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 \
98 | OCFS2_FEATURE_RO_COMPAT_USRQUOTA \
99 | OCFS2_FEATURE_RO_COMPAT_GRPQUOTA)
102 * Heartbeat-only devices are missing journals and other files. The
103 * filesystem driver can't load them, but the library can. Never put
104 * this in OCFS2_FEATURE_INCOMPAT_SUPP, *ever*.
106 #define OCFS2_FEATURE_INCOMPAT_HEARTBEAT_DEV 0x0002
109 * tunefs sets this incompat flag before starting the resize and clears it
110 * at the end. This flag protects users from inadvertently mounting the fs
111 * after an aborted run without fsck-ing.
113 #define OCFS2_FEATURE_INCOMPAT_RESIZE_INPROG 0x0004
115 /* Used to denote a non-clustered volume */
116 #define OCFS2_FEATURE_INCOMPAT_LOCAL_MOUNT 0x0008
118 /* Support for sparse allocation in b-trees */
119 #define OCFS2_FEATURE_INCOMPAT_SPARSE_ALLOC 0x0010
122 * Tunefs sets this incompat flag before starting an operation which
123 * would require cleanup on abort. This is done to protect users from
124 * inadvertently mounting the fs after an aborted run without
127 * s_tunefs_flags on the super block describes precisely which
128 * operations were in progress.
130 #define OCFS2_FEATURE_INCOMPAT_TUNEFS_INPROG 0x0020
132 /* Support for data packed into inode blocks */
133 #define OCFS2_FEATURE_INCOMPAT_INLINE_DATA 0x0040
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
146 /* Support for the extended slot map */
147 #define OCFS2_FEATURE_INCOMPAT_EXTENDED_SLOT_MAP 0x100
149 /* Support for extended attributes */
150 #define OCFS2_FEATURE_INCOMPAT_XATTR 0x0200
153 * backup superblock flag is used to indicate that this volume
154 * has backup superblocks.
156 #define OCFS2_FEATURE_COMPAT_BACKUP_SB 0x0001
159 * The filesystem will correctly handle journal feature bits.
161 #define OCFS2_FEATURE_COMPAT_JBD2_SB 0x0002
164 * Unwritten extents support.
166 #define OCFS2_FEATURE_RO_COMPAT_UNWRITTEN 0x0001
169 * Maintain quota information for this filesystem
171 #define OCFS2_FEATURE_RO_COMPAT_USRQUOTA 0x0002
172 #define OCFS2_FEATURE_RO_COMPAT_GRPQUOTA 0x0004
174 /* The byte offset of the first backup block will be 1G.
175 * The following will be 4G, 16G, 64G, 256G and 1T.
177 #define OCFS2_BACKUP_SB_START 1 << 30
179 /* the max backup superblock nums */
180 #define OCFS2_MAX_BACKUP_SUPERBLOCKS 6
183 * Flags on ocfs2_super_block.s_tunefs_flags
185 #define OCFS2_TUNEFS_INPROG_REMOVE_SLOT 0x0001 /* Removing slots */
188 * Flags on ocfs2_dinode.i_flags
190 #define OCFS2_VALID_FL (0x00000001) /* Inode is valid */
191 #define OCFS2_UNUSED2_FL (0x00000002)
192 #define OCFS2_ORPHANED_FL (0x00000004) /* On the orphan list */
193 #define OCFS2_UNUSED3_FL (0x00000008)
194 /* System inode flags */
195 #define OCFS2_SYSTEM_FL (0x00000010) /* System inode */
196 #define OCFS2_SUPER_BLOCK_FL (0x00000020) /* Super block */
197 #define OCFS2_LOCAL_ALLOC_FL (0x00000040) /* Slot local alloc bitmap */
198 #define OCFS2_BITMAP_FL (0x00000080) /* Allocation bitmap */
199 #define OCFS2_JOURNAL_FL (0x00000100) /* Slot local journal */
200 #define OCFS2_HEARTBEAT_FL (0x00000200) /* Heartbeat area */
201 #define OCFS2_CHAIN_FL (0x00000400) /* Chain allocator */
202 #define OCFS2_DEALLOC_FL (0x00000800) /* Truncate log */
203 #define OCFS2_QUOTA_FL (0x00001000) /* Quota file */
206 * Flags on ocfs2_dinode.i_dyn_features
208 * These can change much more often than i_flags. When adding flags,
209 * keep in mind that i_dyn_features is only 16 bits wide.
211 #define OCFS2_INLINE_DATA_FL (0x0001) /* Data stored in inode block */
212 #define OCFS2_HAS_XATTR_FL (0x0002)
213 #define OCFS2_INLINE_XATTR_FL (0x0004)
214 #define OCFS2_INDEXED_DIR_FL (0x0008)
216 /* Inode attributes, keep in sync with EXT2 */
217 #define OCFS2_SECRM_FL (0x00000001) /* Secure deletion */
218 #define OCFS2_UNRM_FL (0x00000002) /* Undelete */
219 #define OCFS2_COMPR_FL (0x00000004) /* Compress file */
220 #define OCFS2_SYNC_FL (0x00000008) /* Synchronous updates */
221 #define OCFS2_IMMUTABLE_FL (0x00000010) /* Immutable file */
222 #define OCFS2_APPEND_FL (0x00000020) /* writes to file may only append */
223 #define OCFS2_NODUMP_FL (0x00000040) /* do not dump file */
224 #define OCFS2_NOATIME_FL (0x00000080) /* do not update atime */
225 #define OCFS2_DIRSYNC_FL (0x00010000) /* dirsync behaviour (directories only) */
227 #define OCFS2_FL_VISIBLE (0x000100FF) /* User visible flags */
228 #define OCFS2_FL_MODIFIABLE (0x000100FF) /* User modifiable flags */
231 * Extent record flags (e_node.leaf.flags)
233 #define OCFS2_EXT_UNWRITTEN (0x01) /* Extent is allocated but
239 #define OCFS2_IOC_GETFLAGS _IOR('f', 1, long)
240 #define OCFS2_IOC_SETFLAGS _IOW('f', 2, long)
241 #define OCFS2_IOC32_GETFLAGS _IOR('f', 1, int)
242 #define OCFS2_IOC32_SETFLAGS _IOW('f', 2, int)
245 * Space reservation / allocation / free ioctls and argument structure
246 * are designed to be compatible with XFS.
