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"
68 /* Compatibility flags */
69 #define OCFS2_HAS_COMPAT_FEATURE(sb,mask) \
70 ( OCFS2_SB(sb)->s_feature_compat & (mask) )
71 #define OCFS2_HAS_RO_COMPAT_FEATURE(sb,mask) \
72 ( OCFS2_SB(sb)->s_feature_ro_compat & (mask) )
73 #define OCFS2_HAS_INCOMPAT_FEATURE(sb,mask) \
74 ( OCFS2_SB(sb)->s_feature_incompat & (mask) )
75 #define OCFS2_SET_COMPAT_FEATURE(sb,mask) \
76 OCFS2_SB(sb)->s_feature_compat |= (mask)
77 #define OCFS2_SET_RO_COMPAT_FEATURE(sb,mask) \
78 OCFS2_SB(sb)->s_feature_ro_compat |= (mask)
79 #define OCFS2_SET_INCOMPAT_FEATURE(sb,mask) \
80 OCFS2_SB(sb)->s_feature_incompat |= (mask)
81 #define OCFS2_CLEAR_COMPAT_FEATURE(sb,mask) \
82 OCFS2_SB(sb)->s_feature_compat &= ~(mask)
83 #define OCFS2_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
84 OCFS2_SB(sb)->s_feature_ro_compat &= ~(mask)
85 #define OCFS2_CLEAR_INCOMPAT_FEATURE(sb,mask) \
86 OCFS2_SB(sb)->s_feature_incompat &= ~(mask)
88 #define OCFS2_FEATURE_COMPAT_SUPP 0
89 #define OCFS2_FEATURE_INCOMPAT_SUPP 0
90 #define OCFS2_FEATURE_RO_COMPAT_SUPP 0
93 * Heartbeat-only devices are missing journals and other files. The
94 * filesystem driver can't load them, but the library can. Never put
95 * this in OCFS2_FEATURE_INCOMPAT_SUPP, *ever*.
97 #define OCFS2_FEATURE_INCOMPAT_HEARTBEAT_DEV 0x0002
101 * Flags on ocfs2_dinode.i_flags
103 #define OCFS2_VALID_FL (0x00000001) /* Inode is valid */
104 #define OCFS2_UNUSED2_FL (0x00000002)
105 #define OCFS2_ORPHANED_FL (0x00000004) /* On the orphan list */
106 #define OCFS2_UNUSED3_FL (0x00000008)
107 /* System inode flags */
108 #define OCFS2_SYSTEM_FL (0x00000010) /* System inode */
109 #define OCFS2_SUPER_BLOCK_FL (0x00000020) /* Super block */
110 #define OCFS2_LOCAL_ALLOC_FL (0x00000040) /* Slot local alloc bitmap */
111 #define OCFS2_BITMAP_FL (0x00000080) /* Allocation bitmap */
112 #define OCFS2_JOURNAL_FL (0x00000100) /* Slot local journal */
113 #define OCFS2_HEARTBEAT_FL (0x00000200) /* Heartbeat area */
114 #define OCFS2_CHAIN_FL (0x00000400) /* Chain allocator */
115 #define OCFS2_DEALLOC_FL (0x00000800) /* Truncate log */
118 * Journal Flags (ocfs2_dinode.id1.journal1.i_flags)
120 #define OCFS2_JOURNAL_DIRTY_FL (0x00000001) /* Journal needs recovery */
123 * superblock s_state flags
125 #define OCFS2_ERROR_FS (0x00000001) /* FS saw errors */
127 /* Limit of space in ocfs2_dir_entry */
128 #define OCFS2_MAX_FILENAME_LEN 255
130 /* Maximum slots on an ocfs2 file system */
131 #define OCFS2_MAX_SLOTS 255
133 /* Slot map indicator for an empty slot */
134 #define OCFS2_INVALID_SLOT -1
136 #define OCFS2_VOL_UUID_LEN 16
137 #define OCFS2_MAX_VOL_LABEL_LEN 64
139 /* Journal limits (in bytes) */
140 #define OCFS2_MIN_JOURNAL_SIZE (4 * 1024 * 1024)
142 struct ocfs2_system_inode_info
{
148 /* System file index */
150 BAD_BLOCK_SYSTEM_INODE
= 0,
151 GLOBAL_INODE_ALLOC_SYSTEM_INODE
,
152 SLOT_MAP_SYSTEM_INODE
,
153 #define OCFS2_FIRST_ONLINE_SYSTEM_INODE SLOT_MAP_SYSTEM_INODE
154 HEARTBEAT_SYSTEM_INODE
