0001
166 /* The byte offset of the first backup block will be 1G.
167 * The following will be 4G, 16G, 64G, 256G and 1T.
169 #define OCFS2_BACKUP_SB_START 1 << 30
171 /* the max backup superblock nums */
172 #define OCFS2_MAX_BACKUP_SUPERBLOCKS 6
175 * Flags on ocfs2_super_block.s_tunefs_flags
177 #define OCFS2_TUNEFS_INPROG_REMOVE_SLOT 0x0001 /* Removing slots */
180 * Flags on ocfs2_dinode.i_flags
182 #define OCFS2_VALID_FL (0x00000001) /* Inode is valid */
183 #define OCFS2_UNUSED2_FL (0x00000002)
184 #define OCFS2_ORPHANED_FL (0x00000004) /* On the orphan list */
185 #define OCFS2_UNUSED3_FL (0x00000008)
186 /* System inode flags */
187 #define OCFS2_SYSTEM_FL (0x00000010) /* System inode */
188 #define OCFS2_SUPER_BLOCK_FL (0x00000020) /* Super block */
189 #define OCFS2_LOCAL_ALLOC_FL (0x00000040) /* Slot local alloc bitmap */
190 #define OCFS2_BITMAP_FL (0x00000080) /* Allocation bitmap */
191 #define OCFS2_JOURNAL_FL (0x00000100) /* Slot local journal */
192 #define OCFS2_HEARTBEAT_FL (0x00000200) /* Heartbeat area */
193 #define OCFS2_CHAIN_FL (0x00000400) /* Chain allocator */
194 #define OCFS2_DEALLOC_FL (0x00000800) /* Truncate log */
197 * Flags on ocfs2_dinode.i_dyn_features
199 * These can change much more often than i_flags. When adding flags,
200 * keep in mind that i_dyn_features is only 16 bits wide.
202 #define OCFS2_INLINE_DATA_FL (0x0001) /* Data stored in inode block */
203 #define OCFS2_HAS_XATTR_FL (0x0002)
204 #define OCFS2_INLINE_XATTR_FL (0x0004)
205 #define OCFS2_INDEXED_DIR_FL (0x0008)
207 /* Inode attributes, keep in sync with EXT2 */
208 #define OCFS2_SECRM_FL (0x00000001) /* Secure deletion */
209 #define OCFS2_UNRM_FL (0x00000002) /* Undelete */
210 #define OCFS2_COMPR_FL (0x00000004) /* Compress file */
211 #define OCFS2_SYNC_FL (0x00000008) /* Synchronous updates */
212 #define OCFS2_IMMUTABLE_FL (0x00000010) /* Immutable file */
213 #define OCFS2_APPEND_FL (0x00000020) /* writes to file may only append */
214 #define OCFS2_NODUMP_FL (0x00000040) /* do not dump file */
215 #define OCFS2_NOATIME_FL (0x00000080) /* do not update atime */
216 #define OCFS2_DIRSYNC_FL (0x00010000) /* dirsync behaviour (directories only) */
218 #define OCFS2_FL_VISIBLE (0x000100FF) /* User visible flags */
219 #define OCFS2_FL_MODIFIABLE (0x000100FF) /* User modifiable flags */
222 * Extent record flags (e_node.leaf.flags)
224 #define OCFS2_EXT_UNWRITTEN (0x01) /* Extent is allocated but
230 #define OCFS2_IOC_GETFLAGS _IOR('f', 1, long)
231 #define OCFS2_IOC_SETFLAGS _IOW('f', 2, long)
232 #define OCFS2_IOC32_GETFLAGS _IOR('f', 1, int)
233 #define OCFS2_IOC32_SETFLAGS _IOW('f', 2, int)
236 * Space reservation / allocation / free ioctls and argument structure
237 * are designed to be compatible with XFS.
239 * ALLOCSP* and FREESP* are not and will never be supported, but are
240 * included here for completeness.