248 * ALLOCSP* and FREESP* are not and will never be supported, but are
249 * included here for completeness.
251 struct ocfs2_space_resv {
255 __s64 l_len; /* len == 0 means until end of file */
258 __s32 l_pad[4]; /* reserve area */
261 #define OCFS2_IOC_ALLOCSP _IOW ('X', 10, struct ocfs2_space_resv)
262 #define OCFS2_IOC_FREESP _IOW ('X', 11, struct ocfs2_space_resv)
263 #define OCFS2_IOC_RESVSP _IOW ('X', 40, struct ocfs2_space_resv)
264 #define OCFS2_IOC_UNRESVSP _IOW ('X', 41, struct ocfs2_space_resv)
265 #define OCFS2_IOC_ALLOCSP64 _IOW ('X', 36, struct ocfs2_space_resv)
266 #define OCFS2_IOC_FREESP64 _IOW ('X', 37, struct ocfs2_space_resv)
267 #define OCFS2_IOC_RESVSP64 _IOW ('X', 42, struct ocfs2_space_resv)
268 #define OCFS2_IOC_UNRESVSP64 _IOW ('X', 43, struct ocfs2_space_resv)
270 /* Used to pass group descriptor data when online resize is done */
271 struct ocfs2_new_group_input {
272 __u64 group; /* Group descriptor's blkno. */
273 __u32 clusters; /* Total number of clusters in this group */
274 __u32 frees; /* Total free clusters in this group */
275 __u16 chain; /* Chain for this group */
280 #define OCFS2_IOC_GROUP_EXTEND _IOW('o', 1, int)
281 #define OCFS2_IOC_GROUP_ADD _IOW('o', 2,struct ocfs2_new_group_input)
282 #define OCFS2_IOC_GROUP_ADD64 _IOW('o', 3,struct ocfs2_new_group_input)
285 * Journal Flags (ocfs2_dinode.id1.journal1.i_flags)
287 #define OCFS2_JOURNAL_DIRTY_FL (0x00000001) /* Journal needs recovery */
290 * superblock s_state flags
292 #define OCFS2_ERROR_FS (0x00000001) /* FS saw errors */
294 /* Limit of space in ocfs2_dir_entry */
295 #define OCFS2_MAX_FILENAME_LEN 255
297 /* Maximum slots on an ocfs2 file system */
298 #define OCFS2_MAX_SLOTS 255
300 /* Slot map indicator for an empty slot */
301 #define OCFS2_INVALID_SLOT -1
303 #define OCFS2_VOL_UUID_LEN 16
304 #define OCFS2_MAX_VOL_LABEL_LEN 64
306 /* The alternate, userspace stack fields */
307 #define OCFS2_STACK_LABEL_LEN 4
308 #define OCFS2_CLUSTER_NAME_LEN 16
310 /* Journal limits (in bytes) */
311 #define OCFS2_MIN_JOURNAL_SIZE (4 * 1024 * 1024)
314 * Default local alloc size (in megabytes)
316 * The value chosen should be such that most allocations, including new
317 * block groups, use local alloc.
319 #define OCFS2_DEFAULT_LOCAL_ALLOC_SIZE 8
322 * Inline extended attribute size (in bytes)
323 * The value chosen should be aligned to 16 byte boundaries.