,
155 GLOBAL_BITMAP_SYSTEM_INODE
,
156 #define OCFS2_LAST_GLOBAL_SYSTEM_INODE GLOBAL_BITMAP_SYSTEM_INODE
157 ORPHAN_DIR_SYSTEM_INODE
,
158 EXTENT_ALLOC_SYSTEM_INODE
,
159 INODE_ALLOC_SYSTEM_INODE
,
160 JOURNAL_SYSTEM_INODE
,
161 LOCAL_ALLOC_SYSTEM_INODE
,
162 TRUNCATE_LOG_SYSTEM_INODE
,
166 static struct ocfs2_system_inode_info ocfs2_system_inodes
[NUM_SYSTEM_INODES
] = {
167 /* Global system inodes (single copy) */
168 /* The first two are only used from userspace mfks/tunefs */
169 [BAD_BLOCK_SYSTEM_INODE
] = { "bad_blocks", 0, S_IFREG
| 0644 },
170 [GLOBAL_INODE_ALLOC_SYSTEM_INODE
] = { "global_inode_alloc", OCFS2_BITMAP_FL
| OCFS2_CHAIN_FL
, S_IFREG
| 0644 },
172 /* These are used by the running filesystem */
173 [SLOT_MAP_SYSTEM_INODE
] = { "slot_map", 0, S_IFREG
| 0644 },
174 [HEARTBEAT_SYSTEM_INODE
] = { "heartbeat", OCFS2_HEARTBEAT_FL
, S_IFREG
| 0644 },
175 [GLOBAL_BITMAP_SYSTEM_INODE
] = { "global_bitmap", 0, S_IFREG
| 0644 },
177 /* Slot-specific system inodes (one copy per slot) */
178 [ORPHAN_DIR_SYSTEM_INODE
] = { "orphan_dir:%04d", 0, S_IFDIR
| 0755 },
179 [EXTENT_ALLOC_SYSTEM_INODE
] = { "extent_alloc:%04d", OCFS2_BITMAP_FL
| OCFS2_CHAIN_FL
, S_IFREG
| 0644 },
180 [INODE_ALLOC_SYSTEM_INODE
] = { "inode_alloc:%04d", OCFS2_BITMAP_FL
| OCFS2_CHAIN_FL
, S_IFREG
| 0644 },
181 [JOURNAL_SYSTEM_INODE
] = { "journal:%04d", OCFS2_JOURNAL_FL
, S_IFREG
| 0644 },
182 [LOCAL_ALLOC_SYSTEM_INODE
] = { "local_alloc:%04d", OCFS2_BITMAP_FL
| OCFS2_LOCAL_ALLOC_FL
, S_IFREG
| 0644 },
183 [TRUNCATE_LOG_SYSTEM_INODE
] = { "truncate_log:%04d", OCFS2_DEALLOC_FL
, S_IFREG
| 0644 }
186 /* Parameter passed from mount.ocfs2 to module */
187 #define OCFS2_HB_NONE "heartbeat=none"
188 #define OCFS2_HB_LOCAL "heartbeat=local"
191 * OCFS2 directory file types. Only the low 3 bits are used. The
192 * other bits are reserved for now.
194 #define OCFS2_FT_UNKNOWN 0
195 #define OCFS2_FT_REG_FILE 1
196 #define OCFS2_FT_DIR 2
197 #define OCFS2_FT_CHRDEV 3
198 #define OCFS2_FT_BLKDEV 4
199 #define OCFS2_FT_FIFO 5
200 #define OCFS2_FT_SOCK 6
201 #define OCFS2_FT_SYMLINK 7
203 #define OCFS2_FT_MAX 8
206 * OCFS2_DIR_PAD defines the directory entries boundaries
208 * NOTE: It must be a multiple of 4
210 #define OCFS2_DIR_PAD 4
211 #define OCFS2_DIR_ROUND (OCFS2_DIR_PAD - 1)
212 #define OCFS2_DIR_MEMBER_LEN offsetof(struct ocfs2_dir_entry, name)
213 #define OCFS2_DIR_REC_LEN(name_len) (((name_len) + OCFS2_DIR_MEMBER_LEN + \
217 #define OCFS2_LINK_MAX 32000
220 static unsigned char ocfs2_type_by_mode
[S_IFMT
>> S_SHIFT
] = {
221 [S_IFREG
>> S_SHIFT
] = OCFS2_FT_REG_FILE
,
222 [S_IFDIR
>> S_SHIFT
] = OCFS2_FT_DIR
,
223 [S_IFCHR
>> S_SHIFT
] = OCFS2_FT_CHRDEV
,
224 [S_IFBLK
>> S_SHIFT
] = OCFS2_FT_BLKDEV
,
225 [S_IFIFO
>> S_SHIFT
] = OCFS2_FT_FIFO
,
226 [S_IFSOCK
>> S_SHIFT
] = OCFS2_FT_SOCK
,
227 [S_IFLNK
>> S_SHIFT
] = OCFS2_FT_SYMLINK
,
234 #define OCFS2_RAW_SB(dinode) (&((dinode)->id2.i_super))
237 * On disk extent record for OCFS2
238 * It describes a range of clusters on disk.