242 struct ocfs2_space_resv
{
246 __s64 l_len
; /* len == 0 means until end of file */
249 __s32 l_pad
[4]; /* reserve area */
252 #define OCFS2_IOC_ALLOCSP _IOW ('X', 10, struct ocfs2_space_resv)
253 #define OCFS2_IOC_FREESP _IOW ('X', 11, struct ocfs2_space_resv)
254 #define OCFS2_IOC_RESVSP _IOW ('X', 40, struct ocfs2_space_resv)
255 #define OCFS2_IOC_UNRESVSP _IOW ('X', 41, struct ocfs2_space_resv)
256 #define OCFS2_IOC_ALLOCSP64 _IOW ('X', 36, struct ocfs2_space_resv)
257 #define OCFS2_IOC_FREESP64 _IOW ('X', 37, struct ocfs2_space_resv)
258 #define OCFS2_IOC_RESVSP64 _IOW ('X', 42, struct ocfs2_space_resv)
259 #define OCFS2_IOC_UNRESVSP64 _IOW ('X', 43, struct ocfs2_space_resv)
261 /* Used to pass group descriptor data when online resize is done */
262 struct ocfs2_new_group_input
{
263 __u64 group
; /* Group descriptor's blkno. */
264 __u32 clusters
; /* Total number of clusters in this group */
265 __u32 frees
; /* Total free clusters in this group */
266 __u16 chain
; /* Chain for this group */
271 #define OCFS2_IOC_GROUP_EXTEND _IOW('o', 1, int)
272 #define OCFS2_IOC_GROUP_ADD _IOW('o', 2,struct ocfs2_new_group_input)
273 #define OCFS2_IOC_GROUP_ADD64 _IOW('o', 3,struct ocfs2_new_group_input)
276 * Journal Flags (ocfs2_dinode.id1.journal1.i_flags)
278 #define OCFS2_JOURNAL_DIRTY_FL (0x00000001) /* Journal needs recovery */
281 * superblock s_state flags
283 #define OCFS2_ERROR_FS (0x00000001) /* FS saw errors */
285 /* Limit of space in ocfs2_dir_entry */
286 #define OCFS2_MAX_FILENAME_LEN 255
288 /* Maximum slots on an ocfs2 file system */
289 #define OCFS2_MAX_SLOTS 255
291 /* Slot map indicator for an empty slot */
292 #define OCFS2_INVALID_SLOT -1
294 #define OCFS2_VOL_UUID_LEN 16
295 #define OCFS2_MAX_VOL_LABEL_LEN 64
297 /* The alternate, userspace stack fields */
298 #define OCFS2_STACK_LABEL_LEN 4
299 #define OCFS2_CLUSTER_NAME_LEN 16
301 /* Journal limits (in bytes) */
302 #define OCFS2_MIN_JOURNAL_SIZE (4 * 1024 * 1024)
305 * Default local alloc size (in megabytes)
307 * The value chosen should be such that most allocations, including new
308 * block groups, use local alloc.
310 #define OCFS2_DEFAULT_LOCAL_ALLOC_SIZE 8
313 * Inline extended attribute size (in bytes)
314 * The value chosen should be aligned to 16 byte boundaries.
316 #define OCFS2_MIN_XATTR_INLINE_SIZE 256
318 struct ocfs2_system_inode_info
{
324 /* System file index */
326 BAD_BLOCK_SYSTEM_INODE
= 0,
327 GLOBAL_INODE_ALLOC_SYSTEM_INODE
,
328 SLOT_MAP_SYSTEM_INODE
,
329 #define OCFS2_FIRST_ONLINE_SYSTEM_INODE SLOT_MAP_SYSTEM_INODE
330 HEARTBEAT_SYSTEM_INODE
,
331 GLOBAL_BITMAP_SYSTEM_INODE
,
332 #define OCFS2_LAST_GLOBAL_SYSTEM_INODE GLOBAL_BITMAP_SYSTEM_INODE
333 ORPHAN_DIR_SYSTEM_INODE
,
334 EXTENT_ALLOC_SYSTEM_INODE
,
335 INODE_ALLOC_SYSTEM_INODE
,
336 JOURNAL_SYSTEM_INODE
,
337 LOCAL_ALLOC_SYSTEM_INODE
,
338 TRUNCATE_LOG_SYSTEM_INODE
,
342 static struct ocfs2_system_inode_info ocfs2_system_inodes
[NUM_SYSTEM_INODES
] = {
343 /* Global system inodes (single copy) */
344 /* The first two are only used from userspace mfks/tunefs */
345 [BAD_BLOCK_SYSTEM_INODE
] = { "bad_blocks", 0, S_IFREG
| 0644 },
346 [GLOBAL_INODE_ALLOC_SYSTEM_INODE
] = { "global_inode_alloc", OCFS2_BITMAP_FL
| OCFS2_CHAIN_FL
, S_IFREG
| 0644 },
348 /* These are used by the running filesystem */
349 [SLOT_MAP_SYSTEM_INODE