325 #define OCFS2_MIN_XATTR_INLINE_SIZE 256
327 struct ocfs2_system_inode_info {
333 /* System file index */
335 BAD_BLOCK_SYSTEM_INODE = 0,
336 GLOBAL_INODE_ALLOC_SYSTEM_INODE,
337 SLOT_MAP_SYSTEM_INODE,
338 #define OCFS2_FIRST_ONLINE_SYSTEM_INODE SLOT_MAP_SYSTEM_INODE
339 HEARTBEAT_SYSTEM_INODE,
340 GLOBAL_BITMAP_SYSTEM_INODE,
341 USER_QUOTA_SYSTEM_INODE,
342 GROUP_QUOTA_SYSTEM_INODE,
343 #define OCFS2_LAST_GLOBAL_SYSTEM_INODE GROUP_QUOTA_SYSTEM_INODE
344 ORPHAN_DIR_SYSTEM_INODE,
345 EXTENT_ALLOC_SYSTEM_INODE,
346 INODE_ALLOC_SYSTEM_INODE,
347 JOURNAL_SYSTEM_INODE,
348 LOCAL_ALLOC_SYSTEM_INODE,
349 TRUNCATE_LOG_SYSTEM_INODE,
350 LOCAL_USER_QUOTA_SYSTEM_INODE,
351 LOCAL_GROUP_QUOTA_SYSTEM_INODE,
355 static struct ocfs2_system_inode_info ocfs2_system_inodes[NUM_SYSTEM_INODES] = {
356 /* Global system inodes (single copy) */
357 /* The first two are only used from userspace mfks/tunefs */
358 [BAD_BLOCK_SYSTEM_INODE] = { "bad_blocks", 0, S_IFREG | 0644 },
359 [GLOBAL_INODE_ALLOC_SYSTEM_INODE] = { "global_inode_alloc", OCFS2_BITMAP_FL | OCFS2_CHAIN_FL, S_IFREG | 0644 },
361 /* These are used by the running filesystem */
362 [SLOT_MAP_SYSTEM_INODE] = { "slot_map", 0, S_IFREG | 0644 },
363 [HEARTBEAT_SYSTEM_INODE] = { "heartbeat", OCFS2_HEARTBEAT_FL, S_IFREG | 0644 },
364 [GLOBAL_BITMAP_SYSTEM_INODE] = { "global_bitmap", 0, S_IFREG | 0644 },
365 [USER_QUOTA_SYSTEM_INODE] = { "aquota.user", OCFS2_QUOTA_FL, S_IFREG | 0644 },
366 [GROUP_QUOTA_SYSTEM_INODE] = { "aquota.group", OCFS2_QUOTA_FL, S_IFREG | 0644 },
368 /* Slot-specific system inodes (one copy per slot) */
369 [ORPHAN_DIR_SYSTEM_INODE] = { "orphan_dir:%04d", 0, S_IFDIR | 0755 },
370 [EXTENT_ALLOC_SYSTEM_INODE] = { "extent_alloc:%04d", OCFS2_BITMAP_FL | OCFS2_CHAIN_FL, S_IFREG | 0644 },
371 [INODE_ALLOC_SYSTEM_INODE] = { "inode_alloc:%04d", OCFS2_BITMAP_FL | OCFS2_CHAIN_FL, S_IFREG | 0644 },
372 [JOURNAL_SYSTEM_INODE] = { "journal:%04d", OCFS2_JOURNAL_FL, S_IFREG | 0644 },
373 [LOCAL_ALLOC_SYSTEM_INODE] = { "local_alloc:%04d", OCFS2_BITMAP_FL | OCFS2_LOCAL_ALLOC_FL, S_IFREG | 0644 },
374 [TRUNCATE_LOG_SYSTEM_INODE] = { "truncate_log:%04d", OCFS2_DEALLOC_FL, S_IFREG | 0644 },
375 [LOCAL_USER_QUOTA_SYSTEM_INODE] = { "aquota.user:%04d", OCFS2_QUOTA_FL, S_IFREG | 0644 },
376 [LOCAL_GROUP_QUOTA_SYSTEM_INODE] = { "aquota.group:%04d", OCFS2_QUOTA_FL, S_IFREG | 0644 },
379 /* Parameter passed from mount.ocfs2 to module */
380 #define OCFS2_HB_NONE "heartbeat=none"
381 #define OCFS2_HB_LOCAL "heartbeat=local"
384 * OCFS2 directory file types. Only the low 3 bits are used. The
385 * other bits are reserved for now.
387 #define OCFS2_FT_UNKNOWN 0
388 #define OCFS2_FT_REG_FILE 1
389 #define OCFS2_FT_DIR 2
390 #define OCFS2_FT_CHRDEV 3
391 #define OCFS2_FT_BLKDEV 4
392 #define OCFS2_FT_FIFO 5
393 #define OCFS2_FT_SOCK 6
394 #define OCFS2_FT_SYMLINK 7
396 #define OCFS2_FT_MAX 8
399 * OCFS2_DIR_PAD defines the directory entries boundaries
401 * NOTE: It must be a multiple of 4
403 #define OCFS2_DIR_PAD 4
404 #define OCFS2_DIR_ROUND (OCFS2_DIR_PAD - 1)
405 #define OCFS2_DIR_MEMBER_LEN offsetof(struct ocfs2_dir_entry, name)
406 #define OCFS2_DIR_REC_LEN(name_len) (((name_len) + OCFS2_DIR_MEMBER_LEN + \
410 #define OCFS2_LINK_MAX 32000
413 static unsigned char ocfs2_type_by_mode[S_IFMT >> S_SHIFT] = {
414 [S_IFREG >> S_SHIFT] = OCFS2_FT_REG_FILE,
415 [S_IFDIR >> S_SHIFT] = OCFS2_FT_DIR,
416 [S_IFCHR >> S_SHIFT] = OCFS2_FT_CHRDEV,
417 [S_IFBLK >> S_SHIFT] = OCFS2_FT_BLKDEV,
418 [S_IFIFO >> S_SHIFT] = OCFS2_FT_FIFO,
419 [S_IFSOCK >> S_SHIFT] = OCFS2_FT_SOCK,
420 [S_IFLNK >> S_SHIFT] = OCFS2_FT_SYMLINK,
427 #define OCFS2_RAW_SB(dinode) (&((dinode)->id2.i_super))
430 * On disk extent record for OCFS2
431 * It describes a range of clusters on disk.
433 * Length fields are divided into interior and leaf node versions.
434 * This leaves room for a flags field (OCFS2_EXT_*) in the leaf nodes.
