Root/fs/ocfs2/ocfs2_fs.h

1/* -*- mode: c; c-basic-offset: 8; -*-
2 * vim: noexpandtab sw=8 ts=8 sts=0:
3 *
4 * ocfs2_fs.h
5 *
6 * On-disk structures for OCFS2.
7 *
8 * Copyright (C) 2002, 2004 Oracle. All rights reserved.
9 *
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.
13 *
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.
18 *
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.
23 */
24
25#ifndef _OCFS2_FS_H
26#define _OCFS2_FS_H
27
28/* Version */
29#define OCFS2_MAJOR_REV_LEVEL 0
30#define OCFS2_MINOR_REV_LEVEL 90
31
32/*
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.
37 *
38 * All other structures are found from the superblock information.
39 *
40 * OCFS2_SUPER_BLOCK_BLKNO is in blocks, not sectors. eg, for a
41 * blocksize of 2K, it is 4096 bytes into disk.
42 */
43#define OCFS2_SUPER_BLOCK_BLKNO 2
44
45/*
46 * Cluster size limits. The maximum is kept arbitrarily at 1 MB, and could
47 * grow if needed.
48 */
49#define OCFS2_MIN_CLUSTERSIZE 4096
50#define OCFS2_MAX_CLUSTERSIZE 1048576
51
52/*
53 * Blocks cannot be bigger than clusters, so the maximum blocksize is the
54 * minimum cluster size.
55 */
56#define OCFS2_MIN_BLOCKSIZE 512
57#define OCFS2_MAX_BLOCKSIZE OCFS2_MIN_CLUSTERSIZE
58
59/* Filesystem magic number */
60#define OCFS2_SUPER_MAGIC 0x7461636f
61
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"
68#define OCFS2_DIR_TRAILER_SIGNATURE "DIRTRL1"
69#define OCFS2_DX_ROOT_SIGNATURE "DXDIR01"
70#define OCFS2_DX_LEAF_SIGNATURE "DXLEAF1"
71#define OCFS2_REFCOUNT_BLOCK_SIGNATURE "REFCNT1"
72
73/* Compatibility flags */
74#define OCFS2_HAS_COMPAT_FEATURE(sb,mask) \
75    ( OCFS2_SB(sb)->s_feature_compat & (mask) )
76#define OCFS2_HAS_RO_COMPAT_FEATURE(sb,mask) \
77    ( OCFS2_SB(sb)->s_feature_ro_compat & (mask) )
78#define OCFS2_HAS_INCOMPAT_FEATURE(sb,mask) \
79    ( OCFS2_SB(sb)->s_feature_incompat & (mask) )
80#define OCFS2_SET_COMPAT_FEATURE(sb,mask) \
81    OCFS2_SB(sb)->s_feature_compat |= (mask)
82#define OCFS2_SET_RO_COMPAT_FEATURE(sb,mask) \
83    OCFS2_SB(sb)->s_feature_ro_compat |= (mask)
84#define OCFS2_SET_INCOMPAT_FEATURE(sb,mask) \
85    OCFS2_SB(sb)->s_feature_incompat |= (mask)
86#define OCFS2_CLEAR_COMPAT_FEATURE(sb,mask) \
87    OCFS2_SB(sb)->s_feature_compat &= ~(mask)
88#define OCFS2_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
89    OCFS2_SB(sb)->s_feature_ro_compat &= ~(mask)
90#define OCFS2_CLEAR_INCOMPAT_FEATURE(sb,mask) \
91    OCFS2_SB(sb)->s_feature_incompat &= ~(mask)
92
93#define OCFS2_FEATURE_COMPAT_SUPP (OCFS2_FEATURE_COMPAT_BACKUP_SB \
94                     | OCFS2_FEATURE_COMPAT_JBD2_SB)
95#define OCFS2_FEATURE_INCOMPAT_SUPP (OCFS2_FEATURE_INCOMPAT_LOCAL_MOUNT \
96                     | OCFS2_FEATURE_INCOMPAT_SPARSE_ALLOC \
97                     | OCFS2_FEATURE_INCOMPAT_INLINE_DATA \
98                     | OCFS2_FEATURE_INCOMPAT_EXTENDED_SLOT_MAP \
99                     | OCFS2_FEATURE_INCOMPAT_USERSPACE_STACK \
100                     | OCFS2_FEATURE_INCOMPAT_XATTR \
101                     | OCFS2_FEATURE_INCOMPAT_META_ECC \
102                     | OCFS2_FEATURE_INCOMPAT_INDEXED_DIRS \
103                     | OCFS2_FEATURE_INCOMPAT_REFCOUNT_TREE \
104                     | OCFS2_FEATURE_INCOMPAT_DISCONTIG_BG \
105                     | OCFS2_FEATURE_INCOMPAT_CLUSTERINFO)
106#define OCFS2_FEATURE_RO_COMPAT_SUPP (OCFS2_FEATURE_RO_COMPAT_UNWRITTEN \
107                     | OCFS2_FEATURE_RO_COMPAT_USRQUOTA \
108                     | OCFS2_FEATURE_RO_COMPAT_GRPQUOTA)
109
110/*
111 * Heartbeat-only devices are missing journals and other files. The
112 * filesystem driver can't load them, but the library can. Never put
113 * this in OCFS2_FEATURE_INCOMPAT_SUPP, *ever*.
114 */
115#define OCFS2_FEATURE_INCOMPAT_HEARTBEAT_DEV 0x0002
116
117/*
118 * tunefs sets this incompat flag before starting the resize and clears it
119 * at the end. This flag protects users from inadvertently mounting the fs
120 * after an aborted run without fsck-ing.
121 */
122#define OCFS2_FEATURE_INCOMPAT_RESIZE_INPROG 0x0004
123
124/* Used to denote a non-clustered volume */
125#define OCFS2_FEATURE_INCOMPAT_LOCAL_MOUNT 0x0008
126
127/* Support for sparse allocation in b-trees */
128#define OCFS2_FEATURE_INCOMPAT_SPARSE_ALLOC 0x0010
129
130/*
131 * Tunefs sets this incompat flag before starting an operation which
132 * would require cleanup on abort. This is done to protect users from
133 * inadvertently mounting the fs after an aborted run without
134 * fsck-ing.
135 *
136 * s_tunefs_flags on the super block describes precisely which
137 * operations were in progress.
138 */
139#define OCFS2_FEATURE_INCOMPAT_TUNEFS_INPROG 0x0020
140
141/* Support for data packed into inode blocks */
142#define OCFS2_FEATURE_INCOMPAT_INLINE_DATA 0x0040
143
144/*
145 * Support for alternate, userspace cluster stacks. If set, the superblock
146 * field s_cluster_info contains a tag for the alternate stack in use as
147 * well as the name of the cluster being joined.
148 * mount.ocfs2 must pass in a matching stack name.
149 *
150 * If not set, the classic stack will be used. This is compatbile with
151 * all older versions.
152 */
153#define OCFS2_FEATURE_INCOMPAT_USERSPACE_STACK 0x0080
154
155/* Support for the extended slot map */
156#define OCFS2_FEATURE_INCOMPAT_EXTENDED_SLOT_MAP 0x100
157
158/* Support for extended attributes */
159#define OCFS2_FEATURE_INCOMPAT_XATTR 0x0200
160
161/* Support for indexed directores */
162#define OCFS2_FEATURE_INCOMPAT_INDEXED_DIRS 0x0400
163
164/* Metadata checksum and error correction */
165#define OCFS2_FEATURE_INCOMPAT_META_ECC 0x0800
166
167/* Refcount tree support */
168#define OCFS2_FEATURE_INCOMPAT_REFCOUNT_TREE 0x1000
169
170/* Discontigous block groups */
171#define OCFS2_FEATURE_INCOMPAT_DISCONTIG_BG 0x2000
172
173/*
174 * Incompat bit to indicate useable clusterinfo with stackflags for all
175 * cluster stacks (userspace adnd o2cb). If this bit is set,
176 * INCOMPAT_USERSPACE_STACK becomes superfluous and thus should not be set.
177 */
178#define OCFS2_FEATURE_INCOMPAT_CLUSTERINFO 0x4000
179
180/*
181 * backup superblock flag is used to indicate that this volume
182 * has backup superblocks.
183 */
184#define OCFS2_FEATURE_COMPAT_BACKUP_SB 0x0001
185
186/*
187 * The filesystem will correctly handle journal feature bits.
188 */
189#define OCFS2_FEATURE_COMPAT_JBD2_SB 0x0002
190
191/*
192 * Unwritten extents support.
193 */
194#define OCFS2_FEATURE_RO_COMPAT_UNWRITTEN 0x0001
195
196/*
197 * Maintain quota information for this filesystem
198 */
199#define OCFS2_FEATURE_RO_COMPAT_USRQUOTA 0x0002
200#define OCFS2_FEATURE_RO_COMPAT_GRPQUOTA 0x0004
201
202/* The byte offset of the first backup block will be 1G.