] = { "slot_map", 0, S_IFREG
| 0644 },
350 [HEARTBEAT_SYSTEM_INODE
] = { "heartbeat", OCFS2_HEARTBEAT_FL
, S_IFREG
| 0644 },
351 [GLOBAL_BITMAP_SYSTEM_INODE
] = { "global_bitmap", 0, S_IFREG
| 0644 },
353 /* Slot-specific system inodes (one copy per slot) */
354 [ORPHAN_DIR_SYSTEM_INODE
] = { "orphan_dir:%04d", 0, S_IFDIR
| 0755 },
355 [EXTENT_ALLOC_SYSTEM_INODE
] = { "extent_alloc:%04d", OCFS2_BITMAP_FL
| OCFS2_CHAIN_FL
, S_IFREG
| 0644 },
356 [INODE_ALLOC_SYSTEM_INODE
] = { "inode_alloc:%04d", OCFS2_BITMAP_FL
| OCFS2_CHAIN_FL
, S_IFREG
| 0644 },
357 [JOURNAL_SYSTEM_INODE
] = { "journal:%04d", OCFS2_JOURNAL_FL
, S_IFREG
| 0644 },
358 [LOCAL_ALLOC_SYSTEM_INODE
] = { "local_alloc:%04d", OCFS2_BITMAP_FL
| OCFS2_LOCAL_ALLOC_FL
, S_IFREG
| 0644 },
359 [TRUNCATE_LOG_SYSTEM_INODE
] = { "truncate_log:%04d", OCFS2_DEALLOC_FL
, S_IFREG
| 0644 }
362 /* Parameter passed from mount.ocfs2 to module */
363 #define OCFS2_HB_NONE "heartbeat=none"
364 #define OCFS2_HB_LOCAL "heartbeat=local"
367 * OCFS2 directory file types. Only the low 3 bits are used. The
368 * other bits are reserved for now.
370 #define OCFS2_FT_UNKNOWN 0
371 #define OCFS2_FT_REG_FILE 1
372 #define OCFS2_FT_DIR 2
373 #define OCFS2_FT_CHRDEV 3
374 #define OCFS2_FT_BLKDEV 4
375 #define OCFS2_FT_FIFO 5
376 #define OCFS2_FT_SOCK 6
377 #define OCFS2_FT_SYMLINK 7
379 #define OCFS2_FT_MAX 8
382 * OCFS2_DIR_PAD defines the directory entries boundaries
384 * NOTE: It must be a multiple of 4
386 #define OCFS2_DIR_PAD 4
387 #define OCFS2_DIR_ROUND (OCFS2_DIR_PAD - 1)
388 #define OCFS2_DIR_MEMBER_LEN offsetof(struct ocfs2_dir_entry, name)
389 #define OCFS2_DIR_REC_LEN(name_len) (((name_len) + OCFS2_DIR_MEMBER_LEN + \
393 #define OCFS2_LINK_MAX 32000
396 static unsigned char ocfs2_type_by_mode
[S_IFMT
>> S_SHIFT
] = {
397 [S_IFREG
>> S_SHIFT
] = OCFS2_FT_REG_FILE
,
398 [S_IFDIR
>> S_SHIFT
] = OCFS2_FT_DIR
,
399 [S_IFCHR
>> S_SHIFT
] = OCFS2_FT_CHRDEV
,
400 [S_IFBLK
>> S_SHIFT
] = OCFS2_FT_BLKDEV
,
401 [S_IFIFO
>> S_SHIFT
] = OCFS2_FT_FIFO
,
402 [S_IFSOCK
>> S_SHIFT
] = OCFS2_FT_SOCK
,
403 [S_IFLNK
>> S_SHIFT
] = OCFS2_FT_SYMLINK
,
410 #define OCFS2_RAW_SB(dinode) (&((dinode)->id2.i_super))
413 * On disk extent record for OCFS2
414 * It describes a range of clusters on disk.
416 * Length fields are divided into interior and leaf node versions.
417 * This leaves room for a flags field (OCFS2_EXT_*) in the leaf nodes.
419 struct ocfs2_extent_rec
{
420 /*00*/ __le32 e_cpos
; /* Offset into the file, in clusters */
422 __le32 e_int_clusters
; /* Clusters covered by all children */
424 __le16 e_leaf_clusters
; /* Clusters covered by this
427 __u8 e_flags
; /* Extent flags */
430 __le64 e_blkno
; /* Physical disk offset, in blocks */
434 struct ocfs2_chain_rec
{
435 __le32 c_free
; /* Number of free bits in this chain. */
436 __le32 c_total
; /* Number of total bits in this chain */
437 __le64 c_blkno
; /* Physical disk offset (blocks) of 1st group */
440 struct ocfs2_truncate_rec
{
441 __le32 t_start
; /* 1st cluster in this log */
442 __le32 t_clusters
; /* Number of total clusters covered */
446 * On disk extent list for OCFS2 (node in the tree). Note that this
447 * is contained inside ocfs2_dinode or ocfs2_extent_block, so the
448 * offsets are relative to ocfs2_dinode.id2.i_list or
449 * ocfs2_extent_block.h_list, respectively.