436 struct ocfs2_extent_rec {
437 /*00*/ __le32 e_cpos; /* Offset into the file, in clusters */
439 __le32 e_int_clusters; /* Clusters covered by all children */
441 __le16 e_leaf_clusters; /* Clusters covered by this
444 __u8 e_flags; /* Extent flags */
447 __le64 e_blkno; /* Physical disk offset, in blocks */
451 struct ocfs2_chain_rec {
452 __le32 c_free; /* Number of free bits in this chain. */
453 __le32 c_total; /* Number of total bits in this chain */
454 __le64 c_blkno; /* Physical disk offset (blocks) of 1st group */
457 struct ocfs2_truncate_rec {
458 __le32 t_start; /* 1st cluster in this log */
459 __le32 t_clusters; /* Number of total clusters covered */
463 * On disk extent list for OCFS2 (node in the tree). Note that this
464 * is contained inside ocfs2_dinode or ocfs2_extent_block, so the
465 * offsets are relative to ocfs2_dinode.id2.i_list or
466 * ocfs2_extent_block.h_list, respectively.
468 struct ocfs2_extent_list {
469 /*00*/ __le16 l_tree_depth; /* Extent tree depth from this
470 point. 0 means data extents
471 hang directly off this
473 NOTE: The high 8 bits cannot be
474 used - tree_depth is never that big.
476 __le16 l_count; /* Number of extent records */
477 __le16 l_next_free_rec; /* Next unused extent slot */
479 __le64 l_reserved2; /* Pad to
480 sizeof(ocfs2_extent_rec) */
481 /*10*/ struct ocfs2_extent_rec l_recs[0]; /* Extent records */
485 * On disk allocation chain list for OCFS2. Note that this is
486 * contained inside ocfs2_dinode, so the offsets are relative to
487 * ocfs2_dinode.id2.i_chain.
489 struct ocfs2_chain_list {
490 /*00*/ __le16 cl_cpg; /* Clusters per Block Group */
491 __le16 cl_bpc; /* Bits per cluster */
492 __le16 cl_count; /* Total chains in this list */
493 __le16 cl_next_free_rec; /* Next unused chain slot */
495 /*10*/ struct ocfs2_chain_rec cl_recs[0]; /* Chain records */
499 * On disk deallocation log for OCFS2. Note that this is
500 * contained inside ocfs2_dinode, so the offsets are relative to
501 * ocfs2_dinode.id2.i_dealloc.
503 struct ocfs2_truncate_log {
504 /*00*/ __le16 tl_count; /* Total records in this log */
505 __le16 tl_used; /* Number of records in use */
507 /*08*/ struct ocfs2_truncate_rec tl_recs[0]; /* Truncate records */
511 * On disk extent block (indirect block) for OCFS2
513 struct ocfs2_extent_block
515 /*00*/ __u8 h_signature[8]; /* Signature for verification */
517 /*10*/ __le16 h_suballoc_slot; /* Slot suballocator this
518 extent_header belongs to */
519 __le16 h_suballoc_bit; /* Bit offset in suballocator
521 __le32 h_fs_generation; /* Must match super block */
522 __le64 h_blkno; /* Offset on disk, in blocks */
523 /*20*/ __le64 h_reserved3;
524 __le64 h_next_leaf_blk; /* Offset on disk, in blocks,
525 of next leaf header pointing
527 /*30*/ struct ocfs2_extent_list h_list; /* Extent record list */
528 /* Actual on-disk size is one block */
532 * On disk slot map for OCFS2. This defines the contents of the "slot_map"
533 * system file. A slot is valid if it contains a node number >= 0. The
534 * value -1 (0xFFFF) is OCFS2_INVALID_SLOT. This marks a slot empty.
536 struct ocfs2_slot_map {
537 /*00*/ __le16 sm_slots[0];
539 * Actual on-disk size is one block. OCFS2_MAX_SLOTS is 255,
540 * 255 * sizeof(__le16) == 512B, within the 512B block minimum blocksize.
544 struct ocfs2_extended_slot {
545 /*00*/ __u8 es_valid;
546 __u8 es_reserved1[3];
552 * The extended slot map, used when OCFS2_FEATURE_INCOMPAT_EXTENDED_SLOT_MAP
553 * is set. It separates out the valid marker from the node number, and
554 * has room to grow. Unlike the old slot map, this format is defined by
557 struct ocfs2_slot_map_extended {
558 /*00*/ struct ocfs2_extended_slot se_slots[0];
560 * Actual size is i_size of the slot_map system file. It should
561 * match s_max_slots * sizeof(struct ocfs2_extended_slot)
565 struct ocfs2_cluster_info {
566 /*00*/ __u8 ci_stack[OCFS2_STACK_LABEL_LEN];
568 /*08*/ __u8 ci_cluster[OCFS2_CLUSTER_NAME_LEN];
573 * On disk superblock for OCFS2
574 * Note that it is contained inside an ocfs2_dinode, so all offsets
575 * are relative to the start of ocfs2_dinode.id2.