203 * The following will be 4G, 16G, 64G, 256G and 1T.
204 */
205#define OCFS2_BACKUP_SB_START 1 << 30
206
207/* the max backup superblock nums */
208#define OCFS2_MAX_BACKUP_SUPERBLOCKS 6
209
210/*
211 * Flags on ocfs2_super_block.s_tunefs_flags
212 */
213#define OCFS2_TUNEFS_INPROG_REMOVE_SLOT 0x0001 /* Removing slots */
214
215/*
216 * Flags on ocfs2_dinode.i_flags
217 */
218#define OCFS2_VALID_FL (0x00000001) /* Inode is valid */
219#define OCFS2_UNUSED2_FL (0x00000002)
220#define OCFS2_ORPHANED_FL (0x00000004) /* On the orphan list */
221#define OCFS2_UNUSED3_FL (0x00000008)
222/* System inode flags */
223#define OCFS2_SYSTEM_FL (0x00000010) /* System inode */
224#define OCFS2_SUPER_BLOCK_FL (0x00000020) /* Super block */
225#define OCFS2_LOCAL_ALLOC_FL (0x00000040) /* Slot local alloc bitmap */
226#define OCFS2_BITMAP_FL (0x00000080) /* Allocation bitmap */
227#define OCFS2_JOURNAL_FL (0x00000100) /* Slot local journal */
228#define OCFS2_HEARTBEAT_FL (0x00000200) /* Heartbeat area */
229#define OCFS2_CHAIN_FL (0x00000400) /* Chain allocator */
230#define OCFS2_DEALLOC_FL (0x00000800) /* Truncate log */
231#define OCFS2_QUOTA_FL (0x00001000) /* Quota file */
232
233/*
234 * Flags on ocfs2_dinode.i_dyn_features
235 *
236 * These can change much more often than i_flags. When adding flags,
237 * keep in mind that i_dyn_features is only 16 bits wide.
238 */
239#define OCFS2_INLINE_DATA_FL (0x0001) /* Data stored in inode block */
240#define OCFS2_HAS_XATTR_FL (0x0002)
241#define OCFS2_INLINE_XATTR_FL (0x0004)
242#define OCFS2_INDEXED_DIR_FL (0x0008)
243#define OCFS2_HAS_REFCOUNT_FL (0x0010)
244
245/* Inode attributes, keep in sync with EXT2 */
246#define OCFS2_SECRM_FL FS_SECRM_FL /* Secure deletion */
247#define OCFS2_UNRM_FL FS_UNRM_FL /* Undelete */
248#define OCFS2_COMPR_FL FS_COMPR_FL /* Compress file */
249#define OCFS2_SYNC_FL FS_SYNC_FL /* Synchronous updates */
250#define OCFS2_IMMUTABLE_FL FS_IMMUTABLE_FL /* Immutable file */
251#define OCFS2_APPEND_FL FS_APPEND_FL /* writes to file may only append */
252#define OCFS2_NODUMP_FL FS_NODUMP_FL /* do not dump file */
253#define OCFS2_NOATIME_FL FS_NOATIME_FL /* do not update atime */
254/* Reserved for compression usage... */
255#define OCFS2_DIRTY_FL FS_DIRTY_FL
256#define OCFS2_COMPRBLK_FL FS_COMPRBLK_FL /* One or more compressed clusters */
257#define OCFS2_NOCOMP_FL FS_NOCOMP_FL /* Don't compress */
258#define OCFS2_ECOMPR_FL FS_ECOMPR_FL /* Compression error */
259/* End compression flags --- maybe not all used */
260#define OCFS2_BTREE_FL FS_BTREE_FL /* btree format dir */
261#define OCFS2_INDEX_FL FS_INDEX_FL /* hash-indexed directory */
262#define OCFS2_IMAGIC_FL FS_IMAGIC_FL /* AFS directory */
263#define OCFS2_JOURNAL_DATA_FL FS_JOURNAL_DATA_FL /* Reserved for ext3 */
264#define OCFS2_NOTAIL_FL FS_NOTAIL_FL /* file tail should not be merged */
265#define OCFS2_DIRSYNC_FL FS_DIRSYNC_FL /* dirsync behaviour (directories only) */
266#define OCFS2_TOPDIR_FL FS_TOPDIR_FL /* Top of directory hierarchies*/
267#define OCFS2_RESERVED_FL FS_RESERVED_FL /* reserved for ext2 lib */
268
269#define OCFS2_FL_VISIBLE FS_FL_USER_VISIBLE /* User visible flags */
270#define OCFS2_FL_MODIFIABLE FS_FL_USER_MODIFIABLE /* User modifiable flags */
271
272/*
273 * Extent record flags (e_node.leaf.flags)
274 */
275#define OCFS2_EXT_UNWRITTEN (0x01) /* Extent is allocated but
276                         * unwritten */
277#define OCFS2_EXT_REFCOUNTED (0x02) /* Extent is reference
278                         * counted in an associated
279                         * refcount tree */
280
281/*
282 * Journal Flags (ocfs2_dinode.id1.journal1.i_flags)
283 */
284#define OCFS2_JOURNAL_DIRTY_FL (0x00000001) /* Journal needs recovery */
285
286/*
287 * superblock s_state flags
288 */
289#define OCFS2_ERROR_FS (0x00000001) /* FS saw errors */
290
291/* Limit of space in ocfs2_dir_entry */
292#define OCFS2_MAX_FILENAME_LEN 255
293
294/* Maximum slots on an ocfs2 file system */
295#define OCFS2_MAX_SLOTS 255
296
297/* Slot map indicator for an empty slot */
298#define OCFS2_INVALID_SLOT -1
299
300#define OCFS2_VOL_UUID_LEN 16
301#define OCFS2_MAX_VOL_LABEL_LEN 64
302
303/* The cluster stack fields */
304#define OCFS2_STACK_LABEL_LEN 4
305#define OCFS2_CLUSTER_NAME_LEN 16
306
307/* Classic (historically speaking) cluster stack */
308#define OCFS2_CLASSIC_CLUSTER_STACK "o2cb"
309
310/* Journal limits (in bytes) */
311#define OCFS2_MIN_JOURNAL_SIZE (4 * 1024 * 1024)
312
313/*
314 * Inline extended attribute size (in bytes)
315 * The value chosen should be aligned to 16 byte boundaries.
316 */
317#define OCFS2_MIN_XATTR_INLINE_SIZE 256
318
319/*
320 * Cluster info flags (ocfs2_cluster_info.ci_stackflags)
321 */
322#define OCFS2_CLUSTER_O2CB_GLOBAL_HEARTBEAT (0x01)
323
324struct ocfs2_system_inode_info {
325    char *si_name;
326    int si_iflags;
327    int si_mode;
328};
329
330/* System file index */
331enum {
332    BAD_BLOCK_SYSTEM_INODE = 0,
333    GLOBAL_INODE_ALLOC_SYSTEM_INODE,
334    SLOT_MAP_SYSTEM_INODE,
335#define OCFS2_FIRST_ONLINE_SYSTEM_INODE SLOT_MAP_SYSTEM_INODE
336    HEARTBEAT_SYSTEM_INODE,
337    GLOBAL_BITMAP_SYSTEM_INODE,
338    USER_QUOTA_SYSTEM_INODE,
339    GROUP_QUOTA_SYSTEM_INODE,
340#define OCFS2_LAST_GLOBAL_SYSTEM_INODE GROUP_QUOTA_SYSTEM_INODE
341#define OCFS2_FIRST_LOCAL_SYSTEM_INODE ORPHAN_DIR_SYSTEM_INODE
342    ORPHAN_DIR_SYSTEM_INODE,
343    EXTENT_ALLOC_SYSTEM_INODE,
344    INODE_ALLOC_SYSTEM_INODE,
345    JOURNAL_SYSTEM_INODE,
346    LOCAL_ALLOC_SYSTEM_INODE,
347    TRUNCATE_LOG_SYSTEM_INODE,
348    LOCAL_USER_QUOTA_SYSTEM_INODE,
349    LOCAL_GROUP_QUOTA_SYSTEM_INODE,
350#define OCFS2_LAST_LOCAL_SYSTEM_INODE LOCAL_GROUP_QUOTA_SYSTEM_INODE
351    NUM_SYSTEM_INODES
352};
353#define NUM_GLOBAL_SYSTEM_INODES OCFS2_FIRST_LOCAL_SYSTEM_INODE
354#define NUM_LOCAL_SYSTEM_INODES \
355        (NUM_SYSTEM_INODES - OCFS2_FIRST_LOCAL_SYSTEM_INODE)
356
357static struct ocfs2_system_inode_info ocfs2_system_inodes[NUM_SYSTEM_INODES] = {
358    /* Global system inodes (single copy) */
359    /* The first two are only used from userspace mfks/tunefs */
360    [BAD_BLOCK_SYSTEM_INODE] = { "bad_blocks", 0, S_IFREG | 0644 },
361    [GLOBAL_INODE_ALLOC_SYSTEM_INODE] = { "global_inode_alloc", OCFS2_BITMAP_FL | OCFS2_CHAIN_FL, S_IFREG | 0644 },
362
363    /* These are used by the running filesystem */
364    [SLOT_MAP_SYSTEM_INODE] = { "slot_map", 0, S_IFREG | 0644 },