451 struct ocfs2_extent_list
{
452 /*00*/ __le16 l_tree_depth
; /* Extent tree depth from this
453 point. 0 means data extents
454 hang directly off this
456 NOTE: The high 8 bits cannot be
457 used - tree_depth is never that big.
459 __le16 l_count
; /* Number of extent records */
460 __le16 l_next_free_rec
; /* Next unused extent slot */
462 __le64 l_reserved2
; /* Pad to
463 sizeof(ocfs2_extent_rec) */
464 /*10*/ struct ocfs2_extent_rec l_recs
[0]; /* Extent records */
468 * On disk allocation chain list for OCFS2. Note that this is
469 * contained inside ocfs2_dinode, so the offsets are relative to
470 * ocfs2_dinode.id2.i_chain.
472 struct ocfs2_chain_list
{
473 /*00*/ __le16 cl_cpg
; /* Clusters per Block Group */
474 __le16 cl_bpc
; /* Bits per cluster */
475 __le16 cl_count
; /* Total chains in this list */
476 __le16 cl_next_free_rec
; /* Next unused chain slot */
478 /*10*/ struct ocfs2_chain_rec cl_recs
[0]; /* Chain records */
482 * On disk deallocation log for OCFS2. Note that this is
483 * contained inside ocfs2_dinode, so the offsets are relative to
484 * ocfs2_dinode.id2.i_dealloc.
486 struct ocfs2_truncate_log
{
487 /*00*/ __le16 tl_count
; /* Total records in this log */
488 __le16 tl_used
; /* Number of records in use */
490 /*08*/ struct ocfs2_truncate_rec tl_recs
[0]; /* Truncate records */
494 * On disk extent block (indirect block) for OCFS2
496 struct ocfs2_extent_block
498 /*00*/ __u8 h_signature
[8]; /* Signature for verification */
500 /*10*/ __le16 h_suballoc_slot
; /* Slot suballocator this
501 extent_header belongs to */
502 __le16 h_suballoc_bit
; /* Bit offset in suballocator
504 __le32 h_fs_generation
; /* Must match super block */
505 __le64 h_blkno
; /* Offset on disk, in blocks */
506 /*20*/ __le64 h_reserved3
;
507 __le64 h_next_leaf_blk
; /* Offset on disk, in blocks,
508 of next leaf header pointing
510 /*30*/ struct ocfs2_extent_list h_list
; /* Extent record list */
511 /* Actual on-disk size is one block */
515 * On disk slot map for OCFS2. This defines the contents of the "slot_map"
516 * system file. A slot is valid if it contains a node number >= 0. The
517 * value -1 (0xFFFF) is OCFS2_INVALID_SLOT. This marks a slot empty.
519 struct ocfs2_slot_map
{
520 /*00*/ __le16 sm_slots
[0];
522 * Actual on-disk size is one block. OCFS2_MAX_SLOTS is 255,
523 * 255 * sizeof(__le16) == 512B, within the 512B block minimum blocksize.
527 struct ocfs2_extended_slot
{
528 /*00*/ __u8 es_valid
;
529 __u8 es_reserved1
[3];
535 * The extended slot map, used when OCFS2_FEATURE_INCOMPAT_EXTENDED_SLOT_MAP
536 * is set. It separates out the valid marker from the node number, and
537 * has room to grow. Unlike the old slot map, this format is defined by
540 struct ocfs2_slot_map_extended
{
541 /*00*/ struct ocfs2_extended_slot se_slots
[0];
543 * Actual size is i_size of the slot_map system file. It should
544 * match s_max_slots * sizeof(struct ocfs2_extended_slot)
548 struct ocfs2_cluster_info
{
549 /*00*/ __u8 ci_stack
[OCFS2_STACK_LABEL_LEN
];
551 /*08*/ __u8 ci_cluster
[OCFS2_CLUSTER_NAME_LEN
];
556 * On disk superblock for OCFS2
557 * Note that it is contained inside an ocfs2_dinode, so all offsets
558 * are relative to the start of ocfs2_dinode.id2.