577 struct ocfs2_super_block {
578 /*00*/ __le16 s_major_rev_level;
579 __le16 s_minor_rev_level;
581 __le16 s_max_mnt_count;
582 __le16 s_state; /* File system state */
583 __le16 s_errors; /* Behaviour when detecting errors */
584 __le32 s_checkinterval; /* Max time between checks */
585 /*10*/ __le64 s_lastcheck; /* Time of last check */
586 __le32 s_creator_os; /* OS */
587 __le32 s_feature_compat; /* Compatible feature set */
588 /*20*/ __le32 s_feature_incompat; /* Incompatible feature set */
589 __le32 s_feature_ro_compat; /* Readonly-compatible feature set */
590 __le64 s_root_blkno; /* Offset, in blocks, of root directory
592 /*30*/ __le64 s_system_dir_blkno; /* Offset, in blocks, of system
594 __le32 s_blocksize_bits; /* Blocksize for this fs */
595 __le32 s_clustersize_bits; /* Clustersize for this fs */
596 /*40*/ __le16 s_max_slots; /* Max number of simultaneous mounts
597 before tunefs required */
598 __le16 s_tunefs_flag;
599 __le32 s_uuid_hash; /* hash value of uuid */
600 __le64 s_first_cluster_group; /* Block offset of 1st cluster
602 /*50*/ __u8 s_label[OCFS2_MAX_VOL_LABEL_LEN]; /* Label for mounting, etc. */
603 /*90*/ __u8 s_uuid[OCFS2_VOL_UUID_LEN]; /* 128-bit uuid */
604 /*A0*/ struct ocfs2_cluster_info s_cluster_info; /* Selected userspace
606 with INCOMPAT flag. */
607 /*B8*/ __le16 s_xattr_inline_size; /* extended attribute inline size
611 /*C0*/ __le64 s_reserved2[16]; /* Fill out superblock */
615 * NOTE: As stated above, all offsets are relative to
616 * ocfs2_dinode.id2, which is at 0xC0 in the inode.
617 * 0xC0 + 0x140 = 0x200 or 512 bytes. A superblock must fit within
618 * our smallest blocksize, which is 512 bytes. To ensure this,
619 * we reserve the space in s_reserved2. Anything past s_reserved2
620 * will not be available on the smallest blocksize.
625 * Local allocation bitmap for OCFS2 slots
626 * Note that it exists inside an ocfs2_dinode, so all offsets are
627 * relative to the start of ocfs2_dinode.id2.
629 struct ocfs2_local_alloc
631 /*00*/ __le32 la_bm_off; /* Starting bit offset in main bitmap */
632 __le16 la_size; /* Size of included bitmap, in bytes */
635 /*10*/ __u8 la_bitmap[0];
639 * Data-in-inode header. This is only used if i_dyn_features has
640 * OCFS2_INLINE_DATA_FL set.
642 struct ocfs2_inline_data
644 /*00*/ __le16 id_count; /* Number of bytes that can be used
645 * for data, starting at id_data */
648 __u8 id_data[0]; /* Start of user data */
652 * On disk inode for OCFS2
654 struct ocfs2_dinode {
655 /*00*/ __u8 i_signature[8]; /* Signature for validation */
656 __le32 i_generation; /* Generation number */
657 __le16 i_suballoc_slot; /* Slot suballocator this inode
659 __le16 i_suballoc_bit; /* Bit offset in suballocator
661 /*10*/ __le16 i_reserved0;
662 __le16 i_xattr_inline_size;
663 __le32 i_clusters; /* Cluster count */
664 __le32 i_uid; /* Owner UID */
665 __le32 i_gid; /* Owning GID */
666 /*20*/ __le64 i_size; /* Size in bytes */
667 __le16 i_mode; /* File mode */
668 __le16 i_links_count; /* Links count */
669 __le32 i_flags; /* File flags */
670 /*30*/ __le64 i_atime; /* Access time */
671 __le64 i_ctime; /* Creation time */
672 /*40*/ __le64 i_mtime; /* Modification time */
673 __le64 i_dtime; /* Deletion time */
674 /*50*/ __le64 i_blkno; /* Offset on disk, in blocks */
675 __le64 i_last_eb_blk; /* Pointer to last extent
677 /*60*/ __le32 i_fs_generation; /* Generation per fs-instance */
681 /*70*/ __le32 i_attr;
682 __le16 i_orphaned_slot; /* Only valid when OCFS2_ORPHANED_FL
683 was set in i_flags */
684 __le16 i_dyn_features;
686 /*80*/ __le64 i_reserved2[7];
688 __le64 i_pad1; /* Generic way to refer to this
691 __le64 i_rdev; /* Device number */
693 struct { /* Info for bitmap system
695 __le32 i_used; /* Bits (ie, clusters) used */
696 __le32 i_total; /* Total bits (clusters)
699 struct { /* Info for journal system
701 __le32 ij_flags; /* Mounted, version, etc. */
702 __le32 ij_recovery_generation; /* Incremented when the
707 } id1; /* Inode type dependant 1 */
709 struct ocfs2_super_block i_super;
710 struct ocfs2_local_alloc i_lab;
711 struct ocfs2_chain_list i_chain;
712 struct ocfs2_extent_list i_list;
713 struct ocfs2_truncate_log i_dealloc;
714 struct ocfs2_inline_data i_data;
717 /* Actual on-disk size is one block */
721 * On-disk directory entry structure for OCFS2
723 * Packed as this structure could be accessed unaligned on 64-bit platforms
725 struct ocfs2_dir_entry {
726 /*00*/ __le64 inode; /* Inode number */
727 __le16 rec_len; /* Directory entry length */
728 __u8 name_len; /* Name length */
730 /*0C*/ char name[OCFS2_MAX_FILENAME_LEN]; /* File name */
731 /* Actual on-disk length specified by rec_len */
732 } __attribute__ ((packed));
735 * On disk allocator group structure for OCFS2
737 struct ocfs2_group_desc
739 /*00*/ __u8 bg_signature[8]; /* Signature for validation */
740 __le16 bg_size; /* Size of included bitmap in
742 __le16 bg_bits; /* Bits represented by this
744 __le16 bg_free_bits_count; /* Free bits count */
745 __le16 bg_chain; /* What chain I am in. */
746 /*10*/ __le32 bg_generation;
748 __le64 bg_next_group; /* Next group in my list, in
750 /*20*/ __le64 bg_parent_dinode; /* dinode which owns me, in
752 __le64 bg_blkno; /* Offset on disk, in blocks */
753 /*30*/ __le64 bg_reserved2[2];
754 /*40*/ __u8 bg_bitmap[0];
758 * On disk extended attribute structure for OCFS2.