365    [HEARTBEAT_SYSTEM_INODE] = { "heartbeat", OCFS2_HEARTBEAT_FL, S_IFREG | 0644 },
366    [GLOBAL_BITMAP_SYSTEM_INODE] = { "global_bitmap", 0, S_IFREG | 0644 },
367    [USER_QUOTA_SYSTEM_INODE] = { "aquota.user", OCFS2_QUOTA_FL, S_IFREG | 0644 },
368    [GROUP_QUOTA_SYSTEM_INODE] = { "aquota.group", OCFS2_QUOTA_FL, S_IFREG | 0644 },
369
370    /* Slot-specific system inodes (one copy per slot) */
371    [ORPHAN_DIR_SYSTEM_INODE] = { "orphan_dir:%04d", 0, S_IFDIR | 0755 },
372    [EXTENT_ALLOC_SYSTEM_INODE] = { "extent_alloc:%04d", OCFS2_BITMAP_FL | OCFS2_CHAIN_FL, S_IFREG | 0644 },
373    [INODE_ALLOC_SYSTEM_INODE] = { "inode_alloc:%04d", OCFS2_BITMAP_FL | OCFS2_CHAIN_FL, S_IFREG | 0644 },
374    [JOURNAL_SYSTEM_INODE] = { "journal:%04d", OCFS2_JOURNAL_FL, S_IFREG | 0644 },
375    [LOCAL_ALLOC_SYSTEM_INODE] = { "local_alloc:%04d", OCFS2_BITMAP_FL | OCFS2_LOCAL_ALLOC_FL, S_IFREG | 0644 },
376    [TRUNCATE_LOG_SYSTEM_INODE] = { "truncate_log:%04d", OCFS2_DEALLOC_FL, S_IFREG | 0644 },
377    [LOCAL_USER_QUOTA_SYSTEM_INODE] = { "aquota.user:%04d", OCFS2_QUOTA_FL, S_IFREG | 0644 },
378    [LOCAL_GROUP_QUOTA_SYSTEM_INODE] = { "aquota.group:%04d", OCFS2_QUOTA_FL, S_IFREG | 0644 },
379};
380
381/* Parameter passed from mount.ocfs2 to module */
382#define OCFS2_HB_NONE "heartbeat=none"
383#define OCFS2_HB_LOCAL "heartbeat=local"
384#define OCFS2_HB_GLOBAL "heartbeat=global"
385
386/*
387 * OCFS2 directory file types. Only the low 3 bits are used. The
388 * other bits are reserved for now.
389 */
390#define OCFS2_FT_UNKNOWN 0
391#define OCFS2_FT_REG_FILE 1
392#define OCFS2_FT_DIR 2
393#define OCFS2_FT_CHRDEV 3
394#define OCFS2_FT_BLKDEV 4
395#define OCFS2_FT_FIFO 5
396#define OCFS2_FT_SOCK 6
397#define OCFS2_FT_SYMLINK 7
398
399#define OCFS2_FT_MAX 8
400
401/*
402 * OCFS2_DIR_PAD defines the directory entries boundaries
403 *
404 * NOTE: It must be a multiple of 4
405 */
406#define OCFS2_DIR_PAD 4
407#define OCFS2_DIR_ROUND (OCFS2_DIR_PAD - 1)
408#define OCFS2_DIR_MEMBER_LEN offsetof(struct ocfs2_dir_entry, name)
409#define OCFS2_DIR_REC_LEN(name_len) (((name_len) + OCFS2_DIR_MEMBER_LEN + \
410                                          OCFS2_DIR_ROUND) & \
411                     ~OCFS2_DIR_ROUND)
412#define OCFS2_DIR_MIN_REC_LEN OCFS2_DIR_REC_LEN(1)
413
414#define OCFS2_LINK_MAX 32000
415#define OCFS2_DX_LINK_MAX ((1U << 31) - 1U)
416#define OCFS2_LINKS_HI_SHIFT 16
417#define OCFS2_DX_ENTRIES_MAX (0xffffffffU)
418
419#define S_SHIFT 12
420static unsigned char ocfs2_type_by_mode[S_IFMT >> S_SHIFT] = {
421    [S_IFREG >> S_SHIFT] = OCFS2_FT_REG_FILE,
422    [S_IFDIR >> S_SHIFT] = OCFS2_FT_DIR,
423    [S_IFCHR >> S_SHIFT] = OCFS2_FT_CHRDEV,
424    [S_IFBLK >> S_SHIFT] = OCFS2_FT_BLKDEV,
425    [S_IFIFO >> S_SHIFT] = OCFS2_FT_FIFO,
426    [S_IFSOCK >> S_SHIFT] = OCFS2_FT_SOCK,
427    [S_IFLNK >> S_SHIFT] = OCFS2_FT_SYMLINK,
428};
429
430
431/*
432 * Convenience casts
433 */
434#define OCFS2_RAW_SB(dinode) (&((dinode)->id2.i_super))
435
436/*
437 * Block checking structure. This is used in metadata to validate the
438 * contents. If OCFS2_FEATURE_INCOMPAT_META_ECC is not set, it is all
439 * zeros.
440 */
441struct ocfs2_block_check {
442/*00*/ __le32 bc_crc32e; /* 802.3 Ethernet II CRC32 */
443    __le16 bc_ecc; /* Single-error-correction parity vector.
444                   This is a simple Hamming code dependent
445                   on the blocksize. OCFS2's maximum
446                   blocksize, 4K, requires 16 parity bits,
447                   so we fit in __le16. */
448    __le16 bc_reserved1;
449/*08*/
450};
451
452/*
453 * On disk extent record for OCFS2
454 * It describes a range of clusters on disk.
455 *
456 * Length fields are divided into interior and leaf node versions.
457 * This leaves room for a flags field (OCFS2_EXT_*) in the leaf nodes.
458 */
459struct ocfs2_extent_rec {
460/*00*/ __le32 e_cpos; /* Offset into the file, in clusters */
461    union {
462        __le32 e_int_clusters; /* Clusters covered by all children */
463        struct {
464            __le16 e_leaf_clusters; /* Clusters covered by this
465                           extent */
466            __u8 e_reserved1;
467            __u8 e_flags; /* Extent flags */
468        };
469    };
470    __le64 e_blkno; /* Physical disk offset, in blocks */
471/*10*/
472};
473
474struct ocfs2_chain_rec {
475    __le32 c_free; /* Number of free bits in this chain. */
476    __le32 c_total; /* Number of total bits in this chain */
477    __le64 c_blkno; /* Physical disk offset (blocks) of 1st group */
478};
479
480struct ocfs2_truncate_rec {
481    __le32 t_start; /* 1st cluster in this log */
482    __le32 t_clusters; /* Number of total clusters covered */
483};
484
485/*
486 * On disk extent list for OCFS2 (node in the tree). Note that this
487 * is contained inside ocfs2_dinode or ocfs2_extent_block, so the
488 * offsets are relative to ocfs2_dinode.id2.i_list or
489 * ocfs2_extent_block.h_list, respectively.
490 */
491struct ocfs2_extent_list {
492/*00*/ __le16 l_tree_depth; /* Extent tree depth from this
493                       point. 0 means data extents
494                       hang directly off this
495                       header (a leaf)
496                       NOTE: The high 8 bits cannot be
497                       used - tree_depth is never that big.
498                    */
499    __le16 l_count; /* Number of extent records */
500    __le16 l_next_free_rec; /* Next unused extent slot */
501    __le16 l_reserved1;
502    __le64 l_reserved2; /* Pad to
503                       sizeof(ocfs2_extent_rec) */
504/*10*/ struct ocfs2_extent_rec l_recs[0]; /* Extent records */
505};
506
507/*
508 * On disk allocation chain list for OCFS2. Note that this is
509 * contained inside ocfs2_dinode, so the offsets are relative to
510 * ocfs2_dinode.id2.i_chain.
511 */
512struct ocfs2_chain_list {
513/*00*/ __le16 cl_cpg; /* Clusters per Block Group */
514    __le16 cl_bpc; /* Bits per cluster */
515    __le16 cl_count; /* Total chains in this list */
516    __le16 cl_next_free_rec; /* Next unused chain slot */
517    __le64 cl_reserved1;
518/*10*/ struct ocfs2_chain_rec cl_recs[0]; /* Chain records */
519};
520
521/*
522 * On disk deallocation log for OCFS2. Note that this is
523 * contained inside ocfs2_dinode, so the offsets are relative to
524 * ocfs2_dinode.id2.i_dealloc.