560 struct ocfs2_super_block
{
561 /*00*/ __le16 s_major_rev_level
;
562 __le16 s_minor_rev_level
;
564 __le16 s_max_mnt_count
;
565 __le16 s_state
; /* File system state */
566 __le16 s_errors
; /* Behaviour when detecting errors */
567 __le32 s_checkinterval
; /* Max time between checks */
568 /*10*/ __le64 s_lastcheck
; /* Time of last check */
569 __le32 s_creator_os
; /* OS */
570 __le32 s_feature_compat
; /* Compatible feature set */
571 /*20*/ __le32 s_feature_incompat
; /* Incompatible feature set */
572 __le32 s_feature_ro_compat
; /* Readonly-compatible feature set */
573 __le64 s_root_blkno
; /* Offset, in blocks, of root directory
575 /*30*/ __le64 s_system_dir_blkno
; /* Offset, in blocks, of system
577 __le32 s_blocksize_bits
; /* Blocksize for this fs */
578 __le32 s_clustersize_bits
; /* Clustersize for this fs */
579 /*40*/ __le16 s_max_slots
; /* Max number of simultaneous mounts
580 before tunefs required */
581 __le16 s_tunefs_flag
;
582 __le32 s_uuid_hash
; /* hash value of uuid */
583 __le64 s_first_cluster_group
; /* Block offset of 1st cluster
585 /*50*/ __u8 s_label
[OCFS2_MAX_VOL_LABEL_LEN
]; /* Label for mounting, etc. */
586 /*90*/ __u8 s_uuid
[OCFS2_VOL_UUID_LEN
]; /* 128-bit uuid */
587 /*A0*/ struct ocfs2_cluster_info s_cluster_info
; /* Selected userspace
589 with INCOMPAT flag. */
590 /*B8*/ __le16 s_xattr_inline_size
; /* extended attribute inline size
594 /*C0*/ __le64 s_reserved2
[16]; /* Fill out superblock */
598 * NOTE: As stated above, all offsets are relative to
599 * ocfs2_dinode.id2, which is at 0xC0 in the inode.
600 * 0xC0 + 0x140 = 0x200 or 512 bytes. A superblock must fit within
601 * our smallest blocksize, which is 512 bytes. To ensure this,
602 * we reserve the space in s_reserved2. Anything past s_reserved2
603 * will not be available on the smallest blocksize.
608 * Local allocation bitmap for OCFS2 slots
609 * Note that it exists inside an ocfs2_dinode, so all offsets are
610 * relative to the start of ocfs2_dinode.id2.
612 struct ocfs2_local_alloc
614 /*00*/ __le32 la_bm_off
; /* Starting bit offset in main bitmap */
615 __le16 la_size
; /* Size of included bitmap, in bytes */
618 /*10*/ __u8 la_bitmap
[0];
622 * Data-in-inode header. This is only used if i_dyn_features has
623 * OCFS2_INLINE_DATA_FL set.
625 struct ocfs2_inline_data
627 /*00*/ __le16 id_count
; /* Number of bytes that can be used
628 * for data, starting at id_data */
631 __u8 id_data
[0]; /* Start of user data */
635 * On disk inode for OCFS2
637 struct ocfs2_dinode
{
638 /*00*/ __u8 i_signature
[8]; /* Signature for validation */
639 __le32 i_generation
; /* Generation number */
640 __le16 i_suballoc_slot
; /* Slot suballocator this inode
642 __le16 i_suballoc_bit
; /* Bit offset in suballocator
644 /*10*/ __le16 i_reserved0
;
645 __le16 i_xattr_inline_size
;
646 __le32 i_clusters
; /* Cluster count */
647 __le32 i_uid
; /* Owner UID */
648 __le32 i_gid
; /* Owning GID */
649 /*20*/ __le64 i_size
; /* Size in bytes */
650 __le16 i_mode
; /* File mode */
651 __le16 i_links_count
; /* Links count */
652 __le32 i_flags
; /* File flags */
653 /*30*/ __le64 i_atime
; /* Access time */
654 __le64 i_ctime
; /* Creation time */
655 /*40*/ __le64 i_mtime
; /* Modification time */
656 __le64 i_dtime
; /* Deletion time */
657 /*50*/ __le64 i_blkno
; /* Offset on disk, in blocks */
658 __le64 i_last_eb_blk
; /* Pointer to last extent
660 /*60*/ __le32 i_fs_generation
; /* Generation per fs-instance */
664 /*70*/ __le32 i_attr
;
665 __le16 i_orphaned_slot
; /* Only valid when OCFS2_ORPHANED_FL
666 was set in i_flags */
667 __le16 i_dyn_features
;
669 /*80*/ __le64 i_reserved2
[7];
671 __le64 i_pad1
; /* Generic way to refer to this
674 __le64 i_rdev
; /* Device number */
676 struct { /* Info for bitmap system
678 __le32 i_used
; /* Bits (ie, clusters) used */
679 __le32 i_total
; /* Total bits (clusters)
682 struct { /* Info for journal system
684 __le32 ij_flags
; /* Mounted, version, etc. */