240 struct ocfs2_extent_rec
{
241 /*00*/ __le32 e_cpos
; /* Offset into the file, in clusters */
242 __le32 e_clusters
; /* Clusters covered by this extent */
243 __le64 e_blkno
; /* Physical disk offset, in blocks */
247 struct ocfs2_chain_rec
{
248 __le32 c_free
; /* Number of free bits in this chain. */
249 __le32 c_total
; /* Number of total bits in this chain */
250 __le64 c_blkno
; /* Physical disk offset (blocks) of 1st group */
253 struct ocfs2_truncate_rec
{
254 __le32 t_start
; /* 1st cluster in this log */
255 __le32 t_clusters
; /* Number of total clusters covered */
259 * On disk extent list for OCFS2 (node in the tree). Note that this
260 * is contained inside ocfs2_dinode or ocfs2_extent_block, so the
261 * offsets are relative to ocfs2_dinode.id2.i_list or
262 * ocfs2_extent_block.h_list, respectively.
264 struct ocfs2_extent_list
{
265 /*00*/ __le16 l_tree_depth
; /* Extent tree depth from this
266 point. 0 means data extents
267 hang directly off this
269 __le16 l_count
; /* Number of extent records */
270 __le16 l_next_free_rec
; /* Next unused extent slot */
272 __le64 l_reserved2
; /* Pad to
273 sizeof(ocfs2_extent_rec) */
274 /*10*/ struct ocfs2_extent_rec l_recs
[0]; /* Extent records */
278 * On disk allocation chain list for OCFS2. Note that this is
279 * contained inside ocfs2_dinode, so the offsets are relative to
280 * ocfs2_dinode.id2.i_chain.
282 struct ocfs2_chain_list
{
283 /*00*/ __le16 cl_cpg
; /* Clusters per Block Group */
284 __le16 cl_bpc
; /* Bits per cluster */
285 __le16 cl_count
; /* Total chains in this list */
286 __le16 cl_next_free_rec
; /* Next unused chain slot */
288 /*10*/ struct ocfs2_chain_rec cl_recs
[0]; /* Chain records */
292 * On disk deallocation log for OCFS2. Note that this is
293 * contained inside ocfs2_dinode, so the offsets are relative to
294 * ocfs2_dinode.id2.i_dealloc.
296 struct ocfs2_truncate_log
{
297 /*00*/ __le16 tl_count
; /* Total records in this log */
298 __le16 tl_used
; /* Number of records in use */
300 /*08*/ struct ocfs2_truncate_rec tl_recs
[0]; /* Truncate records */
304 * On disk extent block (indirect block) for OCFS2
306 struct ocfs2_extent_block
308 /*00*/ __u8 h_signature
[8]; /* Signature for verification */
310 /*10*/ __le16 h_suballoc_slot
; /* Slot suballocator this
311 extent_header belongs to */
312 __le16 h_suballoc_bit
; /* Bit offset in suballocator
314 __le32 h_fs_generation
; /* Must match super block */
315 __le64 h_blkno
; /* Offset on disk, in blocks */
316 /*20*/ __le64 h_reserved3
;
317 __le64 h_next_leaf_blk
; /* Offset on disk, in blocks,
318 of next leaf header pointing
320 /*30*/ struct ocfs2_extent_list h_list
; /* Extent record list */
321 /* Actual on-disk size is one block */
325 * On disk superblock for OCFS2
326 * Note that it is contained inside an ocfs2_dinode, so all offsets
327 * are relative to the start of ocfs2_dinode.id2.