525 */
526struct ocfs2_truncate_log {
527/*00*/ __le16 tl_count; /* Total records in this log */
528    __le16 tl_used; /* Number of records in use */
529    __le32 tl_reserved1;
530/*08*/ struct ocfs2_truncate_rec tl_recs[0]; /* Truncate records */
531};
532
533/*
534 * On disk extent block (indirect block) for OCFS2
535 */
536struct ocfs2_extent_block
537{
538/*00*/ __u8 h_signature[8]; /* Signature for verification */
539    struct ocfs2_block_check h_check; /* Error checking */
540/*10*/ __le16 h_suballoc_slot; /* Slot suballocator this
541                       extent_header belongs to */
542    __le16 h_suballoc_bit; /* Bit offset in suballocator
543                       block group */
544    __le32 h_fs_generation; /* Must match super block */
545    __le64 h_blkno; /* Offset on disk, in blocks */
546/*20*/ __le64 h_suballoc_loc; /* Suballocator block group this
547                       eb belongs to. Only valid
548                       if allocated from a
549                       discontiguous block group */
550    __le64 h_next_leaf_blk; /* Offset on disk, in blocks,
551                       of next leaf header pointing
552                       to data */
553/*30*/ struct ocfs2_extent_list h_list; /* Extent record list */
554/* Actual on-disk size is one block */
555};
556
557/*
558 * On disk slot map for OCFS2. This defines the contents of the "slot_map"
559 * system file. A slot is valid if it contains a node number >= 0. The
560 * value -1 (0xFFFF) is OCFS2_INVALID_SLOT. This marks a slot empty.
561 */
562struct ocfs2_slot_map {
563/*00*/ __le16 sm_slots[0];
564/*
565 * Actual on-disk size is one block. OCFS2_MAX_SLOTS is 255,
566 * 255 * sizeof(__le16) == 512B, within the 512B block minimum blocksize.
567 */
568};
569
570struct ocfs2_extended_slot {
571/*00*/ __u8 es_valid;
572    __u8 es_reserved1[3];
573    __le32 es_node_num;
574/*10*/
575};
576
577/*
578 * The extended slot map, used when OCFS2_FEATURE_INCOMPAT_EXTENDED_SLOT_MAP
579 * is set. It separates out the valid marker from the node number, and
580 * has room to grow. Unlike the old slot map, this format is defined by
581 * i_size.
582 */
583struct ocfs2_slot_map_extended {
584/*00*/ struct ocfs2_extended_slot se_slots[0];
585/*
586 * Actual size is i_size of the slot_map system file. It should
587 * match s_max_slots * sizeof(struct ocfs2_extended_slot)
588 */
589};
590
591/*
592 * ci_stackflags is only valid if the incompat bit
593 * OCFS2_FEATURE_INCOMPAT_CLUSTERINFO is set.
594 */
595struct ocfs2_cluster_info {
596/*00*/ __u8 ci_stack[OCFS2_STACK_LABEL_LEN];
597    union {
598        __le32 ci_reserved;
599        struct {
600            __u8 ci_stackflags;
601            __u8 ci_reserved1;
602            __u8 ci_reserved2;
603            __u8 ci_reserved3;
604        };
605    };
606/*08*/ __u8 ci_cluster[OCFS2_CLUSTER_NAME_LEN];
607/*18*/
608};
609
610/*
611 * On disk superblock for OCFS2
612 * Note that it is contained inside an ocfs2_dinode, so all offsets
613 * are relative to the start of ocfs2_dinode.id2.
614 */
615struct ocfs2_super_block {
616/*00*/ __le16 s_major_rev_level;
617    __le16 s_minor_rev_level;
618    __le16 s_mnt_count;
619    __le16 s_max_mnt_count;
620    __le16 s_state; /* File system state */
621    __le16 s_errors; /* Behaviour when detecting errors */
622    __le32 s_checkinterval; /* Max time between checks */
623/*10*/ __le64 s_lastcheck; /* Time of last check */
624    __le32 s_creator_os; /* OS */
625    __le32 s_feature_compat; /* Compatible feature set */
626/*20*/ __le32 s_feature_incompat; /* Incompatible feature set */
627    __le32 s_feature_ro_compat; /* Readonly-compatible feature set */
628    __le64 s_root_blkno; /* Offset, in blocks, of root directory
629                       dinode */
630/*30*/ __le64 s_system_dir_blkno; /* Offset, in blocks, of system
631                       directory dinode */
632    __le32 s_blocksize_bits; /* Blocksize for this fs */
633    __le32 s_clustersize_bits; /* Clustersize for this fs */
634/*40*/ __le16 s_max_slots; /* Max number of simultaneous mounts
635                       before tunefs required */
636    __le16 s_tunefs_flag;
637    __le32 s_uuid_hash; /* hash value of uuid */
638    __le64 s_first_cluster_group; /* Block offset of 1st cluster
639                     * group header */
640/*50*/ __u8 s_label[OCFS2_MAX_VOL_LABEL_LEN]; /* Label for mounting, etc. */
641/*90*/ __u8 s_uuid[OCFS2_VOL_UUID_LEN]; /* 128-bit uuid */
642/*A0*/ struct ocfs2_cluster_info s_cluster_info; /* Only valid if either
643                             userspace or clusterinfo
644                             INCOMPAT flag set. */
645/*B8*/ __le16 s_xattr_inline_size; /* extended attribute inline size
646                       for this fs*/
647    __le16 s_reserved0;
648    __le32 s_dx_seed[3]; /* seed[0-2] for dx dir hash.
649                     * s_uuid_hash serves as seed[3]. */
650/*C0*/ __le64 s_reserved2[15]; /* Fill out superblock */
651/*140*/
652
653    /*
654     * NOTE: As stated above, all offsets are relative to
655     * ocfs2_dinode.id2, which is at 0xC0 in the inode.
656     * 0xC0 + 0x140 = 0x200 or 512 bytes. A superblock must fit within
657     * our smallest blocksize, which is 512 bytes. To ensure this,
658     * we reserve the space in s_reserved2. Anything past s_reserved2
659     * will not be available on the smallest blocksize.
660     */
661};
662
663/*
664 * Local allocation bitmap for OCFS2 slots
665 * Note that it exists inside an ocfs2_dinode, so all offsets are
666 * relative to the start of ocfs2_dinode.id2.
667 */
668struct ocfs2_local_alloc
669{
670/*00*/ __le32 la_bm_off; /* Starting bit offset in main bitmap */
671    __le16 la_size; /* Size of included bitmap, in bytes */
672    __le16 la_reserved1;
673    __le64 la_reserved2;
674/*10*/ __u8 la_bitmap[0];
675};
676
677/*
678 * Data-in-inode header. This is only used if i_dyn_features has
679 * OCFS2_INLINE_DATA_FL set.
680 */
681struct ocfs2_inline_data
682{
683/*00*/ __le16 id_count; /* Number of bytes that can be used
684                 * for data, starting at id_data */
685    __le16 id_reserved0;
686    __le32 id_reserved1;
687    __u8 id_data[0]; /* Start of user data */
688};
689
690/*
691 * On disk inode for OCFS2
692 */
693struct ocfs2_dinode {
694/*00*/ __u8 i_signature[8]; /* Signature for validation */
695    __le32 i_generation; /* Generation number */
696    __le16 i_suballoc_slot; /* Slot suballocator this inode
697                       belongs to */
698    __le16 i_suballoc_bit; /* Bit offset in suballocator
699                       block group */
700/*10*/ __le16 i_links_count_hi; /* High 16 bits of links count */
701    __le16 i_xattr_inline_size;
702    __le32 i_clusters; /* Cluster count */
703    __le32 i_uid; /* Owner UID */
704    __le32 i_gid; /* Owning GID */
705/*20*/ __le64 i_size; /* Size in bytes */
706    __le16 i_mode; /* File mode */
707    __le16 i_links_count; /* Links count */
708    __le32 i_flags; /* File flags */
709/*30*/ __le64 i_atime; /* Access time */
710    __le64 i_ctime; /* Creation time */
711/*40*/ __le64 i_mtime; /* Modification time */
712    __le64 i_dtime; /* Deletion time */
713/*50*/ __le64 i_blkno; /* Offset on disk, in blocks */
714    __le64 i_last_eb_blk; /* Pointer to last extent
715                       block */
716/*60*/ __le32 i_fs_generation; /* Generation per fs-instance */
717    __le32 i_atime_nsec;
718    __le32 i_ctime_nsec;
719    __le32 i_mtime_nsec;
720/*70*/ __le32 i_attr;
721    __le16 i_orphaned_slot; /* Only valid when OCFS2_ORPHANED_FL
722                       was set in i_flags */
723    __le16 i_dyn_features;
724    __le64 i_xattr_loc;
725/*80*/ struct ocfs2_block_check i_check; /* Error checking */
726/*88*/ __le64 i_dx_root; /* Pointer to dir index root block */
727/*90*/ __le64 i_refcount_loc;
728    __le64 i_suballoc_loc; /* Suballocator block group this
729                       inode belongs to. Only valid
730                       if allocated from a
731                       discontiguous block group */
732/*A0*/ __le64 i_reserved2[3];
733/*B8*/ union {
734        __le64 i_pad1; /* Generic way to refer to this
735                       64bit union */
736        struct {
737            __le64 i_rdev; /* Device number */
738        } dev1;
739        struct { /* Info for bitmap system
740                       inodes */
741            __le32 i_used; /* Bits (ie, clusters) used */
742            __le32 i_total; /* Total bits (clusters)
743                       available */
744        } bitmap1;
745        struct { /* Info for journal system
746                       inodes */
747            __le32 ij_flags; /* Mounted, version, etc. */
748            __le32 ij_recovery_generation; /* Incremented when the
749                              journal is recovered
750                              after an unclean
751                              shutdown */
752        } journal1;
753    } id1; /* Inode type dependent 1 */
754/*C0*/ union {
755        struct ocfs2_super_block i_super;
756        struct ocfs2_local_alloc i_lab;
757        struct ocfs2_chain_list i_chain;
758        struct ocfs2_extent_list i_list;
759        struct ocfs2_truncate_log i_dealloc;
760        struct ocfs2_inline_data i_data;
761        __u8 i_symlink[0];
762    } id2;
763/* Actual on-disk size is one block */
764};
765
766/*
767 * On-disk directory entry structure for OCFS2
768 *
769 * Packed as this structure could be accessed unaligned on 64-bit platforms
770 */
771struct ocfs2_dir_entry {
772/*00*/ __le64 inode; /* Inode number */
773    __le16 rec_len; /* Directory entry length */
774    __u8 name_len; /* Name length */
775    __u8 file_type;
776/*0C*/ char name[OCFS2_MAX_FILENAME_LEN]; /* File name */
777/* Actual on-disk length specified by rec_len */
778} __attribute__ ((packed));
779
780/*
781 * Per-block record for the unindexed directory btree. This is carefully
782 * crafted so that the rec_len and name_len records of an ocfs2_dir_entry are
783 * mirrored. That way, the directory manipulation code needs a minimal amount
784 * of update.