685 __le32 ij_recovery_generation
; /* Incremented when the
690 } id1
; /* Inode type dependant 1 */
692 struct ocfs2_super_block i_super
;
693 struct ocfs2_local_alloc i_lab
;
694 struct ocfs2_chain_list i_chain
;
695 struct ocfs2_extent_list i_list
;
696 struct ocfs2_truncate_log i_dealloc
;
697 struct ocfs2_inline_data i_data
;
700 /* Actual on-disk size is one block */
704 * On-disk directory entry structure for OCFS2
706 * Packed as this structure could be accessed unaligned on 64-bit platforms
708 struct ocfs2_dir_entry
{
709 /*00*/ __le64 inode
; /* Inode number */
710 __le16 rec_len
; /* Directory entry length */
711 __u8 name_len
; /* Name length */
713 /*0C*/ char name
[OCFS2_MAX_FILENAME_LEN
]; /* File name */
714 /* Actual on-disk length specified by rec_len */
715 } __attribute__ ((packed
));
718 * On disk allocator group structure for OCFS2
720 struct ocfs2_group_desc
722 /*00*/ __u8 bg_signature
[8]; /* Signature for validation */
723 __le16 bg_size
; /* Size of included bitmap in
725 __le16 bg_bits
; /* Bits represented by this
727 __le16 bg_free_bits_count
; /* Free bits count */
728 __le16 bg_chain
; /* What chain I am in. */
729 /*10*/ __le32 bg_generation
;
731 __le64 bg_next_group
; /* Next group in my list, in
733 /*20*/ __le64 bg_parent_dinode
; /* dinode which owns me, in
735 __le64 bg_blkno
; /* Offset on disk, in blocks */
736 /*30*/ __le64 bg_reserved2
[2];
737 /*40*/ __u8 bg_bitmap
[0];
741 * On disk extended attribute structure for OCFS2.
745 * ocfs2_xattr_entry indicates one extend attribute.
747 * Note that it can be stored in inode, one block or one xattr bucket.
749 struct ocfs2_xattr_entry
{
750 __le32 xe_name_hash
; /* hash value of xattr prefix+suffix. */
751 __le16 xe_name_offset
; /* byte offset from the 1st entry in the
752 local xattr storage(inode, xattr block or
754 __u8 xe_name_len
; /* xattr name len, does't include prefix. */
755 __u8 xe_type
; /* the low 7 bits indicate the name prefix
756 * type and the highest bit indicates whether
757 * the EA is stored in the local storage. */
758 __le64 xe_value_size
; /* real xattr value length. */
762 * On disk structure for xattr header.
764 * One ocfs2_xattr_header describes how many ocfs2_xattr_entry records in
765 * the local xattr storage.
767 struct ocfs2_xattr_header
{
768 __le16 xh_count
; /* contains the count of how
769 many records are in the
770 local xattr storage. */
771 __le16 xh_free_start
; /* current offset for storing
773 __le16 xh_name_value_len
; /* total length of name/value
774 length in this bucket. */
775 __le16 xh_num_buckets
; /* Number of xattr buckets
776 in this extent record,
777 only valid in the first
780 struct ocfs2_xattr_entry xh_entries
[0]; /* xattr entry list. */
784 * On disk structure for xattr value root.
786 * When an xattr's value is large enough, it is stored in an external
787 * b-tree like file data. The xattr value root points to this structure.
789 struct ocfs2_xattr_value_root
{
790 /*00*/ __le32 xr_clusters
; /* clusters covered by xattr value. */
792 __le64 xr_last_eb_blk
; /* Pointer to last extent block */
793 /*10*/ struct ocfs2_extent_list xr_list
; /* Extent record list */
797 * On disk structure for xattr tree root.
799 * It is used when there are too many extended attributes for one file. These
800 * attributes will be organized and stored in an indexed-btree.
802 struct ocfs2_xattr_tree_root
{
803 /*00*/ __le32 xt_clusters
; /* clusters covered by xattr. */
805 __le64 xt_last_eb_blk
; /* Pointer to last extent block */
806 /*10*/ struct ocfs2_extent_list xt_list
; /* Extent record list */
809 #define OCFS2_XATTR_INDEXED 0x1
810 #define OCFS2_HASH_SHIFT 5
811 #define OCFS2_XATTR_ROUND 3
812 #define OCFS2_XATTR_SIZE(size) (((size) + OCFS2_XATTR_ROUND) & \
813 ~(OCFS2_XATTR_ROUND))
815 #define OCFS2_XATTR_BUCKET_SIZE 4096
816 #define OCFS2_XATTR_MAX_BLOCKS_PER_BUCKET (OCFS2_XATTR_BUCKET_SIZE \
817 / OCFS2_MIN_BLOCKSIZE)
820 * On disk structure for xattr block.