762 * ocfs2_xattr_entry indicates one extend attribute.
764 * Note that it can be stored in inode, one block or one xattr bucket.
766 struct ocfs2_xattr_entry {
767 __le32 xe_name_hash; /* hash value of xattr prefix+suffix. */
768 __le16 xe_name_offset; /* byte offset from the 1st entry in the
769 local xattr storage(inode, xattr block or
771 __u8 xe_name_len; /* xattr name len, does't include prefix. */
772 __u8 xe_type; /* the low 7 bits indicate the name prefix
773 * type and the highest bit indicates whether
774 * the EA is stored in the local storage. */
775 __le64 xe_value_size; /* real xattr value length. */
779 * On disk structure for xattr header.
781 * One ocfs2_xattr_header describes how many ocfs2_xattr_entry records in
782 * the local xattr storage.
784 struct ocfs2_xattr_header {
785 __le16 xh_count; /* contains the count of how
786 many records are in the
787 local xattr storage. */
788 __le16 xh_free_start; /* current offset for storing
790 __le16 xh_name_value_len; /* total length of name/value
791 length in this bucket. */
792 __le16 xh_num_buckets; /* Number of xattr buckets
793 in this extent record,
794 only valid in the first
797 struct ocfs2_xattr_entry xh_entries[0]; /* xattr entry list. */
801 * On disk structure for xattr value root.
803 * When an xattr's value is large enough, it is stored in an external
804 * b-tree like file data. The xattr value root points to this structure.
806 struct ocfs2_xattr_value_root {
807 /*00*/ __le32 xr_clusters; /* clusters covered by xattr value. */
809 __le64 xr_last_eb_blk; /* Pointer to last extent block */
810 /*10*/ struct ocfs2_extent_list xr_list; /* Extent record list */
814 * On disk structure for xattr tree root.
816 * It is used when there are too many extended attributes for one file. These
817 * attributes will be organized and stored in an indexed-btree.
819 struct ocfs2_xattr_tree_root {
820 /*00*/ __le32 xt_clusters; /* clusters covered by xattr. */
822 __le64 xt_last_eb_blk; /* Pointer to last extent block */
823 /*10*/ struct ocfs2_extent_list xt_list; /* Extent record list */
826 #define OCFS2_XATTR_INDEXED 0x1
827 #define OCFS2_HASH_SHIFT 5
828 #define OCFS2_XATTR_ROUND 3
829 #define OCFS2_XATTR_SIZE(size) (((size) + OCFS2_XATTR_ROUND) & \
830 ~(OCFS2_XATTR_ROUND))
832 #define OCFS2_XATTR_BUCKET_SIZE 4096
833 #define OCFS2_XATTR_MAX_BLOCKS_PER_BUCKET (OCFS2_XATTR_BUCKET_SIZE \
834 / OCFS2_MIN_BLOCKSIZE)
837 * On disk structure for xattr block.
839 struct ocfs2_xattr_block {
840 /*00*/ __u8 xb_signature[8]; /* Signature for verification */
841 __le16 xb_suballoc_slot; /* Slot suballocator this
843 __le16 xb_suballoc_bit; /* Bit offset in suballocator
845 __le32 xb_fs_generation; /* Must match super block */
846 /*10*/ __le64 xb_blkno; /* Offset on disk, in blocks */
848 /*20*/ __le16 xb_flags; /* Indicates whether this block contains
849 real xattr or a xattr tree. */
854 struct ocfs2_xattr_header xb_header; /* xattr header if this
855 block contains xattr */
856 struct ocfs2_xattr_tree_root xb_root;/* xattr tree root if this
862 #define OCFS2_XATTR_ENTRY_LOCAL 0x80
863 #define OCFS2_XATTR_TYPE_MASK 0x7F
864 static inline void ocfs2_xattr_set_local(struct ocfs2_xattr_entry *xe,
868 xe->xe_type |= OCFS2_XATTR_ENTRY_LOCAL;
870 xe->xe_type &= ~OCFS2_XATTR_ENTRY_LOCAL;
873 static inline int ocfs2_xattr_is_local(struct ocfs2_xattr_entry *xe)
875 return xe->xe_type & OCFS2_XATTR_ENTRY_LOCAL;
878 static inline void ocfs2_xattr_set_type(struct ocfs2_xattr_entry *xe, int type)
880 xe->xe_type |= type & OCFS2_XATTR_TYPE_MASK;
883 static inline int ocfs2_xattr_get_type(struct ocfs2_xattr_entry *xe)
885 return xe->xe_type & OCFS2_XATTR_TYPE_MASK;
889 * On disk structures for global quota file
892 /* Magic numbers and known versions for global quota files */
893 #define OCFS2_GLOBAL_QMAGICS {\
894 0x0cf52470, /* USRQUOTA */ \
895 0x0cf52471 /* GRPQUOTA */ \
898 #define OCFS2_GLOBAL_QVERSIONS {\
904 /* Each block of each quota file has a certain fixed number of bytes reserved
905 * for OCFS2 internal use at its end. OCFS2 can use it for things like