785 *
786 * NOTE: Keep this structure aligned to a multiple of 4 bytes.
787 */
788struct ocfs2_dir_block_trailer {
789/*00*/ __le64 db_compat_inode; /* Always zero. Was inode */
790
791    __le16 db_compat_rec_len; /* Backwards compatible with
792                         * ocfs2_dir_entry. */
793    __u8 db_compat_name_len; /* Always zero. Was name_len */
794    __u8 db_reserved0;
795    __le16 db_reserved1;
796    __le16 db_free_rec_len; /* Size of largest empty hole
797                         * in this block. (unused) */
798/*10*/ __u8 db_signature[8]; /* Signature for verification */
799    __le64 db_reserved2;
800    __le64 db_free_next; /* Next block in list (unused) */
801/*20*/ __le64 db_blkno; /* Offset on disk, in blocks */
802    __le64 db_parent_dinode; /* dinode which owns me, in
803                           blocks */
804/*30*/ struct ocfs2_block_check db_check; /* Error checking */
805/*40*/
806};
807
808 /*
809 * A directory entry in the indexed tree. We don't store the full name here,
810 * but instead provide a pointer to the full dirent in the unindexed tree.
811 *
812 * We also store name_len here so as to reduce the number of leaf blocks we
813 * need to search in case of collisions.
814 */
815struct ocfs2_dx_entry {
816    __le32 dx_major_hash; /* Used to find logical
817                     * cluster in index */
818    __le32 dx_minor_hash; /* Lower bits used to find
819                     * block in cluster */
820    __le64 dx_dirent_blk; /* Physical block in unindexed
821                     * tree holding this dirent. */
822};
823
824struct ocfs2_dx_entry_list {
825    __le32 de_reserved;
826    __le16 de_count; /* Maximum number of entries
827                     * possible in de_entries */
828    __le16 de_num_used; /* Current number of
829                     * de_entries entries */
830    struct ocfs2_dx_entry de_entries[0]; /* Indexed dir entries
831                             * in a packed array of
832                             * length de_num_used */
833};
834
835#define OCFS2_DX_FLAG_INLINE 0x01
836
837/*
838 * A directory indexing block. Each indexed directory has one of these,
839 * pointed to by ocfs2_dinode.
840 *
841 * This block stores an indexed btree root, and a set of free space
842 * start-of-list pointers.
843 */
844struct ocfs2_dx_root_block {
845    __u8 dr_signature[8]; /* Signature for verification */
846    struct ocfs2_block_check dr_check; /* Error checking */
847    __le16 dr_suballoc_slot; /* Slot suballocator this
848                         * block belongs to. */
849    __le16 dr_suballoc_bit; /* Bit offset in suballocator
850                         * block group */
851    __le32 dr_fs_generation; /* Must match super block */
852    __le64 dr_blkno; /* Offset on disk, in blocks */
853    __le64 dr_last_eb_blk; /* Pointer to last
854                         * extent block */
855    __le32 dr_clusters; /* Clusters allocated
856                         * to the indexed tree. */
857    __u8 dr_flags; /* OCFS2_DX_FLAG_* flags */
858    __u8 dr_reserved0;
859    __le16 dr_reserved1;
860    __le64 dr_dir_blkno; /* Pointer to parent inode */
861    __le32 dr_num_entries; /* Total number of
862                         * names stored in
863                         * this directory.*/
864    __le32 dr_reserved2;
865    __le64 dr_free_blk; /* Pointer to head of free
866                         * unindexed block list. */
867    __le64 dr_suballoc_loc; /* Suballocator block group
868                           this root belongs to.
869                           Only valid if allocated
870                           from a discontiguous
871                           block group */
872    __le64 dr_reserved3[14];
873    union {
874        struct ocfs2_extent_list dr_list; /* Keep this aligned to 128
875                           * bits for maximum space
876                           * efficiency. */
877        struct ocfs2_dx_entry_list dr_entries; /* In-root-block list of
878                            * entries. We grow out
879                            * to extents if this
880                            * gets too big. */
881    };
882};
883
884/*
885 * The header of a leaf block in the indexed tree.
886 */
887struct ocfs2_dx_leaf {
888    __u8 dl_signature[8];/* Signature for verification */
889    struct ocfs2_block_check dl_check; /* Error checking */
890    __le64 dl_blkno; /* Offset on disk, in blocks */
891    __le32 dl_fs_generation;/* Must match super block */
892    __le32 dl_reserved0;
893    __le64 dl_reserved1;
894    struct ocfs2_dx_entry_list dl_list;
895};
896
897/*
898 * Largest bitmap for a block (suballocator) group in bytes. This limit
899 * does not affect cluster groups (global allocator). Cluster group
900 * bitmaps run to the end of the block.
901 */
902#define OCFS2_MAX_BG_BITMAP_SIZE 256
903
904/*
905 * On disk allocator group structure for OCFS2
906 */
907struct ocfs2_group_desc
908{
909/*00*/ __u8 bg_signature[8]; /* Signature for validation */
910    __le16 bg_size; /* Size of included bitmap in
911                       bytes. */
912    __le16 bg_bits; /* Bits represented by this
913                       group. */
914    __le16 bg_free_bits_count; /* Free bits count */
915    __le16 bg_chain; /* What chain I am in. */
916/*10*/ __le32 bg_generation;
917    __le32 bg_reserved1;
918    __le64 bg_next_group; /* Next group in my list, in
919                       blocks */
920/*20*/ __le64 bg_parent_dinode; /* dinode which owns me, in
921                       blocks */
922    __le64 bg_blkno; /* Offset on disk, in blocks */
923/*30*/ struct ocfs2_block_check bg_check; /* Error checking */
924    __le64 bg_reserved2;
925/*40*/ union {
926        __u8 bg_bitmap[0];
927        struct {
928            /*
929             * Block groups may be discontiguous when
930             * OCFS2_FEATURE_INCOMPAT_DISCONTIG_BG is set.
931             * The extents of a discontigous block group are
932             * stored in bg_list. It is a flat list.
933             * l_tree_depth must always be zero. A
934             * discontiguous group is signified by a non-zero
935             * bg_list->l_next_free_rec. Only block groups
936             * can be discontiguous; Cluster groups cannot.
937             * We've never made a block group with more than
938             * 2048 blocks (256 bytes of bg_bitmap). This
939             * codifies that limit so that we can fit bg_list.
940             * bg_size of a discontiguous block group will
941             * be 256 to match bg_bitmap_filler.