822 struct ocfs2_xattr_block
{
823 /*00*/ __u8 xb_signature
[8]; /* Signature for verification */
824 __le16 xb_suballoc_slot
; /* Slot suballocator this
826 __le16 xb_suballoc_bit
; /* Bit offset in suballocator
828 __le32 xb_fs_generation
; /* Must match super block */
829 /*10*/ __le64 xb_blkno
; /* Offset on disk, in blocks */
831 /*20*/ __le16 xb_flags
; /* Indicates whether this block contains
832 real xattr or a xattr tree. */
837 struct ocfs2_xattr_header xb_header
; /* xattr header if this
838 block contains xattr */
839 struct ocfs2_xattr_tree_root xb_root
;/* xattr tree root if this
845 #define OCFS2_XATTR_ENTRY_LOCAL 0x80
846 #define OCFS2_XATTR_TYPE_MASK 0x7F
847 static inline void ocfs2_xattr_set_local(struct ocfs2_xattr_entry
*xe
,
851 xe
->xe_type
|= OCFS2_XATTR_ENTRY_LOCAL
;
853 xe
->xe_type
&= ~OCFS2_XATTR_ENTRY_LOCAL
;
856 static inline int ocfs2_xattr_is_local(struct ocfs2_xattr_entry
*xe
)
858 return xe
->xe_type
& OCFS2_XATTR_ENTRY_LOCAL
;
861 static inline void ocfs2_xattr_set_type(struct ocfs2_xattr_entry
*xe
, int type
)
863 xe
->xe_type
|= type
& OCFS2_XATTR_TYPE_MASK
;
866 static inline int ocfs2_xattr_get_type(struct ocfs2_xattr_entry
*xe
)
868 return xe
->xe_type
& OCFS2_XATTR_TYPE_MASK
;
872 static inline int ocfs2_fast_symlink_chars(struct super_block
*sb
)
874 return sb
->s_blocksize
-
875 offsetof(struct ocfs2_dinode
, id2
.i_symlink
);
878 static inline int ocfs2_max_inline_data(struct super_block
*sb
)
880 return sb
->s_blocksize
-
881 offsetof(struct ocfs2_dinode
, id2
.i_data
.id_data
);
884 static inline int ocfs2_max_inline_data_with_xattr(struct super_block
*sb
,
885 struct ocfs2_dinode
*di
)
887 unsigned int xattrsize
= le16_to_cpu(di
->i_xattr_inline_size
);
889 if (le16_to_cpu(di
->i_dyn_features
) & OCFS2_INLINE_XATTR_FL
)
890 return sb
->s_blocksize
-
891 offsetof(struct ocfs2_dinode
, id2
.i_data
.id_data
) -
894 return sb
->s_blocksize
-
895 offsetof(struct ocfs2_dinode
, id2
.i_data
.id_data
);
898 static inline int ocfs2_extent_recs_per_inode(struct super_block
*sb
)
902 size
= sb
->s_blocksize
-
903 offsetof(struct ocfs2_dinode
, id2
.i_list
.l_recs
);
905 return size
/ sizeof(struct ocfs2_extent_rec
);
908 static inline int ocfs2_extent_recs_per_inode_with_xattr(
909 struct super_block
*sb
,
910 struct ocfs2_dinode
*di
)
913 unsigned int xattrsize
= le16_to_cpu(di
->i_xattr_inline_size
);
915 if (le16_to_cpu(di
->i_dyn_features
) & OCFS2_INLINE_XATTR_FL
)
916 size
= sb
->s_blocksize
-
917 offsetof(struct ocfs2_dinode
, id2
.i_list
.l_recs
) -
920 size
= sb
->s_blocksize
-
921 offsetof(struct ocfs2_dinode
, id2
.i_list
.l_recs
);
923 return size
/ sizeof(struct ocfs2_extent_rec
);
926 static inline int ocfs2_chain_recs_per_inode(struct super_block
*sb
)
930 size
= sb
->s_blocksize
-
931 offsetof(struct ocfs2_dinode
, id2
.i_chain
.cl_recs
);
933 return size
/ sizeof(struct ocfs2_chain_rec
);
936 static inline u16
ocfs2_extent_recs_per_eb(struct super_block
*sb
)
940 size
= sb
->s_blocksize
-
941 offsetof(struct ocfs2_extent_block
, h_list
.l_recs
);
943 return size
/ sizeof(struct ocfs2_extent_rec
);
946 static inline u16
ocfs2_local_alloc_size(struct super_block
*sb
)
950 size
= sb
->s_blocksize
-
951 offsetof(struct ocfs2_dinode
, id2
.i_lab
.la_bitmap
);
956 static inline int ocfs2_group_bitmap_size(struct super_block