907 #define OCFS2_QBLK_RESERVED_SPACE 8
909 /* Generic header of all quota files */
910 struct ocfs2_disk_dqheader {
911 __le32 dqh_magic; /* Magic number identifying file */
912 __le32 dqh_version; /* Quota format version */
915 #define OCFS2_GLOBAL_INFO_OFF (sizeof(struct ocfs2_disk_dqheader))
917 /* Information header of global quota file (immediately follows the generic
919 struct ocfs2_global_disk_dqinfo {
920 /*00*/ __le32 dqi_bgrace; /* Grace time for space softlimit excess */
921 __le32 dqi_igrace; /* Grace time for inode softlimit excess */
922 __le32 dqi_syncms; /* Time after which we sync local changes to
923 * global quota file */
924 __le32 dqi_blocks; /* Number of blocks in quota file */
925 /*10*/ __le32 dqi_free_blk; /* First free block in quota file */
926 __le32 dqi_free_entry; /* First block with free dquot entry in quota
930 /* Structure with global user / group information. We reserve some space
932 struct ocfs2_global_disk_dqblk {
933 /*00*/ __le32 dqb_id; /* ID the structure belongs to */
934 __le32 dqb_use_count; /* Number of nodes having reference to this structure */
935 __le64 dqb_ihardlimit; /* absolute limit on allocated inodes */
936 /*10*/ __le64 dqb_isoftlimit; /* preferred inode limit */
937 __le64 dqb_curinodes; /* current # allocated inodes */
938 /*20*/ __le64 dqb_bhardlimit; /* absolute limit on disk space */
939 __le64 dqb_bsoftlimit; /* preferred limit on disk space */
940 /*30*/ __le64 dqb_curspace; /* current space occupied */
941 __le64 dqb_btime; /* time limit for excessive disk use */
942 /*40*/ __le64 dqb_itime; /* time limit for excessive inode use */
944 /*50*/ __le64 dqb_pad2;
948 * On-disk structures for local quota file
951 /* Magic numbers and known versions for local quota files */
952 #define OCFS2_LOCAL_QMAGICS {\
953 0x0cf524c0, /* USRQUOTA */ \
954 0x0cf524c1 /* GRPQUOTA */ \
957 #define OCFS2_LOCAL_QVERSIONS {\
962 /* Quota flags in dqinfo header */
963 #define OLQF_CLEAN 0x0001 /* Quota file is empty (this should be after\
964 * quota has been cleanly turned off) */
966 #define OCFS2_LOCAL_INFO_OFF (sizeof(struct ocfs2_disk_dqheader))
968 /* Information header of local quota file (immediately follows the generic
970 struct ocfs2_local_disk_dqinfo {
971 __le32 dqi_flags; /* Flags for quota file */
972 __le32 dqi_chunks; /* Number of chunks of quota structures
974 __le32 dqi_blocks; /* Number of blocks allocated for quota file */
977 /* Header of one chunk of a quota file */
978 struct ocfs2_local_disk_chunk {
979 __le32 dqc_free; /* Number of free entries in the bitmap */
980 u8 dqc_bitmap[0]; /* Bitmap of entries in the corresponding
981 * chunk of quota file */
984 /* One entry in local quota file */
985 struct ocfs2_local_disk_dqblk {
986 /*00*/ __le64 dqb_id; /* id this quota applies to */
987 __le64 dqb_spacemod; /* Change in the amount of used space */
988 /*10*/ __le64 dqb_inodemod; /* Change in the amount of used inodes */
992 static inline int ocfs2_fast_symlink_chars(struct super_block *sb)
994 return sb->s_blocksize -
995 offsetof(struct ocfs2_dinode, id2.i_symlink);
998 static inline int ocfs2_max_inline_data(struct super_block *sb)
1000 return sb->s_blocksize -
1001 offsetof(struct ocfs2_dinode, id2.i_data.id_data);
1004 static inline int ocfs2_max_inline_data_with_xattr(struct super_block *sb,
1005 struct ocfs2_dinode *di)
1007 unsigned int xattrsize = le16_to_cpu(di->i_xattr_inline_size);
1009 if (le16_to_cpu(di->i_dyn_features) & OCFS2_INLINE_XATTR_FL)
1010 return sb->s_blocksize -
1011 offsetof(struct ocfs2_dinode, id2.i_data.id_data) -
1014 return sb->s_blocksize -
1015 offsetof(struct ocfs2_dinode, id2.i_data.id_data);
1018 static inline int ocfs2_extent_recs_per_inode(struct super_block *sb)
1022 size = sb->s_blocksize -
1023 offsetof(struct ocfs2_dinode, id2.i_list.l_recs);
1025 return size / sizeof(struct ocfs2_extent_rec);
1028 static inline int ocfs2_extent_recs_per_inode_with_xattr(
1029 struct super_block *sb,
1030 struct ocfs2_dinode *di)
1033 unsigned int xattrsize = le16_to_cpu(di->i_xattr_inline_size);
1035 if (le16_to_cpu(di->i_dyn_features) & OCFS2_INLINE_XATTR_FL)