942             */
943            __u8 bg_bitmap_filler[OCFS2_MAX_BG_BITMAP_SIZE];
944/*140*/ struct ocfs2_extent_list bg_list;
945        };
946    };
947/* Actual on-disk size is one block */
948};
949
950struct ocfs2_refcount_rec {
951/*00*/ __le64 r_cpos; /* Physical offset, in clusters */
952    __le32 r_clusters; /* Clusters covered by this extent */
953    __le32 r_refcount; /* Reference count of this extent */
954/*10*/
955};
956#define OCFS2_32BIT_POS_MASK (0xffffffffULL)
957
958#define OCFS2_REFCOUNT_LEAF_FL (0x00000001)
959#define OCFS2_REFCOUNT_TREE_FL (0x00000002)
960
961struct ocfs2_refcount_list {
962/*00*/ __le16 rl_count; /* Maximum number of entries possible
963                   in rl_records */
964    __le16 rl_used; /* Current number of used records */
965    __le32 rl_reserved2;
966    __le64 rl_reserved1; /* Pad to sizeof(ocfs2_refcount_record) */
967/*10*/ struct ocfs2_refcount_rec rl_recs[0]; /* Refcount records */
968};
969
970
971struct ocfs2_refcount_block {
972/*00*/ __u8 rf_signature[8]; /* Signature for verification */
973    __le16 rf_suballoc_slot; /* Slot suballocator this block
974                       belongs to */
975    __le16 rf_suballoc_bit; /* Bit offset in suballocator
976                       block group */
977    __le32 rf_fs_generation; /* Must match superblock */
978/*10*/ __le64 rf_blkno; /* Offset on disk, in blocks */
979    __le64 rf_parent; /* Parent block, only valid if
980                       OCFS2_REFCOUNT_LEAF_FL is set in
981                       rf_flags */
982/*20*/ struct ocfs2_block_check rf_check; /* Error checking */
983    __le64 rf_last_eb_blk; /* Pointer to last extent block */
984/*30*/ __le32 rf_count; /* Number of inodes sharing this
985                       refcount tree */
986    __le32 rf_flags; /* See the flags above */
987    __le32 rf_clusters; /* clusters covered by refcount tree. */
988    __le32 rf_cpos; /* cluster offset in refcount tree.*/
989/*40*/ __le32 rf_generation; /* generation number. all be the same
990                     * for the same refcount tree. */
991    __le32 rf_reserved0;
992    __le64 rf_suballoc_loc; /* Suballocator block group this
993                       refcount block belongs to. Only
994                       valid if allocated from a
995                       discontiguous block group */
996/*50*/ __le64 rf_reserved1[6];
997/*80*/ union {
998        struct ocfs2_refcount_list rf_records; /* List of refcount
999                              records */
1000        struct ocfs2_extent_list rf_list; /* Extent record list,
1001                            only valid if
1002                            OCFS2_REFCOUNT_TREE_FL
1003                            is set in rf_flags */
1004    };
1005/* Actual on-disk size is one block */
1006};
1007
1008/*
1009 * On disk extended attribute structure for OCFS2.
1010 */
1011
1012/*
1013 * ocfs2_xattr_entry indicates one extend attribute.
1014 *
1015 * Note that it can be stored in inode, one block or one xattr bucket.
1016 */
1017struct ocfs2_xattr_entry {
1018    __le32 xe_name_hash; /* hash value of xattr prefix+suffix. */
1019    __le16 xe_name_offset; /* byte offset from the 1st entry in the
1020                    local xattr storage(inode, xattr block or
1021                    xattr bucket). */
1022    __u8 xe_name_len; /* xattr name len, doesn't include prefix. */
1023    __u8 xe_type; /* the low 7 bits indicate the name prefix
1024                  * type and the highest bit indicates whether
1025                  * the EA is stored in the local storage. */
1026    __le64 xe_value_size; /* real xattr value length. */
1027};
1028
1029/*
1030 * On disk structure for xattr header.
1031 *
1032 * One ocfs2_xattr_header describes how many ocfs2_xattr_entry records in
1033 * the local xattr storage.
1034 */
1035struct ocfs2_xattr_header {
1036    __le16 xh_count; /* contains the count of how
1037                           many records are in the
1038                           local xattr storage. */
1039    __le16 xh_free_start; /* current offset for storing
1040                           xattr. */
1041    __le16 xh_name_value_len; /* total length of name/value
1042                           length in this bucket. */
1043    __le16 xh_num_buckets; /* Number of xattr buckets
1044                           in this extent record,
1045                           only valid in the first
1046                           bucket. */
1047    struct ocfs2_block_check xh_check; /* Error checking
1048                           (Note, this is only
1049                            used for xattr
1050                            buckets. A block uses
1051                            xb_check and sets
1052                            this field to zero.) */
1053    struct ocfs2_xattr_entry xh_entries[0]; /* xattr entry list. */
1054};
1055
1056/*
1057 * On disk structure for xattr value root.
1058 *
1059 * When an xattr's value is large enough, it is stored in an external
1060 * b-tree like file data. The xattr value root points to this structure.
1061 */
1062struct ocfs2_xattr_value_root {
1063/*00*/ __le32 xr_clusters; /* clusters covered by xattr value. */
1064    __le32 xr_reserved0;
1065    __le64 xr_last_eb_blk; /* Pointer to last extent block */
1066/*10*/ struct ocfs2_extent_list xr_list; /* Extent record list */
1067};
1068
1069/*
1070 * On disk structure for xattr tree root.
1071 *
1072 * It is used when there are too many extended attributes for one file. These
1073 * attributes will be organized and stored in an indexed-btree.
1074 */
1075struct ocfs2_xattr_tree_root {
1076/*00*/ __le32 xt_clusters; /* clusters covered by xattr. */
1077    __le32 xt_reserved0;
1078    __le64 xt_last_eb_blk; /* Pointer to last extent block */
1079/*10*/ struct ocfs2_extent_list xt_list; /* Extent record list */
1080};
1081
1082#define OCFS2_XATTR_INDEXED 0x1
1083#define OCFS2_HASH_SHIFT 5
1084#define OCFS2_XATTR_ROUND 3
1085#define OCFS2_XATTR_SIZE(size) (((size) + OCFS2_XATTR_ROUND) & \
1086                ~(OCFS2_XATTR_ROUND))
1087
1088#define OCFS2_XATTR_BUCKET_SIZE 4096
1089#define OCFS2_XATTR_MAX_BLOCKS_PER_BUCKET (OCFS2_XATTR_BUCKET_SIZE \
1090                         / OCFS2_MIN_BLOCKSIZE)
1091
1092/*
1093 * On disk structure for xattr block.
1094 */
1095struct ocfs2_xattr_block {
1096/*00*/ __u8 xb_signature[8]; /* Signature for verification */
1097    __le16 xb_suballoc_slot; /* Slot suballocator this
1098                    block belongs to. */
1099    __le16 xb_suballoc_bit; /* Bit offset in suballocator
1100                    block group */
1101    __le32 xb_fs_generation; /* Must match super block */
1102/*10*/ __le64 xb_blkno; /* Offset on disk, in blocks */
1103    struct ocfs2_block_check xb_check; /* Error checking */
1104/*20*/ __le16 xb_flags; /* Indicates whether this block contains
1105                    real xattr or a xattr tree. */
1106    __le16 xb_reserved0;
1107    __le32 xb_reserved1;
1108    __le64 xb_suballoc_loc; /* Suballocator block group this
1109                       xattr block belongs to. Only
1110                       valid if allocated from a
1111                       discontiguous block group */
1112/*30*/ union {
1113        struct ocfs2_xattr_header xb_header; /* xattr header if this
1114                            block contains xattr */
1115        struct ocfs2_xattr_tree_root xb_root;/* xattr tree root if this
1116                            block cotains xattr
1117                            tree. */
1118    } xb_attrs;
1119};
1120
1121#define OCFS2_XATTR_ENTRY_LOCAL 0x80
1122#define OCFS2_XATTR_TYPE_MASK 0x7F
1123static inline void ocfs2_xattr_set_local(struct ocfs2_xattr_entry *xe,
1124                     int local)
1125{
1126    if (local)
1127        xe->xe_type |= OCFS2_XATTR_ENTRY_LOCAL;
1128    else
1129        xe->xe_type &= ~OCFS2_XATTR_ENTRY_LOCAL;
1130}
1131
1132static inline int ocfs2_xattr_is_local(struct ocfs2_xattr_entry *xe)
1133{
1134    return xe->xe_type & OCFS2_XATTR_ENTRY_LOCAL;
1135}
1136
1137static inline void ocfs2_xattr_set_type(struct ocfs2_xattr_entry *xe, int type)
1138{
1139    xe->xe_type |= type & OCFS2_XATTR_TYPE_MASK;
1140}
1141
1142static inline int ocfs2_xattr_get_type(struct ocfs2_xattr_entry *xe)
1143{
1144    return xe->xe_type & OCFS2_XATTR_TYPE_MASK;
1145}
1146
1147/*
1148 * On disk structures for global quota file
1149 */
1150
1151/* Magic numbers and known versions for global quota files */