*sb
)
960 size
= sb
->s_blocksize
-
961 offsetof(struct ocfs2_group_desc
, bg_bitmap
);
966 static inline int ocfs2_truncate_recs_per_inode(struct super_block
*sb
)
970 size
= sb
->s_blocksize
-
971 offsetof(struct ocfs2_dinode
, id2
.i_dealloc
.tl_recs
);
973 return size
/ sizeof(struct ocfs2_truncate_rec
);
976 static inline u64
ocfs2_backup_super_blkno(struct super_block
*sb
, int index
)
978 u64 offset
= OCFS2_BACKUP_SB_START
;
980 if (index
>= 0 && index
< OCFS2_MAX_BACKUP_SUPERBLOCKS
) {
981 offset
<<= (2 * index
);
982 offset
>>= sb
->s_blocksize_bits
;
990 static inline u16
ocfs2_xattr_recs_per_xb(struct super_block
*sb
)
994 size
= sb
->s_blocksize
-
995 offsetof(struct ocfs2_xattr_block
,
996 xb_attrs
.xb_root
.xt_list
.l_recs
);
998 return size
/ sizeof(struct ocfs2_extent_rec
);
1001 static inline int ocfs2_fast_symlink_chars(int blocksize
)
1003 return blocksize
- offsetof(struct ocfs2_dinode
, id2
.i_symlink
);
1006 static inline int ocfs2_max_inline_data(int blocksize
)
1008 return blocksize
- offsetof(struct ocfs2_dinode
, id2
.i_data
.id_data
);
1011 static inline int ocfs2_extent_recs_per_inode(int blocksize
)
1016 offsetof(struct ocfs2_dinode
, id2
.i_list
.l_recs
);
1018 return size
/ sizeof(struct ocfs2_extent_rec
);
1021 static inline int ocfs2_chain_recs_per_inode(int blocksize
)
1026 offsetof(struct ocfs2_dinode
, id2
.i_chain
.cl_recs
);
1028 return size
/ sizeof(struct ocfs2_chain_rec
);
1031 static inline int ocfs2_extent_recs_per_eb(int blocksize
)
1036 offsetof(struct ocfs2_extent_block
, h_list
.l_recs
);
1038 return size
/ sizeof(struct ocfs2_extent_rec
);
1041 static inline int ocfs2_local_alloc_size(int blocksize
)
1046 offsetof(struct ocfs2_dinode
, id2
.i_lab
.la_bitmap
);
1051 static inline int ocfs2_group_bitmap_size(int blocksize
)
1056 offsetof(struct ocfs2_group_desc
, bg_bitmap
);
1061 static inline int ocfs2_truncate_recs_per_inode(int blocksize
)
1066 offsetof(struct ocfs2_dinode
, id2
.i_dealloc
.tl_recs
);
1068 return size
/ sizeof(struct ocfs2_truncate_rec
);
1071 static inline uint64_t ocfs2_backup_super_blkno(int blocksize
, int index
)
1073 uint64_t offset
= OCFS2_BACKUP_SB_START
;
1075 if (index
>= 0 && index
< OCFS2_MAX_BACKUP_SUPERBLOCKS
) {
1076 offset
<<= (2 * index
);
1077 offset
/= blocksize
;
1084 static inline int ocfs2_xattr_recs_per_xb(int blocksize
)
1089 offsetof(struct ocfs2_xattr_block
,
1090 xb_attrs
.xb_root
.xt_list
.l_recs
);
1092 return size
/ sizeof(struct ocfs2_extent_rec
);
1094 #endif /* __KERNEL__ */
1097 static inline int ocfs2_system_inode_is_global(int type
)
1099 return ((type
>= 0) &&
1100 (type
<= OCFS2_LAST_GLOBAL_SYSTEM_INODE
));
1103 static inline int ocfs2_sprintf_system_inode_name(char *buf
, int len
,
1109 * Global system inodes can only have one copy. Everything
1110 * after OCFS2_LAST_GLOBAL_SYSTEM_INODE in the system inode
1111 * list has a copy per slot.
1113 if (type
<= OCFS2_LAST_GLOBAL_SYSTEM_INODE
)
1114 chars
= snprintf(buf
, len
, "%s",
1115 ocfs2_system_inodes
[type
].si_name
);
1117 chars
= snprintf(buf
, len
,
1118 ocfs2_system_inodes
[type
].si_name
,
1124 static inline void ocfs2_set_de_type(struct ocfs2_dir_entry
*de
,
1127 de
->file_type
= ocfs2_type_by_mode
[(mode
& S_IFMT
)>>S_SHIFT
];
1130 #endif /* _OCFS2_FS_H */