329 struct ocfs2_super_block
{
330 /*00*/ __le16 s_major_rev_level
;
331 __le16 s_minor_rev_level
;
333 __le16 s_max_mnt_count
;
334 __le16 s_state
; /* File system state */
335 __le16 s_errors
; /* Behaviour when detecting errors */
336 __le32 s_checkinterval
; /* Max time between checks */
337 /*10*/ __le64 s_lastcheck
; /* Time of last check */
338 __le32 s_creator_os
; /* OS */
339 __le32 s_feature_compat
; /* Compatible feature set */
340 /*20*/ __le32 s_feature_incompat
; /* Incompatible feature set */
341 __le32 s_feature_ro_compat
; /* Readonly-compatible feature set */
342 __le64 s_root_blkno
; /* Offset, in blocks, of root directory
344 /*30*/ __le64 s_system_dir_blkno
; /* Offset, in blocks, of system
346 __le32 s_blocksize_bits
; /* Blocksize for this fs */
347 __le32 s_clustersize_bits
; /* Clustersize for this fs */
348 /*40*/ __le16 s_max_slots
; /* Max number of simultaneous mounts
349 before tunefs required */
352 __le64 s_first_cluster_group
; /* Block offset of 1st cluster
354 /*50*/ __u8 s_label
[OCFS2_MAX_VOL_LABEL_LEN
]; /* Label for mounting, etc. */
355 /*90*/ __u8 s_uuid
[OCFS2_VOL_UUID_LEN
]; /* 128-bit uuid */
360 * Local allocation bitmap for OCFS2 slots
361 * Note that it exists inside an ocfs2_dinode, so all offsets are
362 * relative to the start of ocfs2_dinode.id2.
364 struct ocfs2_local_alloc
366 /*00*/ __le32 la_bm_off
; /* Starting bit offset in main bitmap */
367 __le16 la_size
; /* Size of included bitmap, in bytes */
370 /*10*/ __u8 la_bitmap
[0];
374 * On disk inode for OCFS2
376 struct ocfs2_dinode
{
377 /*00*/ __u8 i_signature
[8]; /* Signature for validation */
378 __le32 i_generation
; /* Generation number */
379 __le16 i_suballoc_slot
; /* Slot suballocator this inode
381 __le16 i_suballoc_bit
; /* Bit offset in suballocator
383 /*10*/ __le32 i_reserved0
;
384 __le32 i_clusters
; /* Cluster count */
385 __le32 i_uid
; /* Owner UID */
386 __le32 i_gid
; /* Owning GID */
387 /*20*/ __le64 i_size
; /* Size in bytes */
388 __le16 i_mode
; /* File mode */
389 __le16 i_links_count
; /* Links count */
390 __le32 i_flags
; /* File flags */
391 /*30*/ __le64 i_atime
; /* Access time */
392 __le64 i_ctime
; /* Creation time */
393 /*40*/ __le64 i_mtime
; /* Modification time */
394 __le64 i_dtime
; /* Deletion time */
395 /*50*/ __le64 i_blkno
; /* Offset on disk, in blocks */
396 __le64 i_last_eb_blk
; /* Pointer to last extent
398 /*60*/ __le32 i_fs_generation
; /* Generation per fs-instance */
402 /*70*/ __le64 i_reserved1
[9];
404 __le64 i_pad1
; /* Generic way to refer to this
407 __le64 i_rdev
; /* Device number */
409 struct { /* Info for bitmap system
411 __le32 i_used
; /* Bits (ie, clusters) used */
412 __le32 i_total
; /* Total bits (clusters)
415 struct { /* Info for journal system
417 __le32 ij_flags
; /* Mounted, version, etc. */
420 } id1
; /* Inode type dependant 1 */
422 struct ocfs2_super_block i_super
;
423 struct ocfs2_local_alloc i_lab
;
424 struct ocfs2_chain_list i_chain
;
425 struct ocfs2_extent_list i_list
;
426 struct ocfs2_truncate_log i_dealloc
;
429 /* Actual on-disk size is one block */
433 * On-disk directory entry structure for OCFS2
435 * Packed as this structure could be accessed unaligned on 64-bit platforms
437 struct ocfs2_dir_entry
{
438 /*00*/ __le64 inode
; /* Inode number */
439 __le16 rec_len
; /* Directory entry length */
440 __u8 name_len
; /* Name length */
442 /*0C*/ char name
[OCFS2_MAX_FILENAME_LEN
]; /* File name */
443 /* Actual on-disk length specified by rec_len */
444 } __attribute__ ((packed
));
447 * On disk allocator group structure for OCFS2
449 struct ocfs2_group_desc
451 /*00*/ __u8 bg_signature
[8]; /* Signature for validation */