1152#define OCFS2_GLOBAL_QMAGICS {\
1153    0x0cf52470, /* USRQUOTA */ \
1154    0x0cf52471 /* GRPQUOTA */ \
1155}
1156
1157#define OCFS2_GLOBAL_QVERSIONS {\
1158    0, \
1159    0, \
1160}
1161
1162
1163/* Each block of each quota file has a certain fixed number of bytes reserved
1164 * for OCFS2 internal use at its end. OCFS2 can use it for things like
1165 * checksums, etc. */
1166#define OCFS2_QBLK_RESERVED_SPACE 8
1167
1168/* Generic header of all quota files */
1169struct ocfs2_disk_dqheader {
1170    __le32 dqh_magic; /* Magic number identifying file */
1171    __le32 dqh_version; /* Quota format version */
1172};
1173
1174#define OCFS2_GLOBAL_INFO_OFF (sizeof(struct ocfs2_disk_dqheader))
1175
1176/* Information header of global quota file (immediately follows the generic
1177 * header) */
1178struct ocfs2_global_disk_dqinfo {
1179/*00*/ __le32 dqi_bgrace; /* Grace time for space softlimit excess */
1180    __le32 dqi_igrace; /* Grace time for inode softlimit excess */
1181    __le32 dqi_syncms; /* Time after which we sync local changes to
1182                 * global quota file */
1183    __le32 dqi_blocks; /* Number of blocks in quota file */
1184/*10*/ __le32 dqi_free_blk; /* First free block in quota file */
1185    __le32 dqi_free_entry; /* First block with free dquot entry in quota
1186                 * file */
1187};
1188
1189/* Structure with global user / group information. We reserve some space
1190 * for future use. */
1191struct ocfs2_global_disk_dqblk {
1192/*00*/ __le32 dqb_id; /* ID the structure belongs to */
1193    __le32 dqb_use_count; /* Number of nodes having reference to this structure */
1194    __le64 dqb_ihardlimit; /* absolute limit on allocated inodes */
1195/*10*/ __le64 dqb_isoftlimit; /* preferred inode limit */
1196    __le64 dqb_curinodes; /* current # allocated inodes */
1197/*20*/ __le64 dqb_bhardlimit; /* absolute limit on disk space */
1198    __le64 dqb_bsoftlimit; /* preferred limit on disk space */
1199/*30*/ __le64 dqb_curspace; /* current space occupied */
1200    __le64 dqb_btime; /* time limit for excessive disk use */
1201/*40*/ __le64 dqb_itime; /* time limit for excessive inode use */
1202    __le64 dqb_pad1;
1203/*50*/ __le64 dqb_pad2;
1204};
1205
1206/*
1207 * On-disk structures for local quota file
1208 */
1209
1210/* Magic numbers and known versions for local quota files */
1211#define OCFS2_LOCAL_QMAGICS {\
1212    0x0cf524c0, /* USRQUOTA */ \
1213    0x0cf524c1 /* GRPQUOTA */ \
1214}
1215
1216#define OCFS2_LOCAL_QVERSIONS {\
1217    0, \
1218    0, \
1219}
1220
1221/* Quota flags in dqinfo header */
1222#define OLQF_CLEAN 0x0001 /* Quota file is empty (this should be after\
1223                 * quota has been cleanly turned off) */
1224
1225#define OCFS2_LOCAL_INFO_OFF (sizeof(struct ocfs2_disk_dqheader))
1226
1227/* Information header of local quota file (immediately follows the generic
1228 * header) */
1229struct ocfs2_local_disk_dqinfo {
1230    __le32 dqi_flags; /* Flags for quota file */
1231    __le32 dqi_chunks; /* Number of chunks of quota structures
1232                 * with a bitmap */
1233    __le32 dqi_blocks; /* Number of blocks allocated for quota file */
1234};
1235
1236/* Header of one chunk of a quota file */
1237struct ocfs2_local_disk_chunk {
1238    __le32 dqc_free; /* Number of free entries in the bitmap */
1239    __u8 dqc_bitmap[0]; /* Bitmap of entries in the corresponding
1240                 * chunk of quota file */
1241};
1242
1243/* One entry in local quota file */
1244struct ocfs2_local_disk_dqblk {
1245/*00*/ __le64 dqb_id; /* id this quota applies to */
1246    __le64 dqb_spacemod; /* Change in the amount of used space */
1247/*10*/ __le64 dqb_inodemod; /* Change in the amount of used inodes */
1248};
1249
1250
1251/*
1252 * The quota trailer lives at the end of each quota block.
1253 */
1254
1255struct ocfs2_disk_dqtrailer {
1256/*00*/ struct ocfs2_block_check dq_check; /* Error checking */
1257/*08*/ /* Cannot be larger than OCFS2_QBLK_RESERVED_SPACE */
1258};
1259
1260static inline struct ocfs2_disk_dqtrailer *ocfs2_block_dqtrailer(int blocksize,
1261                                 void *buf)
1262{
1263    char *ptr = buf;
1264    ptr += blocksize - OCFS2_QBLK_RESERVED_SPACE;
1265
1266    return (struct ocfs2_disk_dqtrailer *)ptr;
1267}
1268
1269#ifdef __KERNEL__
1270static inline int ocfs2_fast_symlink_chars(struct super_block *sb)
1271{
1272    return sb->s_blocksize -
1273         offsetof(struct ocfs2_dinode, id2.i_symlink);
1274}
1275
1276static inline int ocfs2_max_inline_data_with_xattr(struct super_block *sb,
1277                           struct ocfs2_dinode *di)
1278{
1279    unsigned int xattrsize = le16_to_cpu(di->i_xattr_inline_size);
1280
1281    if (le16_to_cpu(di->i_dyn_features) & OCFS2_INLINE_XATTR_FL)
1282        return sb->s_blocksize -
1283            offsetof(struct ocfs2_dinode, id2.i_data.id_data) -
1284            xattrsize;
1285    else
1286        return sb->s_blocksize -
1287            offsetof(struct ocfs2_dinode, id2.i_data.id_data);
1288}
1289
1290static inline int ocfs2_extent_recs_per_inode(struct super_block *sb)
1291{
1292    int size;
1293
1294    size = sb->s_blocksize -
1295        offsetof(struct ocfs2_dinode, id2.i_list.l_recs);
1296
1297    return size / sizeof(struct ocfs2_extent_rec);
1298}
1299
1300static inline int ocfs2_extent_recs_per_inode_with_xattr(
1301                        struct super_block *sb,
1302                        struct ocfs2_dinode *di)
1303{
1304    int size;
1305    unsigned int xattrsize = le16_to_cpu(di->i_xattr_inline_size);
1306
1307    if (le16_to_cpu(di->i_dyn_features) & OCFS2_INLINE_XATTR_FL)
1308        size = sb->s_blocksize -
1309            offsetof(struct ocfs2_dinode, id2.i_list.l_recs) -
1310            xattrsize;
1311    else
1312        size = sb->s_blocksize -
1313            offsetof(struct ocfs2_dinode, id2.i_list.l_recs);
1314
1315    return size / sizeof(struct ocfs2_extent_rec);
1316}
1317
1318static inline int ocfs2_extent_recs_per_dx_root(struct super_block *sb)
1319{
1320    int size;
1321
1322    size = sb->s_blocksize -
1323        offsetof(struct ocfs2_dx_root_block, dr_list.l_recs);
1324
1325    return size / sizeof(struct ocfs2_extent_rec);
1326}
1327
1328static inline int ocfs2_chain_recs_per_inode(struct super_block *sb)
1329{
1330    int size;
1331
1332    size = sb->s_blocksize -
1333        offsetof(struct ocfs2_dinode, id2.i_chain.cl_recs);
1334
1335    return size / sizeof(struct ocfs2_chain_rec);
1336}
1337
1338static inline u16 ocfs2_extent_recs_per_eb(struct super_block *sb)
1339{
1340    int size;
1341
1342    size = sb->s_blocksize -
1343        offsetof(struct ocfs2_extent_block, h_list.l_recs);
1344
1345    return size / sizeof(struct ocfs2_extent_rec);
1346}
1347
1348static inline u16 ocfs2_extent_recs_per_gd(struct super_block *sb)
1349{
1350    int size;
1351
1352    size = sb->s_blocksize -
1353        offsetof(struct ocfs2_group_desc, bg_list.l_recs);
1354
1355    return size / sizeof(struct ocfs2_extent_rec);
1356}
1357
1358static inline int ocfs2_dx_entries_per_leaf(struct super_block *sb)
1359{
1360    int size;
1361
1362    size = sb->s_blocksize -
1363        offsetof(struct ocfs2_dx_leaf, dl_list.de_entries);
1364
1365    return size / sizeof(struct ocfs2_dx_entry);
1366}
1367
1368static inline int ocfs2_dx_entries_per_root(struct super_block *sb)
1369{
1370    int size;
1371
1372    size = sb->s_blocksize -
1373        offsetof(struct ocfs2_dx_root_block, dr_entries.de_entries);
1374
1375    return size / sizeof(struct ocfs2_dx_entry);
1376}
1377
1378static inline u16 ocfs2_local_alloc_size(struct super_block *sb)
1379{
1380    u16 size;
1381
1382    size = sb->s_blocksize -
1383        offsetof(struct ocfs2_dinode, id2.i_lab.la_bitmap);
1384
1385    return size;
1386}
1387
1388static inline int ocfs2_group_bitmap_size(struct super_block *sb,
1389                      int suballocator,
1390                      u32 feature_incompat)
1391{
1392    int size = sb->s_blocksize -
1393        offsetof(struct ocfs2_group_desc, bg_bitmap);
1394
1395    /*
1396     * The cluster allocator uses the entire block. Suballocators have
1397     * never used more than OCFS2_MAX_BG_BITMAP_SIZE. Unfortunately, older
1398     * code expects bg_size set to the maximum. Thus we must keep
1399     * bg_size as-is unless discontig_bg is enabled.