1036 size = sb->s_blocksize -
1037 offsetof(struct ocfs2_dinode, id2.i_list.l_recs) -
1040 size = sb->s_blocksize -
1041 offsetof(struct ocfs2_dinode, id2.i_list.l_recs);
1043 return size / sizeof(struct ocfs2_extent_rec);
1046 static inline int ocfs2_chain_recs_per_inode(struct super_block *sb)
1050 size = sb->s_blocksize -
1051 offsetof(struct ocfs2_dinode, id2.i_chain.cl_recs);
1053 return size / sizeof(struct ocfs2_chain_rec);
1056 static inline u16 ocfs2_extent_recs_per_eb(struct super_block *sb)
1060 size = sb->s_blocksize -
1061 offsetof(struct ocfs2_extent_block, h_list.l_recs);
1063 return size / sizeof(struct ocfs2_extent_rec);
1066 static inline u16 ocfs2_local_alloc_size(struct super_block *sb)
1070 size = sb->s_blocksize -
1071 offsetof(struct ocfs2_dinode, id2.i_lab.la_bitmap);
1076 static inline int ocfs2_group_bitmap_size(struct super_block *sb)
1080 size = sb->s_blocksize -
1081 offsetof(struct ocfs2_group_desc, bg_bitmap);
1086 static inline int ocfs2_truncate_recs_per_inode(struct super_block *sb)
1090 size = sb->s_blocksize -
1091 offsetof(struct ocfs2_dinode, id2.i_dealloc.tl_recs);
1093 return size / sizeof(struct ocfs2_truncate_rec);
1096 static inline u64 ocfs2_backup_super_blkno(struct super_block *sb, int index)
1098 u64 offset = OCFS2_BACKUP_SB_START;
1100 if (index >= 0 && index < OCFS2_MAX_BACKUP_SUPERBLOCKS) {
1101 offset <<= (2 * index);
1102 offset >>= sb->s_blocksize_bits;
1110 static inline u16 ocfs2_xattr_recs_per_xb(struct super_block *sb)
1114 size = sb->s_blocksize -
1115 offsetof(struct ocfs2_xattr_block,
1116 xb_attrs.xb_root.xt_list.l_recs);
1118 return size / sizeof(struct ocfs2_extent_rec);
1121 static inline int ocfs2_fast_symlink_chars(int blocksize)
1123 return blocksize - offsetof(struct ocfs2_dinode, id2.i_symlink);
1126 static inline int ocfs2_max_inline_data(int blocksize)
1128 return blocksize - offsetof(struct ocfs2_dinode, id2.i_data.id_data);
1131 static inline int ocfs2_extent_recs_per_inode(int blocksize)
1136 offsetof(struct ocfs2_dinode, id2.i_list.l_recs);
1138 return size / sizeof(struct ocfs2_extent_rec);
1141 static inline int ocfs2_chain_recs_per_inode(int blocksize)
1146 offsetof(struct ocfs2_dinode, id2.i_chain.cl_recs);
1148 return size / sizeof(struct ocfs2_chain_rec);
1151 static inline int ocfs2_extent_recs_per_eb(int blocksize)
1156 offsetof(struct ocfs2_extent_block, h_list.l_recs);
1158 return size / sizeof(struct ocfs2_extent_rec);
1161 static inline int ocfs2_local_alloc_size(int blocksize)
1166 offsetof(struct ocfs2_dinode, id2.i_lab.la_bitmap);
1171 static inline int ocfs2_group_bitmap_size(int blocksize)
1176 offsetof(struct ocfs2_group_desc, bg_bitmap);
1181 static inline int ocfs2_truncate_recs_per_inode(int blocksize)
1186 offsetof(struct ocfs2_dinode, id2.i_dealloc.tl_recs);
1188 return size / sizeof(struct ocfs2_truncate_rec);
1191 static inline uint64_t ocfs2_backup_super_blkno(int blocksize, int index)
1193 uint64_t offset = OCFS2_BACKUP_SB_START;
1195 if (index >= 0 && index < OCFS2_MAX_BACKUP_SUPERBLOCKS) {
1196 offset <<= (2 * index);
1197 offset /= blocksize;
1204 static inline int ocfs2_xattr_recs_per_xb(int blocksize)
1209 offsetof(struct ocfs2_xattr_block,
1210 xb_attrs.xb_root.xt_list.l_recs);
1212 return size / sizeof(struct ocfs2_extent_rec);
1214 #endif /* __KERNEL__ */
1217 static inline int ocfs2_system_inode_is_global(int type)
1219 return ((type >= 0) &&
1220 (type <= OCFS2_LAST_GLOBAL_SYSTEM_INODE));
1223 static inline int ocfs2_sprintf_system_inode_name(char *buf, int len,
1229 * Global system inodes can only have one copy. Everything
1230 * after OCFS2_LAST_GLOBAL_SYSTEM_INODE in the system inode
1231 * list has a copy per slot.
1233 if (type <= OCFS2_LAST_GLOBAL_SYSTEM_INODE)
1234 chars = snprintf(buf, len, "%s",
1235 ocfs2_system_inodes[type].si_name);
1237 chars = snprintf(buf, len,
1238 ocfs2_system_inodes[type].si_name,
1244 static inline void ocfs2_set_de_type(struct ocfs2_dir_entry *de,
1247 de->file_type = ocfs2_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
1250 #endif /* _OCFS2_FS_H */