452 __le16 bg_size
; /* Size of included bitmap in
454 __le16 bg_bits
; /* Bits represented by this
456 __le16 bg_free_bits_count
; /* Free bits count */
457 __le16 bg_chain
; /* What chain I am in. */
458 /*10*/ __le32 bg_generation
;
460 __le64 bg_next_group
; /* Next group in my list, in
462 /*20*/ __le64 bg_parent_dinode
; /* dinode which owns me, in
464 __le64 bg_blkno
; /* Offset on disk, in blocks */
465 /*30*/ __le64 bg_reserved2
[2];
466 /*40*/ __u8 bg_bitmap
[0];
470 static inline int ocfs2_fast_symlink_chars(struct super_block
*sb
)
472 return sb
->s_blocksize
-
473 offsetof(struct ocfs2_dinode
, id2
.i_symlink
);
476 static inline int ocfs2_extent_recs_per_inode(struct super_block
*sb
)
480 size
= sb
->s_blocksize
-
481 offsetof(struct ocfs2_dinode
, id2
.i_list
.l_recs
);
483 return size
/ sizeof(struct ocfs2_extent_rec
);
486 static inline int ocfs2_chain_recs_per_inode(struct super_block
*sb
)
490 size
= sb
->s_blocksize
-
491 offsetof(struct ocfs2_dinode
, id2
.i_chain
.cl_recs
);
493 return size
/ sizeof(struct ocfs2_chain_rec
);
496 static inline u16
ocfs2_extent_recs_per_eb(struct super_block
*sb
)
500 size
= sb
->s_blocksize
-
501 offsetof(struct ocfs2_extent_block
, h_list
.l_recs
);
503 return size
/ sizeof(struct ocfs2_extent_rec
);
506 static inline u16
ocfs2_local_alloc_size(struct super_block
*sb
)
510 size
= sb
->s_blocksize
-
511 offsetof(struct ocfs2_dinode
, id2
.i_lab
.la_bitmap
);
516 static inline int ocfs2_group_bitmap_size(struct super_block
*sb
)
520 size
= sb
->s_blocksize
-
521 offsetof(struct ocfs2_group_desc
, bg_bitmap
);
526 static inline int ocfs2_truncate_recs_per_inode(struct super_block
*sb
)
530 size
= sb
->s_blocksize
-
531 offsetof(struct ocfs2_dinode
, id2
.i_dealloc
.tl_recs
);
533 return size
/ sizeof(struct ocfs2_truncate_rec
);
536 static inline int ocfs2_fast_symlink_chars(int blocksize
)
538 return blocksize
- offsetof(struct ocfs2_dinode
, id2
.i_symlink
);
541 static inline int ocfs2_extent_recs_per_inode(int blocksize
)
546 offsetof(struct ocfs2_dinode
, id2
.i_list
.l_recs
);
548 return size
/ sizeof(struct ocfs2_extent_rec
);
551 static inline int ocfs2_chain_recs_per_inode(int blocksize
)
556 offsetof(struct ocfs2_dinode
, id2
.i_chain
.cl_recs
);
558 return size
/ sizeof(struct ocfs2_chain_rec
);
561 static inline int ocfs2_extent_recs_per_eb(int blocksize
)
566 offsetof(struct ocfs2_extent_block
, h_list
.l_recs
);
568 return size
/ sizeof(struct ocfs2_extent_rec
);
571 static inline int ocfs2_local_alloc_size(int blocksize
)
576 offsetof(struct ocfs2_dinode
, id2
.i_lab
.la_bitmap
);
581 static inline int ocfs2_group_bitmap_size(int blocksize
)
586 offsetof(struct ocfs2_group_desc
, bg_bitmap
);
591 static inline int ocfs2_truncate_recs_per_inode(int blocksize
)
596 offsetof(struct ocfs2_dinode
, id2
.i_dealloc
.tl_recs
);
598 return size
/ sizeof(struct ocfs2_truncate_rec
);
600 #endif /* __KERNEL__ */
603 static inline int ocfs2_system_inode_is_global(int type
)
605 return ((type
>= 0) &&
606 (type
<= OCFS2_LAST_GLOBAL_SYSTEM_INODE
));
609 static inline int ocfs2_sprintf_system_inode_name(char *buf
, int len
,
615 * Global system inodes can only have one copy. Everything
616 * after OCFS2_LAST_GLOBAL_SYSTEM_INODE in the system inode
617 * list has a copy per slot.
619 if (type
<= OCFS2_LAST_GLOBAL_SYSTEM_INODE
)
620 chars
= snprintf(buf
, len
,
621 ocfs2_system_inodes
[type
].si_name
);
623 chars
= snprintf(buf
, len
,
624 ocfs2_system_inodes
[type
].si_name
,
630 static inline void ocfs2_set_de_type(struct ocfs2_dir_entry
*de
,
633 de
->file_type
= ocfs2_type_by_mode
[(mode
& S_IFMT
)>>S_SHIFT
];
636 #endif /* _OCFS2_FS_H */