1400     */
1401    if (suballocator &&
1402        (feature_incompat & OCFS2_FEATURE_INCOMPAT_DISCONTIG_BG))
1403        size = OCFS2_MAX_BG_BITMAP_SIZE;
1404
1405    return size;
1406}
1407
1408static inline int ocfs2_truncate_recs_per_inode(struct super_block *sb)
1409{
1410    int size;
1411
1412    size = sb->s_blocksize -
1413        offsetof(struct ocfs2_dinode, id2.i_dealloc.tl_recs);
1414
1415    return size / sizeof(struct ocfs2_truncate_rec);
1416}
1417
1418static inline u64 ocfs2_backup_super_blkno(struct super_block *sb, int index)
1419{
1420    u64 offset = OCFS2_BACKUP_SB_START;
1421
1422    if (index >= 0 && index < OCFS2_MAX_BACKUP_SUPERBLOCKS) {
1423        offset <<= (2 * index);
1424        offset >>= sb->s_blocksize_bits;
1425        return offset;
1426    }
1427
1428    return 0;
1429
1430}
1431
1432static inline u16 ocfs2_xattr_recs_per_xb(struct super_block *sb)
1433{
1434    int size;
1435
1436    size = sb->s_blocksize -
1437        offsetof(struct ocfs2_xattr_block,
1438             xb_attrs.xb_root.xt_list.l_recs);
1439
1440    return size / sizeof(struct ocfs2_extent_rec);
1441}
1442
1443static inline u16 ocfs2_extent_recs_per_rb(struct super_block *sb)
1444{
1445    int size;
1446
1447    size = sb->s_blocksize -
1448        offsetof(struct ocfs2_refcount_block, rf_list.l_recs);
1449
1450    return size / sizeof(struct ocfs2_extent_rec);
1451}
1452
1453static inline u16 ocfs2_refcount_recs_per_rb(struct super_block *sb)
1454{
1455    int size;
1456
1457    size = sb->s_blocksize -
1458        offsetof(struct ocfs2_refcount_block, rf_records.rl_recs);
1459
1460    return size / sizeof(struct ocfs2_refcount_rec);
1461}
1462
1463static inline u32
1464ocfs2_get_ref_rec_low_cpos(const struct ocfs2_refcount_rec *rec)
1465{
1466    return le64_to_cpu(rec->r_cpos) & OCFS2_32BIT_POS_MASK;
1467}
1468#else
1469static inline int ocfs2_fast_symlink_chars(int blocksize)
1470{
1471    return blocksize - offsetof(struct ocfs2_dinode, id2.i_symlink);
1472}
1473
1474static inline int ocfs2_max_inline_data_with_xattr(int blocksize,
1475                           struct ocfs2_dinode *di)
1476{
1477    if (di && (di->i_dyn_features & OCFS2_INLINE_XATTR_FL))
1478        return blocksize -
1479            offsetof(struct ocfs2_dinode, id2.i_data.id_data) -
1480            di->i_xattr_inline_size;
1481    else
1482        return blocksize -
1483            offsetof(struct ocfs2_dinode, id2.i_data.id_data);
1484}
1485
1486static inline int ocfs2_extent_recs_per_inode(int blocksize)
1487{
1488    int size;
1489
1490    size = blocksize -
1491        offsetof(struct ocfs2_dinode, id2.i_list.l_recs);
1492
1493    return size / sizeof(struct ocfs2_extent_rec);
1494}
1495
1496static inline int ocfs2_chain_recs_per_inode(int blocksize)
1497{
1498    int size;
1499
1500    size = blocksize -
1501        offsetof(struct ocfs2_dinode, id2.i_chain.cl_recs);
1502
1503    return size / sizeof(struct ocfs2_chain_rec);
1504}
1505
1506static inline int ocfs2_extent_recs_per_eb(int blocksize)
1507{
1508    int size;
1509
1510    size = blocksize -
1511        offsetof(struct ocfs2_extent_block, h_list.l_recs);
1512
1513    return size / sizeof(struct ocfs2_extent_rec);
1514}
1515
1516static inline int ocfs2_extent_recs_per_gd(int blocksize)
1517{
1518    int size;
1519
1520    size = blocksize -
1521        offsetof(struct ocfs2_group_desc, bg_list.l_recs);
1522
1523    return size / sizeof(struct ocfs2_extent_rec);
1524}
1525
1526static inline int ocfs2_local_alloc_size(int blocksize)
1527{
1528    int size;
1529
1530    size = blocksize -
1531        offsetof(struct ocfs2_dinode, id2.i_lab.la_bitmap);
1532
1533    return size;
1534}
1535
1536static inline int ocfs2_group_bitmap_size(int blocksize,
1537                      int suballocator,
1538                      uint32_t feature_incompat)
1539{
1540    int size = sb->s_blocksize -
1541        offsetof(struct ocfs2_group_desc, bg_bitmap);
1542
1543    /*
1544     * The cluster allocator uses the entire block. Suballocators have
1545     * never used more than OCFS2_MAX_BG_BITMAP_SIZE. Unfortunately, older
1546     * code expects bg_size set to the maximum. Thus we must keep
1547     * bg_size as-is unless discontig_bg is enabled.
1548     */
1549    if (suballocator &&
1550        (feature_incompat & OCFS2_FEATURE_INCOMPAT_DISCONTIG_BG))
1551        size = OCFS2_MAX_BG_BITMAP_SIZE;
1552
1553    return size;
1554}
1555
1556static inline int ocfs2_truncate_recs_per_inode(int blocksize)
1557{
1558    int size;
1559
1560    size = blocksize -
1561        offsetof(struct ocfs2_dinode, id2.i_dealloc.tl_recs);
1562
1563    return size / sizeof(struct ocfs2_truncate_rec);
1564}
1565
1566static inline uint64_t ocfs2_backup_super_blkno(int blocksize, int index)
1567{
1568    uint64_t offset = OCFS2_BACKUP_SB_START;
1569
1570    if (index >= 0 && index < OCFS2_MAX_BACKUP_SUPERBLOCKS) {
1571        offset <<= (2 * index);
1572        offset /= blocksize;
1573        return offset;
1574    }
1575
1576    return 0;
1577}
1578
1579static inline int ocfs2_xattr_recs_per_xb(int blocksize)
1580{
1581    int size;
1582
1583    size = blocksize -
1584        offsetof(struct ocfs2_xattr_block,
1585             xb_attrs.xb_root.xt_list.l_recs);
1586
1587    return size / sizeof(struct ocfs2_extent_rec);
1588}
1589#endif /* __KERNEL__ */
1590
1591
1592static inline int ocfs2_system_inode_is_global(int type)
1593{
1594    return ((type >= 0) &&
1595        (type <= OCFS2_LAST_GLOBAL_SYSTEM_INODE));
1596}
1597
1598static inline int ocfs2_sprintf_system_inode_name(char *buf, int len,
1599                          int type, int slot)
1600{
1601    int chars;
1602
1603        /*
1604         * Global system inodes can only have one copy. Everything
1605         * after OCFS2_LAST_GLOBAL_SYSTEM_INODE in the system inode
1606         * list has a copy per slot.
1607         */
1608    if (type <= OCFS2_LAST_GLOBAL_SYSTEM_INODE)
1609        chars = snprintf(buf, len, "%s",
1610                 ocfs2_system_inodes[type].si_name);
1611    else
1612        chars = snprintf(buf, len,
1613                 ocfs2_system_inodes[type].si_name,
1614                 slot);
1615
1616    return chars;
1617}
1618
1619static inline void ocfs2_set_de_type(struct ocfs2_dir_entry *de,
1620                    umode_t mode)
1621{
1622    de->file_type = ocfs2_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
1623}
1624
1625static inline int ocfs2_gd_is_discontig(struct ocfs2_group_desc *gd)
1626{
1627    if ((offsetof(struct ocfs2_group_desc, bg_bitmap) +
1628         le16_to_cpu(gd->bg_size)) !=
1629        offsetof(struct ocfs2_group_desc, bg_list))
1630        return 0;
1631    /*
1632     * Only valid to check l_next_free_rec if
1633     * bg_bitmap + bg_size == bg_list.
1634     */
1635    if (!gd->bg_list.l_next_free_rec)
1636        return 0;
1637    return 1;
1638}
1639#endif /* _OCFS2_FS_H */
1640
1641

Archive Download this file



interactive